Testing the Efficacy of a Cognitive Bias Modification Task for Cognitively Vulnerable Individuals

Home , Cognitive vulnerability

Too Much of a Good Thing: Testing the Efficacy of a Cognitive Bias Modification Task for Cognitively Vulnerable Individuals

Gerald J. Haeffel, David C. Rozek, Jennifer L. Hames & Jessica Technow

Cognitive Therapy and Research

ISSN 0147-5916 Volume 36 Number 5

Cogn Ther Res (2012) 36:493-501 DOI 10.1007/s10608-011-9379-6

1 23 Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media, LLC. This e-offprint is for personal use only and shall not be self- archived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication.

1 23 Author's personal copy

Cogn Ther Res (2012) 36:493–501 DOI 10.1007/s10608-011-9379-6

ORIGINAL ARTICLE

Too Much of a Good Thing: Testing the Efﬁcacy of a Cognitive Bias Modiﬁcation Task for Cognitively Vulnerable Individuals

Gerald J. Haeffel • David C. Rozek • Jennifer L. Hames • Jessica Technow

Published online: 13 July 2011 Ó Springer Science+Business Media, LLC 2011

Abstract A growing body of research indicates that training might be most effective in reducing cognitive cognitive bias modification (CBM) may be an effective vulnerability when initially used in small doses. intervention for individuals with anxiety. However, few studies have tested whether CBM works for other disorders Keywords Depression Á Cognitive vulnerability Á characterized by negative cognitive biases such as Cognitive bias modification Á Attention depression. This experiment tested the efficacy of CBM for altering depressogenic self-worth biases. Consistent with hypotheses, results showed that CBM was more effective Introduction than a control condition in teaching participants a more adaptive cognitive style (i.e., increased attention for According to the cognitive theories of depression adaptive self-worth stimuli). Those who successfully (Abramson et al. 1989; Beck 1967), some individuals have learned a more adaptive style reported fewer depressive a cognitive vulnerability that interacts with stress to pro- symptoms and exhibited greater persistence (i.e., less duce depression. Specifically, people are vulnerable to helplessness) on a difficult laboratory task. However, the depression because they tend to view stressful life events as task proved difficult for cognitively vulnerable partici- having negative implications for their future and for their pants; they were not able to maintain their new learning self-worth. Recent research has provided strong support for over the entire course of the CBM training. Rather, their the cognitive vulnerability hypothesis (Abramson et al. negative cognitive bias began to re-emerge after only 20 2002). Prospective studies have consistently found that learning trials. These results indicate that CBM attention cognitive vulnerability interacts with stressful life events to predict the development of depressive symptoms and depressive disorders (Abramson et al. 1999; Alloy et al. 2006; Gibb et al. 2006; Haeffel et al. 2007; Hankin et al. This research was supported in part by an Undergraduate Research 2004; Metalsky and Joiner 1992). Opportunity Grant to Jennifer Hames. Taken together, prior studies indicate that individuals with a cognitive vulnerability are at heightened risk for G. J. Haeffel (&) Á D. C. Rozek Á J. L. Hames Á J. Technow Department of Psychology, University of Notre Dame, depression. Thus, it is important to create interventions that Haggar Hall, Notre Dame, IN 46556, USA can reduce this vulnerability, and in turn, increase resil- e-mail: [email protected] ience to depression. One strategy for reducing cognitive vulnerability is with an intervention such as cognitive- Present Address: J. L. Hames behavioral therapy (CBT), which focuses on helping indi- Department of Psychology, Florida State University, viduals make more adaptive cognitions. Research has Tallahassee, FL, USA shown that CBT is as effective as medication, has no side effects, and may even have a relapse prevention effect J. Technow Department of Psychology, University of Denver, (Hollon et al. 2002). However, CBT is also time intensive Denver, CO, USA (e.g., [12 sessions) and requires trained professionals. 123 Author's personal copy

