J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

Contents lists available at ScienceDirect

Journal of Behavior Therapy and Experimental Psychiatry journal homepage: www.elsevier.com/locate/jbtep

The manipulation of alcohol-related interpretation biases by means of Cognitive Bias Modification e Interpretation (CBM-I)* Marcella L. Woud a, *, Moniek H.M. Hutschemaekers b, Mike Rinck b, Eni S. Becker b a b

Mental Health Research and Treatment Center, Department of Psychology, University of Bochum, Germany Behavioural Science Institute, Radboud University Nijmegen, The Netherlands

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 July 2014 Received in revised form 19 January 2015 Accepted 2 March 2015 Available online 11 March 2015

Background and objectives: There is a large body of evidence demonstrating that alcohol abuse and misuse is characterized by alcohol-related interpretation biases (IBs). The present study tested whether alcohol-related IBs can be trained, and whether this has an effect on alcohol-related associations and drinking behavior. A newly developed alcohol Cognitive Bias Modification e Interpretation (CBM-I) training was employed. The potential moderating effect of executive control on CBM-I training effects was tested. Method: Participants were hazardously male drinking students. A classical Stroop was used to assess levels of executive control. Half of the sample was trained to interpret ambiguous alcohol-related scenarios in an alcohol-related manner (alcohol training group), whereas the other half was trained to interpret ambiguous alcohol-related scenarios in a neutral manner (neutral training group). A Single Target Implicit Association Test (STIAT) was used to test whether the training would generalize to implicit alcohol-related associations (target words: alcohol, attributes: positive vs. neutral). To test the training's effect on drinking behavior, a bogus taste test and a one week follow-up measure assessing participant's real life drinking behavior were used. Results: The CBM-I training was partly successful: When presented with novel ambiguous alcoholrelated scenarios, participants of the alcohol training group interpreted these scenarios as more alcohol-related after the training. However, there was no reduction in alcohol-related IBs in the neutral training group. Results of the STIAT demonstrated that both training groups showed stronger positive than neutral alcohol-related associations. However, there were no between-group differences in alcoholrelated associations. Moreover, the CBM-I training's effect was not moderated by levels of executive control. Finally, no group differences were found on levels of alcohol consumption (bogus taste test and at one week follow-up). Limitations: The neutral training might have been operationalized sub-optimally. A multi-session training might have resulted in stronger effects. Conclusions: These findings are the first to show that alcohol-related IBs can be trained. However, the training effect only partly generalized so more research is needed to advance our understanding of alcohol CBM-I effects. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Alcohol-related interpretation bias CBM-I training Hazardous drinkers Implicit associations Taste test Executive control

1. Introduction

* This study was conducted when Marcella L. Woud was still at Behavioural Science Institute, Radboud University Nijmegen, Montessorilaan 3, 6525 HR Nijmegen, The Netherlands. * Corresponding author. Mental Health Research and Treatment Center, Department of Psychology, University of Bochum, Germany, Massenbergstraße 9-13, 44787 Bochum, Germany. Tel.: þ49 (0)234/32 21502. E-mail address: [email protected] (M.L. Woud).

http://dx.doi.org/10.1016/j.jbtep.2015.03.001 0005-7916/© 2015 Elsevier Ltd. All rights reserved.

According to models of addictive behaviors (e.g., Wiers et al., 2007), alcohol misuse and dependency are characterized by alcohol-related interpretation biases (IBs). This relationship has received substantial empirical support (for a review, see Stacy & Wiers, 2010). An alcohol-related IB can be defined best as an individual's tendency to more often endorse an alcohol-related than neutral interpretation when confronted with ambiguous,

62

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

potentially alcohol-relevant information. Word association tasks are well-established paradigms to investigate such biases (e.g., Stacy, Ames, & Grenard, 2006; Stacy & Wiers, 2010). During these tasks, participants give their first spontaneous association to ambiguous cues that could be interpreted as either alcohol-related or alcohol-unrelated. Various operationalizations have been employed, for example, single word cues (e.g., ‘shot’: gun shot or drink; ‘pitcher’: baseball term or jug) or open-ended ambiguous scenarios (e.g. from Woud, Fitzgerald, Wiers, Rinck, & Becker, 2012: ‘At the festival. You and your friends are attending a festival. You want to have a big night out. So you and your friend are quickly going to the … ’, and for similar studies, see e.g., Ames & Stacy, 1998; Salemink & Wiers, 2014; Stacy, 1995, 1997; Woud et al., 2015). Results have consistently showed that individuals with high levels of alcohol consumption exhibit alcohol-related IBs, i.e., they generated more alcohol-related than alcohol-unrelated interpretations in response to ambiguous alcohol-related cues. Although these findings are important, they are limited: results were obtained via cross-sectional designs, and thus merely show that alcohol-related IBs are positively related to levels of alcohol consumption. That is, such data show that alcohol-related IBs are a correlate of alcohol abuse and misuse (Kraemer et al., 1997). Consequently, it remains unclear whether alcohol-related IBs causally contribute to alcohol abuse and misuse. However, examining the causal hypothesis is of great interest, from a theoretical and clinical perspective. There is only one methodological approach that provides an adequate option here, namely an experimental design. Here, it needs to be tested whether 1. alcohol-related IBs can be modified; and whether 2. this modification leads to changes in levels of alcohol consumption (for an overview of how to structurally examine the typology of a causal risk factor, see Kraemer et al. (1997); and for similar reasoning in the CBM literature, see e.g., MacLeod, Rutherford, Campbell, Ebsworthy, & Holker, 2002). The present study addressed the causal hypotheses in the context of alcohol-related IBs: It employed an experimental design that examined whether alcohol-related IBs can be trained (i.e., increased and decreased, given the ‘proof of principle’ character of this study), and whether this would generalize to alcohol-related associations and drinking behavior in a training-compatible manner. Procedures to manipulate cognitive biases employ methods developed within the Cognitive Bias Modification (CBM) literature (Koster, Fox, & MacLeod, 2009; Woud & Becker, 2014). To modify IBs, variants of the scenario-based training paradigm developed by Mathews and Mackintosh (2000) have frequently been used (Cognitive Bias Modification e Interpretation; CBM-I). CBM-I requires participants to read ambiguous sentences that end in a tobe-completed word fragment. Participants have to complete the fragment, which then produces an outcome consistent with, for example, a functional or dysfunctional interpretation. By means of a two-phase procedure, it is then tested whether trainingcompatible changes in interpretation could be established: During the encoding-phase, participants are presented with novel ambiguous scenarios. In a subsequent recognition-phase, participants have to interpret these scenarios. Moreover, generalization to other cognitive concepts and/or (dysfunctional) behavioral responses is tested. To date, CBM-I has been used successfully in emotional pathology (for reviews, see e.g., Hallion & Ruscio, 2011; Hertel & Mathews, 2011; MacLeod & Mathews, 2012), e.g., in the context of depression (e.g., Blackwell & Holmes, 2010; Williams, Blackwell, Mackenzie, Holmes, & Andrews, 2013; Yiend et al., 2014), social anxiety (Salemink, van den Hout, & Kindt, 2007, 2009) or analog posttraumatic stress (e.g., Woud, Holmes, Postma, Dagleish, & Mackintosh, 2012; Woud, Postma, Holmes, & Mackintosh, 2013). These studies robustly showed that IBs can be

modified. However, findings concerning the trainings' generalization are rather mixed (Fox, Mackintosh, & Holmes, 2014). The goal of the present study was to test whether alcoholrelated IBs can also be manipulated (i.e., increased and decreased) using a CBM-I training procedure. Previous alcohol-related CBM research has focused on alcohol-related attentional biases (Field & Eastwood, 2005; Schoenmakers, Wiers, Jones, Bruce, & Jansen, 2007) and alcohol approach tendencies (Eberl et al., 2012; Wiers, Eberl, Rinck, Becker, & Lindenmeyer, 2011; Wiers et al., 2010). In the present CBM-I training, male hazardous drinking students were trained either to interpret ambiguous alcohol-related scenarios in an alcohol-related manner (alcohol training group, ATG), or to interpret the scenarios in an alcohol-unrelated manner (neutral training group, NTG). Participants were presented with openended, ambiguous alcohol-related scenarios and were instructed to finish each scenario by completing a word fragment. Completing the word fragment produced an outcome consistent with either an alcohol-related or alcohol-unrelated interpretation (depending on the training condition). By means of the encoding-recognition procedure it was then tested whether the CBM-I training was successful. We expected the ATG to show an increase in alcoholrelated IBs post training, whereas the NTG should show a decrease in alcohol-related IBs post training. To test the training's generalization effects, a Single Target Implicit Associations Test (STIAT; Wigboldus, Holland, & van Knippenberg, 2004) was used, assessing positive versus neutral alcohol-related associations. We expected participants in the ATG to show stronger positive than neutral alcohol-related associations post training, and vice versa for the NTG. Moreover, we expected the ATG to show stronger positive alcohol-related associations than the NTG. This expectation is based on the conceptual overlap between the CBM-I training and the STIAT (i.e., interpretations vs. associations). Moreover, the results of Wiers et al. (2010, 2011) showed that CBM training affected alcohol-related associations in a training-compatible manner (assessed via the Implicit Association Test, IAT; Greenwald, McGhee, & Schwartz, 1998). A bogus taste test was used to assess participants' drinking behavior. Moreover, we measured participants' drinking behavior a week after the training. On both measures, we expected the ATG to consume more alcohol than the NTG. Finally, we investigated whether the training's effect depended on individual levels of executive control (EC). This was motivated by the findings of Salemink and Wiers (2014). They employed a CBM-I training in the context of adolescent anxiety (i.e., positive versus control training) and found that individuals with low levels of EC benefited the most from the positive training. Following this, we expected that individuals with low levels of EC, compared to individuals with high levels of EC, would show the strongest training effect, in the ATG and NTG. 2. Methods 2.1. Participants Seventy-four male students of Radboud University Nijmegen participated (Mage ¼ 22.16, SD ¼ 2.58). The study's description explained that participants would actually take part in two studies to disguise the connection between CBM-I training and taste test (see e.g., Houben, Schoenmakers, & Wiers, 2010). Participants were screened via a ‘health-check’ which included the Alcohol Use Disorders Identification Test (AUDIT; Saunders, Aasland, Babor, de la Fuente, & Grant, 1993). If a participant's AUDIT score was 8 or higher, he was invited to take part (AUDIT total sample M ¼ 13.38, SD ¼ 4.41) (for additional studies using this cut-off, see e.g., Wiers, Rinck, Dictus, & Van den Wildenberg, 2009; Woud, Fitzgerald, et al., 2012). The health-check also included a variety of filler questions.

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

Other eligibility criteria were age between 18 and 30 and fluent in written Dutch. 2.2. Materials 2.2.1. Self-report measures 2.2.1.1. Lab health-check. A second health-check was administered at the lab. Of main interest here were the questions concerning thirst and urge to drink (11-point Likert scale). It also included filler questions and three mood measures (i.e., tensed vs. relaxed, passive vs. active, unhappy vs. happy, 11-point Likert scale). 2.2.1.2. Time-line-follow-back (TLFB). During the lab session, alcohol consumption was measured with an adapted version of the TLFB questionnaire (Sobell & Sobell, 1992; Wiers, Hoogeveen, Sergeant, & Gunning, 1997). Participants summed up for every day of the past week, how many and what kind of standard alcoholic drinks they had consumed, and how many drinks they would typically drink on an average day. In addition, they indicated the number of occasions on which they had drunk five or more standard glasses of alcohol during the previous 2 weeks. 2.2.1.3. One week follow-up questionnaire. This questionnaire contained the Time Line Follow Back (TLFB) questionnaire during which participants' actual drinking behavior during the 7 days following the study was assessed. Moreover, participants' mood was assessed (negative vs. positive, 11-point Likert scale). Finally, awareness was checked via 4 questions: (1) What do you think is the aim of this study?, (2) What do you think was the aim of the computer task where you were supposed to complete word fragments, (3) Do you think that the word completion task influenced you and if yes, how?, (4) Do you think that the taste test influenced you and if yes, how? Of main interest here were the results of question 1, 2 and 3. 2.2.2. Classical stroop During the classical Stroop task (Stroop, 1935) participants categorized stimuli according to their print color. On the compatible card, the words ‘yellow’, ‘green’, ‘red’, ‘blue’ and ‘white’ were shown, and word meaning and print color were matched. The same words were shown on the incompatible card. However, there was no match between meaning and color. Finally, there was a card