494 Cogn Ther Res (2012) 36:493–501

These characteristics make dissemination to the general than patients assigned to a control condition. This study public difficult. Ideally, an intervention should be highly provides the first evidence that engaging in a CBM-A task accessible and cost effective. To this end, researchers have can have both long-term and clinically significant effects. begun to test an alternative intervention strategy called More research is still needed, however, to confirm the long- cognitive bias modification (CBM; Koster et al. 2009). term efficacy of CBM-A, and also to determine whether it CBM typically uses a computer task to target a specific can reduce risk for the initial onset of clinically significant pattern of cognitive processing that is thought to confer risk anxiety. for psychopathology (e.g., an attentional bias toward neg- The results of studies testing CBM-A for anxiety are ative stimuli). CBM helps to correct the bias by having promising and indicate that it may be an effective strategy people repeatedly practice a competing stimulus contin- for changing negative cognitive patterns (see Hakamata gency (e.g., an attentional bias toward neutral or positive et al. 2010 for an extensive review). However, few studies stimuli). have tested whether the efficacy of CBM-A translates to The CBM approach has been used almost exclusively in the area of depression (Holmes et al. 2009; Joormann et al. the area of anxiety (Amir et al. 2009; Schmidt et al. 2009). 2009). To date, there are only 2 studies that have applied Research has shown that individuals with high levels of the dot-probe paradigm from the anxiety literature to the anxiety are more likely to attend to threatening stimuli area of depression. In the first study, Wells and Beevers (e.g., threatening words) than individuals with low levels of (2010) trained a sample of dysphoric college students to anxiety. Importantly, these biases appear to precede and attend to neutral rather than negative stimuli. After four predict anxiety reactivity (Salemink et al. 2007; Wilson sessions, they found that students who received the dot- et al. 2006). Thus, the CBM approach has been used to probe training reported fewer depressive symptoms than train participants to attend to neutral rather than threatening students who were in a control group. This was the first stimuli. This typically has been accomplished by modify- study to support the use of an attention-training task for ing an attentional dot-probe task (MacLeod et al. 1986). In reducing depression. However, the results of the second a traditional dot-probe task, a threatening and neutral word study to test CBM-A for depression were not as promising. pair is presented to participants with one word above and In this study, Baert et al. (2010) found that the dot-probe the other below a central fixation point. A dot-probe then training was not effective in altering participants’ negative appears in the position of one of the words immediately cognitive patterns. Moreover, the training actually led to after its presentation. Participants are required to indicate increased symptom severity for some participants (those the position of the dot by using the numeric keypad. The with moderate to severe depressive symptoms). They speed at which participants can detect the dot-probe is an concluded that CBM-A should be counter indicated for indication of attention allocation. individuals with significant levels of depressive symptoms. MacLeod et al. (2002) modified the dot-probe paradigm The mixed findings from these two studies indicate that to create a CBM task that could change the attentional the translation of CBM-A from anxiety to depression might biases found in anxious individuals; we will refer to CBM not be a straightforward one. One explanation for why that focuses specifically on altering attentional biases as CBM-A might be less effective for depression than anxiety CBM-A (attention). Rather than having the dot-probes is that those at risk for depression tend to have limited appear randomly behind both neutral and threatening cognitive resources (e.g., decreased working memory words, this task had the dot-probes appear more often in the capacity). We suspect that low levels of cognitive resources vicinity of the neutral words, thus creating a word-dot may affect a cognitively vulnerable individual’s ability to contingency. After repeated trials, participants learned the learn and consolidate the new cognitive associations taught word-dot contingency and began to attend to neutral rather by CBM-A tasks. than threatening words (because they had learned that this CBM-A is cognitively demanding. It requires cogni- was where the dot-probe was most likely to occur). Using tively vulnerable individuals to suppress their prepotent MacLeod’s CBM-A task (and variations of it), researchers tendency to attend to negative stimuli so that they can have shown that they can reduce attentional biases, and in attend to more neutral stimuli. This process of suppressing turn, reduce anxiety for up to 48 h after completing the a well-established pattern of thinking in order to form a training (Amir et al. 2009; Li et al. 2008; Schmidt et al. new pattern of thinking can be very difficult. There is a 2009; See et al. 2009). Moreover, a recent study by large body of research documenting the trouble that indi- Schmidt et al. (2009) provided evidence that CBM-A can viduals have forming new cognitive associations when even lead to the remission of clinically significant anxiety. there exists a competing and well-established associations They found that patients diagnosed with Social Anxiety in the same domain (e.g., McNeil and Alibali 2005). The Disorder who were randomly assigned to a CBM-A con- ‘‘winner’’ of the competition between old and new thinking dition were significantly more likely to remit from anxiety depends, in part, on the availability of cognitive resources. 123 Author's personal copy