63

showing meaningless colored strings of XXX. Each card contained 40 stimuli, i.e., 8 stimuli distributed across 5 columns. The ‘XXX card’ was shown first, whereas the other cards were presented randomly (initiated by a press on the spacebar via the experimenter). After a participant had named the last word's color, the experimenter pressed again and the card disappeared. Reaction times were saved by the computer, and errors were recorded by the experimenter who was blind to the card type. 2.2.3. Alcohol Cognitive Bias Modification e interpretation (CBM-I) training 2.2.3.1. Training phase. The main CBM-I training contained 65 ambiguous alcohol-related scenarios and 10 neutral filler scenarios. There were also two ‘CBM-I reactivation trainings’ which were administered in order to boost the training's effects. Both reactivation sessions contained 10 ambiguous alcohol-related scenarios and 1 neutral filler scenario. Scenarios described typical student life situations with a clear positive connotation in which alcohol could be involved (e.g., being with friends, joining a party). Each scenario started with a title and contained three lines. The last line ended in a to-be-completed word fragment such that the meaning of the scenario remained ambiguous until word completion. Participants were asked to complete the word fragment. Word fragments were designed such that only one possible solution could complete the scenario's meaning, and these words produced an outcome consistent with either an alcohol-related or alcohol-unrelated interpretation. To encourage thorough processing of the sentences' meanings, 32 scenarios were followed by a comprehension question (for examples, see Table 1). A training trial was as follows: The title and the three lines were presented consecutively, and participants were required to press the space bar when they had read the sentence. To ensure that participants read the scenarios properly, the title and sentences remained on the screen for at least 2 s. After the last sentence, pressing the space bar revealed the word fragment. Once participants knew the correct missing letter, they were required to press the space bar again, and then to type in the first missing letter. If the letter was correct, the completed word appeared. If the letter was incorrect, error feedback was provided until the correct missing letter was pressed. Then, either a comprehension question followed

Table 1 Examples of stimuli during the CBM-I training and Encoding Recognition phase. Phase

Type

Contents

Training

Title Alcohol training sentence

Cinema You are going to the cinema with some friends. This includes buying something yummy to eat and to drink. You buy some … b e _ r (beer) M & _ 's (M&M's) Are you going to a concert? Studying together You and your study buddy are working on an assignment. You notice that your way of working differs a lot. You are very … a c c _ r a t e (accurate) i n a c c _ r a t e (inaccurate) Day at the beach You had a day at the beach with some friends. You enjoyed the sun throughout the whole day. At the end of the day, you all fancy something … y u m _ y (yummy) Are you at the beach? Day at the beach At the end of the day, you all fancy a cold beer. At the end of the day, you all fancy French fries. At the end of the day, you all fancy something special. At the end of the day, you all fancy something they sell at the beach kiosk.

Alcohol word fragment Neutral word fragment Comprehension question Title Neutral training sentence

Encoding

Recognition

Word fragment A Word fragment B Title Encoding sentence Word fragment Comprehension question Title Alcohol-related target Alcohol-unrelated target Foil 1 Foil 2

64

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

or a new scenario title appeared. Participants were randomly allocated to the two training conditions. 2.2.3.2. Measuring alcohol-related IBs. This included a two-phase procedure. During encoding, ten novel ambiguous alcohol-related scenarios were presented, starting with a title followed by three lines. Participants were again instructed to complete the word fragment. However, this did not resolve the scenarios' ambiguity. After each scenario, participants were required to rate how well they could imagine themselves in the situation (10-point Likert scale). Four scenarios were followed by comprehension questions (for examples, see Table 1). There were two scenarios sets and the order of sets was counterbalanced. In the recognition phase, the original ten encoding-phase scenario titles were presented again, followed by four sentences. Participants rated how close in meaning each sentence seemed to the original scenario (4-point Likert scale). One sentence represented an alcohol-related interpretation and one an alcoholunrelated interpretation (targets). The two additional sentences did not provide a resolution of the alcohol-related ambiguity (foils; see Table 1). 2.2.4. Single Target Implicit Associations Test (STIAT) The STIAT (Wigboldus et al., 2004) included six positive and six neutral attribute words, and six alcohol words as targets. During the attribute discrimination block, participants categorized words according to their valence (positive vs. neutral). Participants were asked to press one key (letter ‘A’) in response to positive attributes, and the other key (number ‘6’, numlock field) in response to neutral attributes. Each attribute was presented once. In the second block, the six alcohol-related targets were introduced. Half of the participants pressed the left key (‘A’), and the other half pressed the right key (‘6’). The following combined block was a combination of target and attribute discrimination (presented in mixed random order) and included 24 trials: The six alcohol targets appeared once and the six attributes that required the same response were also shown once (i.e., 12 trials). Because the other set of attributes was assigned to only one key, these attributes were presented twice as often, i.e., 2  6. During the reversed target discrimination block, participants practiced the reversed response assignment for alcohol targets. Participants who had previously pressed the positive key in response to alcohol now had to respond with the neutral key, the other half vice versa. The final reversed combined block combined the unchanged attribute discrimination with the reversed target discrimination (24 trials). Block sequence and key assignment were counterbalanced. After incorrect responses, a red ‘X’ appeared. The two combined blocks are labeled as follows: compatible block: alcohol targets and positive attributes share a response key; incompatible block: alcohol targets and neutral attributes share a response key. 2.2.5. Taste test The bogus taste test included 3 different alcohol-free beers (introduced as regular beers) and 3 different types of cola (200 ml per glass, numbered with 1, 2 or 3 used for ratings). The brandnumber assignment was quasi-randomized. Participants were informed about the brands and the 10 min time limit and were told that they could drink as much as they liked. The ratingquestionnaire contained four tasks per beverage: evaluation of the beverages' taste (11-point Likert scale), assignment of brands to the glasses, indication of which glass they would prefer to drink if they could choose among them, rating how easy the assignment was (11-point Likert scale).