Cogn Ther Res (2012) 36:493–501 495

Cognitive resources are necessary for suppressing estab- would lead to decreases in depressive symptoms and lished knowledge in order to learn new and competing greater persistence on a difficult laboratory task. material. Thus, the effectiveness of CBM-A relies heavily on the ability of the cognitively vulnerable individual to use cognitive resources to inhibit their well-established Method attentional pattern and form a more adaptive pattern.1 We suspect that CBM-A may be difficult for individuals Participants with a cognitive vulnerability because they tend to have reduced cognitive resources. A growing body of research Participants were 61 undergraduates from a mid-sized demonstrates that cognitive vulnerability is associated with private university in the Midwestern United States; they deficits in cognitive capacity, cognitive flexibility, task- were recruited via a volunteer sign-up procedure. Sixty- switching, concentration, attention, and memory (e.g., nine percent of the sample reported their ethnicity as Lyubomirsky et al. 1999; Nolen-Hoeksema et al. 2008; Caucasian, 13% Asian, 10% Hispanic, 3% African Amer- Ray et al. 2005; Watkins and Brown 2002). This work ican, and 5% endorsed an ‘‘other’’ category. One partici- strongly suggests that vulnerable individuals might not pant was excluded from data analyses because of extreme have the cognitive resources needed to maintain the con- responding (i.e., greater than 3 standard deviations) on tingencies learned during CBM-A. Cognitive resources multiple measures; thus, the final sample consisted of 60 have been compared to a muscle (Muraven and Baumeister participants (39 women, 21 men; M age = 20.09). Partic- 2000); their strength and power are finite, and can be ipants were compensated with extra credit points for a depleted with repeated use. Thus, we predicted that CBM-A psychology course. might eventually wear-out the already weakened ‘‘cognitive muscle’’ of vulnerable individuals. Once the resources Measures are depleted by CBM-A then the training effects should begin to diminish and the negative attentional biases Cognitive Bias Modification Task re-emerge. In summary, there is a paucity of research on CBM-A The CBM-A task was modeled after the attentional dot- for depression, and further research is sorely needed. Thus, probe task created by MacLeod et al. (2002). In this task, a the goal of the current experiment was to add to the body of word pair (one adaptive and one maladaptive) is presented literature on CBM-A for depression as well as to help to to participants with one word above and the other below a reconcile inconsistencies in this area. We tested three central fixation point. A dot-probe then appears in the hypotheses. First, we hypothesized that participants would position of one of the words and participants are required to exhibit lower levels of cognitive vulnerability after CBM-A detect the location of a dot-probe as quickly as possible by training. Consistent with prior research, we reasoned that pressing either up or down on an input box. We made one participants would be able to learn the contingencies they modification to this basic CBM-A task. Specifically, we practiced during CBM-A. However, we also hypothesized included a priming stimulus before the presentation of the that the efficacy of CBM-A would deteriorate over time, word pair. We added a prime because the cognitive theo- particularly for participants with a cognitive vulnerability. ries of depression propose a vulnerability-stress model. The CBM-A task requires individuals to repeatedly control This means that cognitive vulnerability is most likely to or inhibit their prepotent tendency to attend to negative emerge in the presence of stress. In other words, cognitive inferences (i.e., inhibit their well-established cognitive biases lie dormant until activated (or primed) by the req- vulnerability). We hypothesized that this resource-inten- uisite negative context. Thus, it is imperative that indi- sive process would fatigue participants with a cognitive viduals be primed to ensure that their negative cognitive vulnerability over time, and in turn, allow for their more biases are activated and, in turn, amenable to modification. established negative pattern of thinking to re-emerge. At the start of each trial, participants were presented Finally, we hypothesized that CBM-A (when successful) with a black central fixation cross (on a white background) would lead to changes in participants’ mood and actual for 1,000 ms. Then, participants were primed (750 ms) behavior. Specifically, we predicted that CBM-A training with one of 20 negative scenarios (e.g., ‘‘rejected by friend’’; see appendix for all scenarios used). Next, the word-pair, which consists of one adaptive and one maladaptive self-worth inference (see appendix for adjectives 1 It is important to note that even associative and implicit learning used), was presented for 1,000 ms. The adaptive and tasks (similar to the dot-probe training task) require cognitive resources and working memory (Carter et al. 2003; Jonides et al. maladaptive word lists were matched for average length 1998; Reber and Kotovsky 1997; Ridderinkhof et al. 2004). and frequency of use (Carroll et al. 1971). After the word 123 Author's personal copy

496 Cogn Ther Res (2012) 36:493–501 pair disappeared, a dot appeared in the location of one of to solve a series of anagrams. They were told to hit the the words and remained there until the participant indicated space bar once they solved the anagram or, if they could its location by pressing a top or bottom button on an input not solve the anagram, then they should hit the space bar to box. In order to create the contingency between negative receive the next anagram. The time spent trying to solve scenarios and adaptive words, the dot-probe was presented the three anagrams was used as a behavioral measure of behind the adaptive word approximately 95% of the time. persistence (i.e., resistance to helplessness). The control condition did not create a contingency; rather, in this condition, the dot-probe was equally likely to appear Depressive Symptoms behind the adaptive and maladaptive words (i.e., the dob- probe was presented behind the adaptive word 50% of the The Mood and Anxiety Symptom Questionnaire (MASQ; time. Participants completed 2 blocks of 40 trials (80 trials Watson et al. 1995) was used to assess levels of depressive total) in both the training and control conditions. symptoms at baseline and upon completion of the CBM-A Participants’ response latencies from only the correct task. The MASQ is a self-report questionnaire that assesses trials were analyzed. Consistent with prior research, self- symptoms specific to depression and anxiety based on the corrected trials as well as reaction times of less than tripartite theory of anxiety and depression (Clark and 100 ms and more than 1,000 ms were removed (Joormann Watson 1991). In this study, we used the anhedonic sub- and Gotlib 2007); omitted trials accounted for a small scale, which contains 22 items that assess symptoms percentage of all trials (\6%). Average reaction times for hypothesized to be specific to depression such as low matched (dot-probe appeared in the same location that the positive affect (example items: ‘‘Thought about death or maladaptive word appeared) and mismatched trials (dot- suicide,’’ ‘‘Felt really up or lively,’’ ‘‘Felt cheerful’’). In probe appeared in the location opposite of where the this study, participants were instructed to rate the items maladaptive word appeared) were calculated for each set of with regard to how the felt ‘‘right now’’ [it is important to twenty trials (0–20, 21–40, 41–60, and 61–80). The aver- note that a number of studies have shown that the MASQ is age matched score was then subtracted from the average still a reliable and valid measure of depressive symptoms mismatched score at each interval to create a measure of even when the instructions are changed to assess either cognitive bias with greater reaction time difference scores shorter or longer durations than originally intended (e.g., indicating a greater bias for attending to maladaptive Hankin 2005; Haeffel and Mathew 2010; Sarin et al. stimuli. 2005)]. The MASQ has demonstrated good reliability and validity in previous research (Watson et al. 1995). Cognitive Vulnerability to Depression Procedure The Cognitive Style Questionnaire (CSQ; Haeffel et al. 2008; Alloy et al. 2000) was used to assess participants’ At baseline, participants completed a brief demographics baseline levels of cognitive vulnerability to depression. The questionnaire and measures of cognitive vulnerability CSQ is widely used self-report questionnaire that assesses (CSQ) and depressive symptoms (MASQ). They were then the cognitive vulnerability factor featured in the hope- randomly assigned to either a CBM-A or control inter- lessness theory of depression. The CSQ assesses partici- vention. After the intervention, participants completed the pants’ causal attributions for the 12 hypothetical negative measure of depressive symptoms (MASQ) and a difficult events on dimensions of stability and globality; in addition, laboratory task (anagram failure task; Haeffel et al. 2007; participants rate the probable consequences of each event MacLeod et al. 2002; Mogg et al. 1990). Participants were and the self-worth implications of each event. Mean-item then debriefed about the unsolvable nature of many of the scores can range from 1 to 7, with higher scores reflecting anagrams and the purpose of the experiment. more negative cognitive styles. The CSQ has good internal consistency, reliability, and validity (Haeffel et al. 2008). Results Difficult Laboratory Task Three hypotheses were tested. First, we hypothesized that Participants were asked to solve a set of 10 difficult and most participants would be able to learn the contingencies unsolvable anagrams (Tresselt and Mayzner 1966). Par- they practiced during CBM-A. However, we also hypoth- ticipants first completed two example anagrams (that were esized that the efficacy of CBM-A would deteriorate over solvable) and were told that, ‘‘being good at solving ana- time for participants with a cognitive vulnerability. Finally, grams requires good vocabulary and good memory.’’ Par- we hypothesized that CBM-A (when successful) would lead ticipants were then instructed that they had unlimited time to changes in mood and behavior. Specifically, we predicted 123 Author's personal copy