Following this, a second taste test followed as part of the cover story. Here, participants tasted five different M&Ms and had to guess their color. 2.3. Procedure After participants signed the informed consent for study 1, they completed the lab health-check (including the first urge and mood assessment) and the Stroop. The pre-training Encoding Recognition

Health-check online version Health-check lab version: Thirst, urge assessment (T1) and mood Encoding-recognition pre-training CBM-I : Alcohol / Neutral Encoding - recognition post-training CBM-I reactivation training

Urge/Mood asesssment (T2)

STIAT

Urge/Mood assessment (T3)

CBM-I reactivation training

Urge /Mood assessment (T4)

Taste test

TLFB lab version TLFB follow-up, awareness check and mood Fig. 1. Flowchart procedure.

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

task, the CBM-I training, and the post-training Encoding Recognition task followed. Next, the first reactivation CBM-I training and second urge and mood assessment were administered, followed by the STIAT. Then, participants completed the third urge and mood assessment and another reactivation CBM-I training. Next, participants completed the fourth urge and mood assessment and were told that they had finished study 1. Consequently, they signed informed consent for study 2. This was followed by the taste tests and the TFLB. After that, participants were informed that the beer was alcohol-free and their awareness of this was checked. Finally, participants were informed about the one-week follow up which was e-mailed (see Fig. 1 for flowchart). Ethical approval for the study was granted by the ethics committee of the Behavioural Science Institute. 3. Results 3.1. Participant characteristics One participant was excluded as he was tested despite not meeting the AUDIT requirement (final sample: N ¼ 73; alcohol training group (ATG) n ¼ 35, neutral training group (NTG): n ¼ 38). There were no significant between-group differences on the lab health-check (i.e., age, thirst, urge, and mood), drinking behavior during the week before the experiment (i.e., TLFB lab version), ratings on alcohol and neutral target sentences on the Encoding Recognition task pre-training and Stroop scores (for means, standard deviations and significances, see Table 2). However, the groups differed significantly on the AUDIT, t(71) ¼ 2.18, p ¼ .03, with higher scores in the ATG than NTG (ATG: M ¼ 14.6, SD ¼ 4.21; NTG: M ¼ 12.42, SD ¼ 4.32). Hence, AUDIT scores were entered as covariate in between-group analyses. To examine changes in urge, an ANCOVA including Time (T1, T2, T3, T4) as within-subjects factor, Group (ATG, NTG) as betweensubjects factor, and AUDIT as covariate. The Time  Group interaction was not significant, F(3,68) ¼ .58, p ¼.63, and neither were the main effects, F's < .19, p's > .05 (for means and standard deviations, see Table 2).1 To examine mood changes, three repeated measures ANCOVAs were conducted, each including Time (T1, T2, T3, T4) as withinsubjects factor, Group (ATG; NTG) as between-subjects factor, and AUDIT as covariate. Of most importance here was the Group  Time interaction: tensed versus relaxed: F(3,68) ¼ 1.74, p ¼ .17; passive versus active: F(3,68) ¼ .06, p ¼ .98; unhappy versus happy: F(3,68) ¼ 4.58, p < .01. Hence, levels of happiness changed over the course of the study. However, none of the between-group comparisons was significant (for significances per time point, and means and standard deviations of all mood states, see Table 2). 3.2. CBM-I training: manipulation check To test whether the training changed the alcohol-related IBs in a training-congruent manner, an ANCOVA was conducted, with Time (pre-training, post-training) and Sentence Type (alcohol target, neutral target) as within-subjects factors, Group (ATG, NTG) and Sentence Set (AB, BA) as between-subjects factors, and AUDIT as covariate. Results showed a marginal significant interaction of Group  Time  Sentence Type, F(1,68) ¼ 3.17, p ¼ .08, eta2 ¼ .04. Following this, we conducted two ANOVAs with Time (pre-training, post-training) and Sentence Type (alcohol target, neutral target) as

1 For the sake of brevity, we only report the outcomes relevant for our hypotheses/main outcomes per analysis.

65

Table 2 Demographic data, urge and mood: Means, standard deviations and significances per group. Measure

Alcohol training group (n ¼ 35) M (SD)

M (SD)

Age Thirst TLFB lab versiona AUDIT Alcohol target Neutral target Urge T1 Urge T2 Urge T3 Urge T4 Tensed/Relaxed T1 Tensed/Relaxed T2 Tensed/Relaxed T3 Tensed/Relaxed T4 Passive/Active TI Passive/Active T2 Passive/Active T3 Passive/Active T4 Happy/Unhappy T1 Happy/Unhappy T2 Happy/Unhappy T3 Happy/Unhappy T4 Mood follow-upb

22.09 5.86 3.65 14.60 2.57 2.73 2.71 4.06 3.97 4.09 7.17 7.31 7.23 7.40 5.94 5.97 6.17 5.94 7.54 7.26 7.20 7.11 7.07

22.21 5.95 3.02 12.42 2.43 2.71 2.71 3.53 3.50 3.58 7.47 7.53 7.03 7.32 6.24 6.13 6.34 6.08 7.13 7.32 7.26 7.29 7.12

(2.48) (1.46) (1.83) (4.21) (.44) (.41) (2.32) (2.52) (2.61) (2.79) (1.98) (1.60) (1.42) (1.44) (1.59) (1.65) (1.60) (1.63) (1.04) (1.01) (1.13) (1.23) (1.31)

Neutral training group (n ¼ 38)

(2.73) (1.71) (1.84) (4.32) (.62) (.54) (1.92) (2.25) (2.45) (2.47) (1.39) (1.48) (1.35) (1.32) (1.68) (1.69) (1.83) (1.70) (1.30) (1.09) (1.13) (1.14) (1.39)

Significances

t(71) t(71) t(70) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(71) t(62)

¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼

.20, p ¼ .84 .24, p ¼ .81 1.44, p ¼ .15 2.18, p ¼ .03 1.06, p ¼ .29 .19, p ¼ .85 .01, p ¼ .99 .95, p ¼ .35 .80, p ¼ .43 .82, p ¼ .41 .76, p ¼ .45 .59, p ¼ .56 .63, p ¼ .53 .26, p ¼ .80 .77, p ¼ .45 .41, p ¼ .69 .42, p ¼ .67 .35, p ¼ .73 1.49, p ¼ .14 .24, p ¼ .81 .24, p ¼ .81 .63, p ¼ .53 .15, p ¼ .88

Note. TLFB: Time Line Follow Back questionnaire; AUDIT: Alcohol Use Disorder Identification Test; Alcohol target and neutral target: Similarity ratings for alcohol and neutral target sentences of Encoding Recognition Task pre CBM-I training; Urge/ Mood states: pre-training: T1, post first reactivation CBM-I training: T2, post STIAT: T3, post second reactivation CBM-I training: T4. a The data of one participant was missing. b This analysis only includes participants who completed the follow-up questionnaire.

within-subjects factors, and Sentence Set (AB, BA) as betweensubjects factor separately for the ATG and NTG. 3.2.1. NTG The Time  Sentence Type interaction was not significant, F(1,36) ¼ .52, p ¼ .47 (neutral target: pre-training: EMM ¼ 2.71, SE ¼ .09, post-training: EMM ¼ 2.81, SE ¼ .08; alcohol target: pretraining: EMM ¼ 2.42, SE ¼ .1, post-training: EMM ¼ 2.46, SE ¼ .1). 3.2.2. ATG The Time  Sentence Type interaction was significant, F(1,33) ¼ 7.31, p ¼ .01, eta2 ¼ .18. Therefore, we conducted two additional repeated measures ANOVAs for each Sentence Type: Time (pre-training, post-training) was used as within-subjects factor and Sentence Set (AB, BA) as between-subjects factor. For the alcohol target sentences, we found a significant main effect of Time: F(1,33) ¼ 11.82, p > .01, eta2 ¼ .26, with higher means for similarity ratings on alcohol target sentences after the training (pre-training: EMM ¼ 2.57, SE ¼ .08, post-training: EMM ¼ 2.74, SE ¼ .07). For the neutral target sentences, the main effect of Time was not significant: F(1,33) ¼ .53, p ¼ .47 (pre-training: EMM ¼ 2.72, SE ¼ .07, post-training: EMM ¼ 2.67, SE ¼ .08). 3.3. Single target association test (STIAT) We used the D600-measure scoring algorithm, which combined test trials during compatible and incompatible blocks into an overall score (see Glashouwer, Smulders, de Jong, Roefs, & Wiers, 2013; Greenwald, Nosek, & Banaji, 2003). One sample t-tests showed that the STIAT scores differed significantly from zero in both training groups: ATG: t(34) ¼ 3.1, p < .01, d ¼ .54 (M ¼ .22,

66

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

SD ¼ .41); NTG: t(37) ¼ 2.17, p ¼ .04, d ¼ .35 (M ¼ .12, SD ¼ .34), showing that both groups had stronger positive than neutral alcohol-related associations (i.e., a positive D600 score). A univariate ANCOVA was conducted to test between-group differences, including the between-subjects factor Group (ATG, NTG) and Order STIAT Blocks (compatible-incompatible, incompatible-compatible), and AUDIT scores as covariate. However, the main effect of Group was not significant, F(1,68) ¼ .34, p ¼ .56, indicating that the two groups did not differ on alcohol-related associations (ATG: EMM ¼ .19, SE ¼ .06; NTG: EMM ¼ .14, SE ¼ .05). 3.4. Taste test and one week follow-up Two participants were excluded: One participant knew that the beer was alcohol free and another participant continued to drink although the taste test was finished. A univariate ANCOVA was conducted with Group (ATG, NTG) as between-subjects factor, AUDIT scores as covariate and amount drunk as dependent variable. The main effect of Group was not significant, F(1,68) ¼ 1.44, p ¼ .24, indicating that the ATG did not drink more than the NTG (ATG: EMM ¼ 37.98, SE ¼ 6.13; NTG: EMM ¼ 48.18, SE ¼ 5.7). For the ANCOVA on amount of soda drunk, we found a marginal significant main effect of Group, F(1,68) ¼ 3.56, p ¼ .06, eta2 ¼ .05 (ATG: EMM ¼ 36.57, SE ¼ 4.9; NTG: EMM ¼ 49.49, SE ¼ 4.58). Regarding the group difference on alcohol drunk during the 7 days following the experiment (i.e., online TLFB, data of 8 participants were missing), results of a third ANCOVA showed also no significant main effect of Group, F(1,62) ¼ .31, p ¼ .58 (ATG: EMM ¼ 3.18, SE ¼ .34; NTG: EMM ¼ 2.92, SE ¼ .32). 3.5. Stroop Prior to the analysis, RTs of compatible cards were subtracted from RTs of incompatible cards, and this difference score served as dependent variable for all analyses. Results of a one sample t-test showed the expected Stroop interference, t(71) ¼ 12.21, p < .001, d ¼ 1.44 (M ¼ 5611, SD ¼ 3900). To examine the moderating role of EC, a repeated measure ANCOVA was conducted, including the between-subjects factors Group (ATG, NTG) and Order Cards (compatible-incompatible, incompatible-compatible), the withinsubjects factor Time (difference score alcohol-neutral targets pre, difference score alcohol-neutral targets post), and Stroop and AUDIT scores as covariates (all variables were z-standardized first, see Aiken & West, 1991). Of most importance here was the Group  Time  Stroop interaction. However, this effect was not significant, F(1,66) ¼ 2.41, p ¼ .13. 3.6. Awareness check As with the online TLFB questionnaire, 8 participants did not return it. Moreover, 5 participants did not complete the awareness questions (final sample: N ¼ 65). Answers were divided into three categories: not aware, having a vague idea, aware. A participant was considered as aware if his answer included terms indicating a manipulation towards or away from alcohol, and that this manipulation was supposed to affect his alcohol-related cognitions and/or drinking behavior. Results were as follows: Question one (i.e., general aim study): two participants of the ATG had a vague idea; Question two (i.e., aim word completion task): one participant of the ATG was aware, six participants had a vague idea (ATG n ¼ 5, NTG n ¼ 1); Question three (i.e., influence word completion task): seven participants had a vague idea (ATG n ¼ 5, NTG n ¼ 2). To conclude, participants were generally unaware of the study's aim and the tasks' functions.