Cogn Ther Res (2012) 36:493–501 497 that CBM-A training would lead to decreases in depressive 20 symptoms and greater persistence on a difﬁcult laboratory 10 task. Participants in the CBM-A and control conditions did not differ signiﬁcantly on any of the baseline variables (age, 0 gender, cognitive vulnerability, and depressive symptoms). -10

Effect of CBM-A Training on Cognitive Biases -20 -30 To test the efficacy of CBM-A training over time, we conducted a 2 (condition: CBM-A or control) 9 2 (cog- -40 Reaction Time (ms) nitive vulnerability: high or low as determined by a median -50 split on the CSQ) 9 4 (trial set: 0–20, 21–40, 41–60, 61–80) mixed-factor ANCOVA. The dependent variable -60 was cognitive bias (i.e., the tendency to attend to adaptive -70 relative to maladaptive stimuli during the CBM-A task). 0-20 21-40 41-60 61-80 Taking a conservative analytic approach, MASQ score at Trial baseline was used as covariate in analyses to control for Training-Vulnerable Control-Vulnerable any individual differences in initial level of depressive Training-NonVulnerable Control-NonVulnerable symptoms. As predicted, there was a significant between- subjects effect for condition, F(1, 52) = 13.79, P \ .001, Fig. 1 Cognitive bias as a function of condition (CBM-A training or g2 = .21. Corroborating prior research, participants in the control), cognitive vulnerability (vulnerable vs. non-vulnerable), and p time (trial number). Note that smaller reaction time values indicate a CBM-A condition were able to learn the cognitive asso- greater bias for adaptive relative to maladaptive stimuli ciations taught by the task; overall, they were much more likely to attend to adaptive stimuli relative to maladaptive By parsing the trials more closely together (every 10 stimuli (M reaction time =-27.41 ms, SE = 5.68 ms) trials), we found clear evidence of learning in the cogni- than participants in the control condition (M reaction tively vulnerable group of participants. Corroborating the time = .54 ms, SE = 4.83 ms; note that smaller values initial results, there was a significant interaction of condi- indicate a greater bias for adaptive stimuli). tion, cognitive vulnerability, and trial set, F(3,135) = 2.60, However, as predicted, the main effect of condition was 2 P = .056, gp = .05. As Fig. 2 shows, individuals with a qualified by the interaction of condition, cognitive vul- cognitive vulnerability quickly acquired the new associa- nerability, and trial set, F(3, 156) = 4.52, P = .005, tions taught by CBM-A, however, this learning began to 2 gp = .08. As shown in Fig. 1, the efficacy of CBM-A for deteriorate after about 20 training trials. Interaction con- cognitively vulnerable individuals weakened considerably trasts confirmed the pattern of the interaction, F(1, 45) = over time whereas the efficacy of CBM-A for non-vul- 2 4.67, P = .04, gp = .09. nerable individuals remained steady across trials. Interac- tion contrasts confirmed the linear pattern of results, F(1, Effect of CBM-A Training on Mood and Behavior 2 52) = 9.05, P = .004, gp = .15 (neither the quadratic nor any of the higher-order trends were significant). We tested the hypothesis that CBM-A training would lead to The figure also showed an unexpected result for the changes in depressive symptoms and helplessness behavior. training condition. In contrast to non-vulnerable individu- Specifically, we tested whether CBM-A outcome (i.e., the als in the training condition, vulnerable individuals in this cognitive bias score for trials 60–80) predicted levels of condition did not show clear evidence of learning the depressive symptoms upon completion of training and the contingencies taught by CBM-A. At first blush, it appears amount of time spent on a difficult laboratory task. We that vulnerable individuals in the training condition actu- conducted an ANCOVA with CBM-A training end-state ally started with a positivity bias whereas non-vulnerable (successful or non-successful as operationalized by a median learned the positivity bias over the first 20 trials (which split of the cognitive bias score for trials 60–80) as the then remained steady over the duration of the task). independent variable and MASQ score post-intervention and However, another explanation for this discrepancy is that time spent on the laboratory stressor as the dependent vari- the trials were averaged in groups (every 20 trials) that ables, respectively. MASQ and CSQ scores at baseline were were too large to detect rapid learning. To test this expla- used as covariate in all analyses to control for any individual nation, we divided the data from trials 0–40 into groups of differences in initial level of depressive symptoms or cog- 10 rather than groups of 20. nitive vulnerability. As predicted, there was a significant 123 Author's personal copy

498 Cogn Ther Res (2012) 36:493–501

20 56 55 54 0 53 52 -20 51 50 49

-40 MASQ Score 48 47 -60 46

Reaction Time (ms) 45 Successful CBM Non-Successful CBM -80 25

-100 0-10 11-20 21-30 31-40 20 Trial Training-Vulnerable Control-Vulnerable 15

Training-NonVulnerable Control-NonVulnerable 10

Fig. 2 Cognitive bias as a function of condition (CBM-A training or Time (Minutes) control), cognitive vulnerability (vulnerable vs. non-vulnerable), and 5 time (trial number). Note that smaller reaction time values indicate a greater bias for adaptive relative to maladaptive stimuli 0 Successful CBM Non-Successful CBM

Fig. 3 Top panel: Level of depressive symptoms upon completion of effect of CBM-A end-state on both depressive symptoms (at CBM-A training as function of CBM-A outcome (successful vs. non- 2 successful). Bottom panel: Time spent on a difficult laboratory task as the level of a trend; F[1, 23] = 3.61, P = .07, gp = .14) and a function of CBM-A outcome (successful vs. non-successful) time spent on the laboratory task (F[1, 22] = 4.79, P = .04, 2 gp = .18) with effect sizes in the medium to large range. As shown in Fig. 3, participants who exhibited the contingencies taught by the CBM-A task reported significantly lower The results of the experiment also highlight potential levels of depressive symptoms and spent significantly more problems with CBM-A for depression. Our findings are the time on the difficult laboratory task than participants who did first to show that cognitively vulnerable individuals might not successfully learn the CBM-A contingencies (even after struggle to maintain newly learned associations over the controlling for initial levels of depressive symptoms and duration of a CBM-A training session. After approximately cognitive vulnerability). 20 trials, the negative cognitive patterns of vulnerable participants began to re-emerge. These findings corroborate those of Baert et al. (2010) who recently found that CBM- Discussion A was ineffective in altering cognitive biases (and actually led to increased depression in some participants). The This study contributes to the small, but important body of results also add to a large body of research in cognitive research testing the efficacy of CBM-A tasks for depres- psychology, which shows that people have great difficulty sion. Consistent with hypotheses, results showed that forming new cognitive associations when there exists a CBM-A was more effective than a control condition in competing and well-established associations in the same altering attentional biases. Moreover, those who exhibited domain (McNeil and Alibali 2005; Diamond and Kirkham a more adaptive attentional style after CBM-A training 2005; Zevin and Seidenberg 2002). Our results indicate reported fewer depressive symptoms and had greater per- that CBM-A training should be provided, at least initially, sistence (i.e., less helplessness) on a difficult laboratory in low-doses. It is important to remember, however, that we task. These results map onto the findings from the anxiety tested participants after a single 20-min training session. literature, corroborate the recent findings of Wells and With repeated training (e.g., everyday for 1 month), it may Beevers (2010), and provide evidence that CBM-A has be that the contingencies learned during CBM-A could potential for helping reduce cognitive vulnerability to be strengthened, and over time, require less cognitive depression. resources (i.e., become more automatic).