4. Discussion The present study tested the effects of a newly developed alcohol CBM-I training. Results showed that the ATG interpreted novel ambiguous alcohol-related scenarios as more alcohol-related after the training. However, the NTG did not show a reduction in alcohol-related IBs after the training. Regarding the Single Target Implicit Association Test (STIAT; Wigboldus et al., 2004), both training groups showed stronger positive than neutral implicit alcohol-related associations. However, there were no between group differences. Regarding participants' drinking behavior, results did not show significant group differences during the taste test and in actual drinking behavior the week after the training. Moreover, results showed that EC did not moderate the training's effect. Finally, there were no between-group differences concerning changes in urge and mood over the course of the study. To conclude, the present alcohol CBM-I training was partly successful. A first point of discussion concerns the absence of a training effect in the neutral training group. As participants were a relatively homogenous sample with respect to their drinking behavior, a change in bias in both directions should have been possible (MacLeod et al., 2002). Hence, what could offer an explanation are the characteristics of the two CBM training groups. Both groups were presented with ambiguous alcohol-related scenarios. In the alcohol training group, these scenarios were consistently completed with an alcohol-related word fragment, i.e., a homogenous category. In the neutral training group, however, the scenarios were consistently completed with an alcohol-unrelated word fragment (e.g., food or alcohol-unrelated activities). Hence, it seems likely that learning during the neutral training was more difficult and thus did not reveal a training effect in that group. Following the discrepant results of the manipulation check, interpreting the results of the measures applied to test the training's generalization effect is not that straightforward. To start with, we found that both training groups showed stronger positive than neutral alcohol-related associations on the STIAT. As there was no training effect in the NTG, this result seems logical. Moreover, we tested hazardous drinking students who can hold positive alcoholrelated associations (e.g., Houben & Wiers, 2007; but for different findings, see e.g., Houben & Wiers, 2008; Wiers, Van Woerden, Smulders, & De Jong, 2002). Generally, the within-group findings are in line with those of Wiers et al. (2010, and see Wiers et al., 2011). Their alcohol-CBM study manipulated alcohol-related approach-avoidance tendencies and results showed generalization effects to the Implicit Association Test (IAT; Greenwald et al., 1998): Participants who were trained to avoid alcohol showed stronger alcoholeavoidance associations. However, such an effect was not observed in the alcohol approach condition although this group was successfully trained. We also tested for between group differences on alcohol-related associations. Results did not show such differences, which could be partly explained by the absence of a training effect in the NTG. As Wiers et al. (2010) did not conduct between-group analyses, we cannot compare our data with that study. Hence, a tentative conclusion at this stage is that generalization effects to implicit-alcohol-related associations have been observed though might depend on (subtle) boundary conditions. Results of the taste test and one week follow-up showed that the ATG did not drink more than the NTG, on either measure. This could be explained by the fact that the training involved scenarios including positive social contexts. The taste test, however, took place in the lab. Hence, there was no contextual match between training and taste test. These findings resemble results of Schoenmakers et al.'s attention re-training study (2007). In contrast, CBM studies by Field and Eastwood (2005) and Wiers et al. (2010) did reveal effects on taste tests. However, in Wiers et al.'s study (2010)

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

this effect only occurred in successfully trained participants. A reanalysis of the present data, however, did not change the taste test results. Procedural differences could offer an explanation here: Wiers et al. (2010) applied the taste test immediately after the training, whereas we applied it after the STIAT. However, given the reactivation CBM sessions this might not be the most plausible explanation. Second, Wiers et al. (2010) applied an approach avoidance training, whereas we applied a CBM-I training, so the trainings might differ in regarding their conceptual overlap with the taste test. The approach avoidance training taps into motivational tendencies, whereas CBM-I taps into higher-order cognitive processes. Hence, the approach avoidance training might have a greater conceptual overlap with the taste test, and thus might have more potential to affect (alcohol-related) behavioral responses (for contradicting evidence, see Field & Eastwood, 2005). Finally, we did not find a moderating role of EC. That is, the CBMI training's effect did not depend on individuals levels of EC. This is surprising, given the very promising results of Salemink and Wiers (2014), and the numerous examples of EC being a moderator for alcohol-related implicit cognitive processing (e.g., Grenard et al., 2008; Houben & Wiers, 2009; Thush et al., 2008, but for an exception, see Woud, Becker, Rinck, & Salemink, 2015). The most parsimonious explanation for not finding such a moderation is that we tested a fairly homogenous sample (i.e., male university students). Hence, it is likely that there was not enough variation in Stroop scores to find a moderating effect. This study has at least three limitations. First, we used a singlesession training whereas a multi-session training might have resulted in stronger CBM effects. Second, the increase in alcoholrelated IBs in the ATG may be due to priming effects rather than a bias change. Third, the NTG might not have been operationalized adequately. However, when considered from a clinical perspective, optimizing this condition is crucial, and therefore is a clear target for future research. Moreover, future studies are needed that test variations of taste test procedures that are conceptually closer to the CBM-I training contents in order to find generalization effects to actual drinking behavior. To summarize, results of the present study were only partly in line with our expectations. Hence, more research is needed to further advance our understanding of the effects of alcohol CBM-I. However, given the many examples in the area of emotional pathology, there are various exciting research avenues ahead. Role of funding organizations This research was funded by the Behavioural Science Institute (BSI), Radboud University Nijmegen, The Netherlands. The BSI had no role in study design; in the collection, analysis and interpretation of data; in the writing of the paper; and in the decision to submit the article for publication. Acknowledgments We are grateful to Anne Schwenzfeier who helped with the data collection and to Klaske Glashouwer who provided the syntax for the scoring algorithm of the D600-measure. References Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. London: Sage. Ames, S. L., & Stacy, A. W. (1998). Implicit cognition in the prediction of substance use among drug offenders. Psychology of Addictive Behaviors, 12, 1e10. http:// dx.doi.org/10.1037/0893-164X.12.4.272. Blackwell, S. E., & Holmes, E. A. (2010). Modifying interpretation and imagination in clinical depression: a single case series using cognitive bias

67

modification. Applied Cognitive Psychology, 24, 338e350. http://dx.doi.org/ 10.1002/acp.1680. Eberl, C., Wiers, R. W., Pawelczack, S., Rinck, M., Becker, E. S., & Lindenmeyer, J. (2012). Approach bias modification in alcohol dependence: do clinical effects replicate and for whom does it work best? Developmental Cognitive Neuroscience, 4, 38e51. http://dx.doi.org/10.1016/j.dcn.2012.11.002. Field, M., & Eastwood, B. (2005). Experimental manipulation of attentional bias increases the motivation to drink alcohol. Drug and Alcohol Dependence, 183, 350e357. http://dx.doi.org/10.1007/s00213-005-0202-5. Fox, E., Mackintosh, B., & Holmes, E. A. (2014). Travellers' tales in cognitive bias modification research: a commentary on the special issue. Cognitive Therapy and Research, 38, 239e247. http://dx.doi.org/10.1007/s10608-014-9604-1. Glashouwer, K. A., Smulders, F. T. Y., de Jong, P. J., Roefs, A., & Wiers, R. W. (2013). Measuring automatic associations: Validation of algorithms for the Implicit Association Test (IAT) in a laboratory setting. Journal of Behavior Therapy and Experimental Psychiatry, 44, 105e113. http://dx.doi.org/10.1016/ j.jbtep.2012.07.015. Greenwald, A. G., McGhee, D. E., & Schwartz, L. K. (1998). Measuring individual differences in implicit cognitions: the Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464e1480. http://dx.doi.org/10.1037/00223514.74.6.1464. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197e216. http://dx.doi.org/10.1037/00223514.85.2.197. Grenard, J. L., Ames, S. L., Wiers, R. W., Thush, C., Sussman, S., & Stacy, A. W. (2008). Working memory capacity moderates the predictive effects of drug-related associations on substance use. Psychology of Addictive Behaviors, 22, 426e432. http://dx.doi.org/10.1037/0893-164X.22.3.426. Hallion, L. S., & Ruscio, H. M. (2011). A meta-analysis of the effect of cognitive bias modification on anxiety and depression. Psychological Bulletin, 137, 940e958. http://dx.doi.org/10.1037/a0024355. Hertel, P. T., & Mathews, A. (2011). Cognitive bias modification: past perspectives, current findings, and future applications. Perspectives on Psychological Science, 6, 521e536. http://dx.doi.org/10.1177/1745691611421205. Houben, K., Schoenmakers, T. M., & Wiers, R. W. (2010). I didn't feel like drinking but I don't know why: the effects of evaluative conditioning on alcohol-related attitudes, craving and behavior. Addictive Behaviors, 35(12), 1161e1163. http:// dx.doi.org/10.1016/j.addbeh.2010.08.012. Houben, K., & Wiers, R. W. (2007). Are drinkers implicitly positive about drinking alcohol? Personalizing the alcohol-IAT to reduce negative extrapersonal contamination. Alcohol and Alcoholism, 42(4), 301e307. http://dx.doi.org/ 10.1093/alcalc/agm015. Houben, K., & Wiers, R. W. (2008). Implicitly positive about alcohol? Implicit positive associations predict drinking behavior. Addictive Behaviors, 33(8), 979e986. http://dx.doi.org/10.1016/j.addbeh.2008.03.002. Houben, K., & Wiers, R. W. (2009). Response inhibition moderates the relationship between implicit associations and drinking behavior. Alcoholism: Clinical and Experimental Research, 33, 626e633. http://dx.doi.org/10.1111/j.15300277.2008.00877.x. Koster, E. H. W., Fox, E., & MacLeod, C. (2009). Introduction to the special section on cognitive bias modification in emotional disorders. Journal of Abnormal Psychology, 118, 1e4. http://dx.doi.org/10.1037/a0014379. Kraemer, H. C., Kazdin, A. E., Offord, D. R., Kessler, R. C., Jensen, P. S., & Kupfer, D. J. (1997). Coming to terms with terms of risk. Archives General Psychiatry, 54, 337e343. http://dx.doi.org/10.1001/archpsyc.1997.01830160065009. MacLeod, C., & Mathews, A. (2012). Cognitive bias modification approaches to anxiety. Annual Review of Clinical Psychology, 8, 189e217. http://dx.doi.org/ 10.1146/annurev-clinpsy-032511e143052. MacLeod, C., Rutherford, E., Campbell, L., Ebsworthy, G., & Holker, L. (2002). Selective attention and emotional vulnerability: assessing the causal basis of their association through the experimental manipulation of attentional bias. Journal of Abnormal Psychology, 111, 107e123. http://dx.doi.org/10.1037/0021843X.111.1.107. Mathews, A., & Mackintosh, B. (2000). Induced emotional interpretation bias and anxiety. Journal of Abnormal Psychology, 109, 602e615. http://dx.doi.org/ 10.1037//0021-843X.109.4.602. Salemink, E., van den Hout, M. A., & Kindt, M. (2007). Trained interpretive bias and anxiety. Behaviour Research and Therapy, 45, 329e340. http://dx.doi.org/ 10.1016/j.brat.2006.03.011. Salemink, E., van den Hout, M. A., & Kindt, M. (2009). Effects of positive interpretive bias modification in highly anxious individuals. Journal of Anxiety Disorders, 23, 676e683. http://dx.doi.org/10.1016/j.janxdis.2009.02.006. Salemink, E., & Wiers, R. W. (2014). Alcohol-related memory associations in positive and negative affect situations: drinking motives, working memory capacity, and prospective drinking. Psychology of Addictive Behaviors, 28, 105e113. http:// dx.doi.org/10.1037/a0032806. Saunders, J. B., Aasland, O. G., Babor, T. F., de la Fuente, J. R., & Grant, M. (1993). Development of the Alcohol Use Disorders Screening Test (AUDIT). WHO collaborative project on early detection of persons with harmful alcohol consumption. II. Addiction, 88, 791e804. http://dx.doi.org/10.1111/j.13600443.1993.tb02093.x. 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, 399e405. http://dx.doi.org/10.1111/j.1360-0443.2006.01718.x.