123 Author's personal copy

Cogn Ther Res (2012) 36:493–501 499

The current findings also provide a clue for how to might be counterproductive (at least initially). In addition, reconcile the different results found by the previous two analyses aimed at understanding trial-by-trial learning can CBM-A studies for depression. Wells and Beevers (2010) help to determine the number of trials and sessions nec- found that CBM-A was effective in reducing depressive essary for optimal CBM-A efficacy. A final strength of this symptoms whereas Baert et al. (2010) found that CBM-A study was the use of a highly specific measure of depres- might actually lead to increases in depressive symptoms. sive symptoms (the MASQ). Previous studies have been Baert and colleagues attributed the opposite findings to limited by symptom measures (e.g., the Beck Depression differences in participants’ levels of depression. Partici- Inventory) that lack specificity and are saturated with high pants in the Wells and Beevers study had lower levels of levels of negative affect (Clark and Watson 1991). Using depression than those in the Baert and colleagues study. the MASQ enabled us to examine the specificity of our However, our results suggest that the discrepant results results to the depressive symptoms. That said, it is could be also be due to participants having different levels important to note that measures such as the MASQ (and the of cognitive vulnerability. Cognitive vulnerability is con- Beck Depression Inventory) were not designed to measure sistently and robustly correlated with depression levels. short-term changes in depressive symptoms. Rather, these Thus, it is possible (or even likely) that participants in the measures are typically used to measure symptom change Wells and Beevers’ study had lower levels of cognitive over the course of weeks, months, and event years. We vulnerability than those in the Baert study. This means that chose the MASQ because it was used in one of the two some of the participants in the Wells and Beevers’ study prior CBM-A studies of depression (Baert et al. 2010), it is probably did not have the pre-existing and highly engrained specific to depression, and it contains items (mood adjec- negative associations that make learning a CBM-A con- tives) that are more amenable to short-term change than the tingency difficult. Participants’ higher levels of cognitive items on the Beck Depression Inventory (e.g., change in vulnerability in the Baert et al. study might also explain weight). Both our study and Baert et al. (2010) support the why they found that CBM-A was detrimental for some use of the MASQ as a short-term measure of change; participants. There is some evidence that suppressing a however, future work in this area is still needed to validate highly engrained pattern of thinking can actually lead to an these more long-term measures for use in short-term increase in the thoughts that are being suppressed. For experimental designs. example, Erskine et al. (in press) recently found that when Limitations of the current study should also be noted. smokers suppressed thoughts about smoking there was an First, it would be premature to make conclusions about the initial decrease in smoking. However, as cognitive effect of CBM-A on clinically significant forms of resources diminished over time, there was a rebound effect depression because the current study used a relatively and individuals exhibited an increase in smoking. Thus, it health sample and we only assessed depressive symptoms is possible that once the cognitive resources of the highly (rather than disorder). Second, the study examined college vulnerable individuals in Baert et al.’s study were depleted freshmen. Although freshmen are ideal for testing pre- by CBM-A they actually experienced an increase in the vention interventions because they are at the peak age for negative cognitions that they were previously suppressing. developing depression (Hankin et al. 1998) and are likely This increase in negative cognitions could have led to to experience high levels of stress, the results may not increased depression levels. generalize to community samples. Finally, it is important to The current study had both strengths and limitations. For note the preliminary nature of these findings and the need example, a significant strength of this study is that it is one for replication in a larger sample. Indeed, this is the first of the few to test whether or not the efficacy of CBM-A study to examine the moderating effects of cognitive vul- translates from anxiety to depression. In addition, we used nerability on trial-by-trial learning in CBM-A for depres- an experimental design with multiple outcome measures sion. Moreover, our CBM-A task did not map directly onto (i.e., self-report and behavior). The study also provided the CBM-A task used by the prior two studies in this area. some of the first support for the novel hypothesis that Unlike prior work, we used a priming stimulus in our task learning would deteriorate for cognitively vulnerable to ensure the activation of cognitive biases during our task. individuals over time. These results highlight the impor- Our use of a prime provided a greater level of theoretical tance of examining the trial-by-trial learning in CBM-A validity as well as a more rigorous test of the CBM-A tasks. If we had only tested the main effect of condition, we hypothesis; however, it also does not allow us to directly would have overlooked important information about the equate our results to the two prior studies in this area. Our efficacy of CBM-A for individuals with cognitive vulner- CBM-A task was likely more difficult for cognitively ability. By using a fine-grained analysis, we were able to vulnerable individuals than prior CBM-A tasks because identify a subgroup of individuals (i.e., those with a cog- their cognitive biases were continually activated by the nitive vulnerability) for which long sessions of CBM-A prime during our task. However, this is likely what happens 123 Author's personal copy

500 Cogn Ther Res (2012) 36:493–501 in the real world when encountering a stressful life event. In I. H. Gotlib & C. L. Hammen (Eds.), Handbook of depression. Thus, we contend that by not using a prime, there is danger New York: Guilford. Abramson, L. Y., Alloy, L. B., Hogan, M. E., Whitehouse, W. G., in overestimating the positive effects CBM-A as it is Donovan, P., Rose, D., et al. (1999). Cognitive vulnerability to unclear if the training will hold up when participants need depression: Theory and evidence. Journal of Cognitive Psycho- it most—during times of stress. therapy: An International Quarterly, 13, 5–20. In conclusion, our results suggest that CBM-A has the Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopeless- ness depression: A theory-based subtype of depression. Psycho- potential to be an easily disseminated and effective inter- logical Review, 96, 358–372. vention for reducing cognitive vulnerability to depression. Alloy, L. B., Abramson, L. Y., Hogan, M. E., Whitehouse, W. G., Rose, However, there are some important caveats to consider. D. T., Robinson, M. S., et al. (2000). The Temple-Wisconsin CBM-A is a learning process, which requires the effortful cognitive vulnerability to depression project: Lifetime history of axis I psychopathology in individuals at high and low cognitive risk suppression of a pre-potent response. This suppression is for depression. Journal of Abnormal Psychology, 109, 403–418. cognitively taxing and can affect learning over time. In our Alloy, L. B., Abramson, L. Y., Whitehouse, W. G., Hogan, M. E., experiment, cognitively vulnerable participants quickly Panzarella, C., & Rose, D. T. (2006). Prospective incidence of learned the contingencies taught by CBM-A; unfortunately, first onsets and recurrences of depression in individuals at high and low cognitive risk for depression. Journal of Abnormal this new learning faded rapidly over repeated trials. It will Psychology, 115, 145–156. be important for future studies to test whether the efficacy Amir, N., Beard, C., Burns, M., & Bomyea, J. (2009). Attention of CBM-A can be better sustained with repeated training. modification program in individuals with generalized anxiety disorder. Journal of Abnormal Psychology, 118, 28–33. Baert, S., De Raedt, R., Schacht, R., & Koster, E. H. W. (2010). Attentional bias training in depression: Therapeutic effects Appendix depend on depression severity. Journal of Behavior Therapy and Experimental Psychiatry, 41, 265–274. Negative Scenarios Beck, A. T. (1967). Depression: Clinical, experimental, and theoretical aspects. New York: Harper & Row. Carroll, J. B., Davies, P., & Richman, B. (1971). Word frequency Negative scenarios: failed a class presentation, rejected by book. New York: American Publishing Co., Inc. friend, got low grade on a paper, couldn’t finish your Carter, R. M., Hofstotter, C., Tsuchiya, N., & Koch, C. (2003). homework, partner breaks up with you, you are unhappy, Working memory and fear conditioning. Proceedings of the National Academy of Science, 100, 1399–1404. job performance criticized, betrayed by family member, Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and GPA is low, parents disapproved of your behavior, got in depression: Psychometric evidence and taxonomic implications. serious trouble, felt unattractive, not in a romantic rela- Journal of Abnormal Psychology, 100, 316–336. tionship, fired from job, failed to achieve goal, people not Diamond, A., & Kirkham, N. (2005). Not quite as grown-up as we like to think—parallels between cognition in childhood and interested in you, have no close friends, failed a test, crit- adulthood. Psychological Science, 16, 291–297. icized by professor, rejected by romantic partner Erskine, J.A., Georgiou, G.J., & Kvavilashvili, L. (in press). I suppress, therefore I smoke: Effects of thought suppression on Maladaptive Adjectives smoking behavior. Psychological Science. Gibb, B. E., Beevers, C. G., Andover, M. S., & Holleran, K. (2006). The hopelessness theory of depression: A prospective multi- Maladaptive adjectives: dumb, failure, incompetent, inad- wave test of the vulnerability-stress hypothesis. Cognitive equate, incapable, inept, inferior, loser, nobody, pathetic, Therapy and Research, 30, 763–772. pitiful, stupid, unable, unfit, unimportant, unlovable, Haeffel, G. J., Gibb, B. E., Abramson, L. Y., Alloy, L. B., Metalsky, G. I., Joiner, T., et al. (2008). Measuring cognitive vulnerability unwanted, useless, weak, worthless to depression: Development and validation of the cognitive style questionnaire. Clinical Psychology Review, 28, 824–836. Adaptive Adjectives Haeffel, G. J., Abramson, L. Y., Brazy, P., Shah, J., Teachman, B., & Nosek, B. (2007). Explicit and implicit cognition: A preliminary test of a dual-process theory of cognitive vulnerability. Behav- Adaptive adjectives: able, capable, competent, confident, iour Research and Therapy, 45, 1155–1167. desirable, determined, hopeful, important, inspire, lovable, Haeffel, G. J., & Mathew, A. R. (2010). Inside thoughts and outside motivated, proficient, skillful, smart, talented, useful, influences: Cognitive vulnerability moderates the effect of valuable, winner, worthy decreases in perceived social support on depressive symptoms. Journal of Social and Clinical Psychology, 29, 283–302. Hakamata, Y., Lissek, S., Bar-Haim, Y., Britton, J. C., Fox, N. A., Leibenluft, E., et al. (2010). Attention bias modification treatment: A meta-analysis toward the establishment of novel References treatment for anxiety. Biological Psychiatry, 68, 982–990. Hankin, B. L. (2005). Childhood maltreatment and psychopathology: Abramson, L. Y., Alloy, L. B., Hankin, B. L., Haeffel, G. J., MacCoon, Prospective tests of attachment, cognitive vulnerability, and D. G., & Gibb, B. E. (2002). Cognitive vulnerability-stress models stress as mediating processes. Cognitive Therapy and Research, of depression in a self-regulatory and psychobiological context. 29, 645–671.