68

M.L. Woud et al. / J. Behav. Ther. & Exp. Psychiat. 49 (2015) 61e68

Sobell, L. C., & Sobell, M. B. (1992). Time line follow-back: a technique for assessing self-reported alcohol consumption. In R. Z. Litten, & J. P. Allen (Eds.), Measuring alcohol consumption: Psychosocial and biochemical methods (pp. 41e71). Totowa, NJ: Humana Press. http://dx.doi.org/10.1007/978-1-4612-0357-5_3. Stacy, A. W. (1995). Memory associations and ambiguous cues in models of alcohol and marijuana use. Experimental and Clinical Psychopharmacology, 3, 183e194. http://dx.doi.org/10.1037/1064-1297.3.2.183. Stacy, A. W. (1997). Memory activation and expectancy as prospective predictors of alcohol and marihuana use. Journal of Abnormal Psychology, 106, 61e73. http:// dx.doi.org/10.1037/0021-843X.106.1.61. Stacy, A. W., Ames, S. L., & Grenard, J. L. (2006). Word association tests of associative memory and implicit processes: theoretical and assessment issues. In R. W. Wiers, & A. W. Stacy (Eds.), Handbook of implicit cognition and addiction (pp. 75e90). Thousand Oaks, CA: Sage Publishers. Stacy, A. W., & Wiers, R. W. (2010). Implicit cognition and addiction: a tool for explaining paradoxical behavior. Annual Review of Clinical Psychology, 6, 551e575. http://dx.doi.org/10.1146/annurev.clinpsy.121208.131444. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643e662. http://dx.doi.org/10.1037/h0054651. Thush, C., Wiers, R. W., Ames, S. L., Grenard, J. L., Sussman, S., & Stacy, A. W. (2008). Interactions between implicit and explicit cognition and working memory capacity in the prediction of alcohol use in at-risk adolescents. Drug and Alcohol Dependence, 94, 116e124. http://dx.doi.org/10.1016/ j.drugalcdep.2007.10.019. Wiers, R. W., Bartholow, B. D., Van den Wildenberg, E., Thush, C., Engels, R. C. M. E., Sher, K. J., et al. (2007). Automatic and controlled processes and the development of addictive behaviors in adolescents: a review and a model. Pharmacology, Biochemistry and Behavior, 86, 263e283. http://dx.doi.org/10.1016/ j.pbb.2006.09.021. Wiers, R. W., Eberl, C., Rinck, M., Becker, E. S., & Lindenmeyer, J. (2011). Retraining automatic action tendencies changes alcoholic patients' approach bias for alcohol and improves treatment outcome. Psychological Science, 22, 490e497. http://dx.doi.org/10.1177/0956797611400615. Wiers, R. W., Hoogeveen, K. J., Sergeant, J. A., & Gunning, W. B. (1997). High and low dose expectancies and the differential associations with drinking in male and female adolescents and young adults. Addiction, 92, 871e888. http://dx.doi.org/ 10.1111/j.1360-0443.1997.tb02956.x. Wiers, R. W., Rinck, M., Dictus, M., & Van den Wildenberg, E. (2009). Relatively strong automatic appetitive action-tendencies in male carriers of the OPRM1 Gallele. Genes Brain and Behavior, 8, 101e106. http://dx.doi.org/10.1111/j.1601183X.2008.00454.x.

Wiers, R. W., Rinck, M., Kordts-Freudinger, R., Dictus, M., Houben, K., van den Wildenberg, E., et al. (2010). Retraining automatic action-tendencies to approach alcohol in hazardous drinkers. Addiction, 105, 279e287. http:// dx.doi.org/10.1111/j.1360-0443.2009.02775.x. Wiers, R. W., Van Woerden, N., Smulders, F. T., & De Jong, P. J. (2002). Implicit and explicit alcohol-related cognitions in heavy and light drinkers. Journal of Abnormal Psychology, 111, 648e658. http://dx.doi.org/10.1037/0021843X.111.4.648. Wigboldus, D. H. J., Holland, R. W., & van Knippenberg, A. (2004). Single target implicit associations (Unpublished manuscript). Williams, A. D., Blackwell, S. E., Mackenzie, A., Holmes, E. A., & Andrews, G. (2013). Combining imagination and reason in the treatment of depression: a randomized controlled trial of internet-based cognitive bias modification and internet-CBT for depression. Journal of Consulting and Clinical Psychology, 81, 793e799. http://dx.doi.org/10.1037/a0033247. Woud, M. L., & Becker, E. S. (2014). Editorial for the special issue on cognitive bias modification techniques: an introduction to a time traveller's tale. Cognitive Therapy and Research, 38, 83e88. http://dx.doi.org/10.1007/s10608-014-9605-0. Woud, M. L., Becker, E. S., Rinck, M., & Salemink, E. (2015). The relationship between drinking motives and alcohol-related interpretation biases. Journal of Behavior Therapy and Experimental Psychiatry, 47, 102e110. Woud, M. L., Fitzgerald, D. A., Wiers, R. W., Rinck, M., & Becker, E. S. (2012). ‘Getting into the spirit’: alcohol-related interpretation bias in heavy-drinking students. Psychology of Addictive Behaviors, 26, 627e632. http://dx.doi.org/10.1037/ a00290255. Woud, M. L., Holmes, E. A., Postma, P., Dalgleish, T., & Mackintosh, B. (2012). Ameliorating intrusive memories of distressing experiences using computerized reappraisal training. Emotion, 12, 778e784. http://dx.doi.org/10.1037/ a0024992. Woud, M. L., Pawelczak, S., Rinck, M., Lindenmeyer, J., Souren, P., Wiers, R. W., et al. (2014). Alcohol-related interpretation bias in alcohol-dependent patients. Alcoholism: Clinical and Experimental Research, 38, 1151e1159. http://dx.doi.org/ 10.1111/acer.12334. Woud, M. L., Postma, P., Holmes, E. A., & Mackintosh, B. (2013). Reducing analogue trauma symptoms by computerized reappraisal trainingeConsidering a cognitive prophylaxis? Journal of Behavior Therapy and Experimental Psychiatry, 44, 312e315. http://dx.doi.org/10.1016/j.jbtep.2013.01.003. Yiend, J., Lee, J. S., Tekes, S., Atkins, L., Mathews, A., Vrinten, M., et al. (2014). Modifying interpretation in a clinically depressed sample using ‘Cognitive Bias Modification-Errors’: a double blind randomised controlled trial. Cognitive Therapy and Research, 38, 1e14. http://dx.doi.org/10.1007/s10608-013-9571-y.