123 Author's personal copy

Cogn Ther Res (2012) 36:493–501 501

Hankin, B. L., Abramson, L. Y., Miller, N., & Haeffel, G. J. (2004). Mogg, K., Mathews, A., Bird, C., & MacGregor-Morris, R. (1990). Cognitive vulnerability-stress theories of depression: Examining Effects of stress and anxiety on the processing of threat stimuli. affective specificity in the prediction of depression versus Journal of Personality and Social Psychology, 59, 1230–1237. anxiety in three prospective studies. Cognitive Therapy and Muraven, M., & Baumeister, R. F. (2000). Self-regulation and Research, 28, 309–345. depletion of limited resources: Does self-control resemble a Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P., McGee, R., muscle? Psychological Bulletin, 126, 247–259. & Angell, K. E. (1998). Development of depression from Nolen-Hoeksema, S. N., Wisco, B. E., & Lyubomirsky, S. (2008). preadolescence to young adulthood: Emerging gender differ- Rethinking rumination. Perspectives on Psychological Science, ences in a 10-year longitudinal study. Journal of Abnormal 3, 400–423. Psychology, 107, 128–140. Ray, R. D., Ochsner, K. N., Cooper, J. C., Robertson, E. R., Gabrieli, J. Hollon, S. D., Thase, M. E., & Markowitz, J. C. (2002). Treatment D. E., & Gross, J. J. (2005). Individual differences in trait rumination and prevention of depression. Psychological Science in the and the neural systems supporting cognitive reappraisal. Cognitive, Public Interest, 3, 39–77. Affective & Behavioral Neuroscience, 5, 156–168. Holmes, E. A., Lang, T. J., & Shah, D. M. (2009). Developing Reber, P. J., & Kotovsky, K. (1997). Implicit learning in problem interpretation bias modification as a ‘‘cognitive vaccine’’ for solving: The role of working memory capacity. Journal of depressed mood: Imagining positive events makes you feel Experimental Psychology: General, 126, 178–203. better than thinking about them verbally. Journal of Abnormal Ridderinkhof, K. R., Ullsperger, M., Crone, E. A., & Nieuwenhuis, S. Psychology, 118, 76–88. (2004). The role of the medial frontal cortex in cognitive control. Joormann, J., & Gotlib, I. H. (2007). Selective attention to emotional Science, 306, 443–447. faces following recovery from depression. Journal of Abnormal Salemink, E., van den Hout, M., & Kindt, M. (2007). Trained Psychology, 116, 80–85. interpretive bias and anxiety. Behaviour Research and Therapy, Joormann, J., Hertel, P. T., LeMoult, J., & Gotlib, I. H. (2009). 45, 329–340. Training forgetting of negative material in depression. Journal of Sarin, S., Abela, J. R. Z., & Auerbach, R. P. (2005). The response Abnormal Psychology, 118, 34–43. styles theory of depression: A test of specificity and causal Jonides, J., Smith, E. E., Marshuetz, C., Koeppe, R. A., & Reuter- mediation. Cognition and Emotion, 19, 751–761. Lorenze, P. A. (1998). Inhibition in verbal working memory Schmidt, N. B., Richey, J. A., Buckner, J. D., & Timpano, K. R. revealed by brain activation. Proceedings of the National (2009). Attention training for generalized social anxiety disorder. Academy of Sciences, 95, 8410–8413. Journal of Abnormal Psychology, 118, 5–14. Koster, E. H. W., Fox, E., & MacLeod, C. (2009). Introduction to the See, J., MacLeod, C., & Bridle, R. (2009). The reduction of anxiety special section on cognitive bias modification in emotional vulnerability through the modification of attentional bias: A real disorders. Journal of Abnormal Psychology, 118, 1–4. world study using a home-based cognitive bias modification Li, S., Tan, J., Qian, M., & Liu, X. (2008). Continual training of procedure. Journal of Abnormal Psychology, 118, 65–75. attentional bias in social anxiety. Behaviour Research and Tresselt, M. E., & Mayzner, M. (1966). Normative times for a sample Therapy, 46, 905–912. of 134 solution words and 378 associated anagrams. Psycho- Lyubomirsky, S., Tucker, K. L., Caldwell, N. D., & Berg, K. (1999). nomic Monograph Supplements, 1, 293–298. Why ruminators are poor problem solvers: clues from the Watkins, E., & Brown, R. G. (2002). Rumination and executive phenomenology of dysphoric rumination. Journal of Personality function in depression: an experimental study. Journal of and Social Psychology, 77, 1041–1060. Neurology, Neurosurgery and Psychiatry, 72, 400–402. MacLeod, C., Mathews, A., & Tata, P. (1986). Attentional bias in Watson, D., Clark, L. A., Weber, K., Assenheimer, J. S., Strauss, M. emotional disorders. Journal of Abnormal Psychology, 95, 15–20. E., & McCormick, R. A. (1995). Testing a tripartite model: II. MacLeod, C., Rutherford, E., Campbell, L., Ebsworthy, G., & Holker, Exploring the symptom structure of anxiety and depression in L. (2002). Selective attention and emotional vulnerability: student, adult, and patient samples. Journal of Abnormal Assessing the causal basis of their association through the Psychology, 104, 15–25. experimental manipulation of attentional bias. Journal of Wells, T. T., & Beevers, C. G. (2010). Biased attention and Abnormal Psychology, 111, 107–123. dysphoria: Manipulating selective attention reduces subsequent McNeil, N. M., & Alibali, M. W. (2005). Why won’t you change your depressive symptoms. Cognition & Emotion, 24, 719–728. mind? Knowledge of operational patterns hinders learning and Wilson, E. J., MacLeod, C., Mathews, A., & Rutherford, E. M. performance on equations. Child Development, 76, 883–899. (2006). The causal role of interpretive bias in anxiety reactivity. Metalsky, G. I., & Joiner, T. E. (1992). Vulnerability to depressive Journal of Abnormal Psychology, 115, 103–111. symptomatology: A prospective test of the diathesis-stress and Zevin, J. D., & Seidenberg, M. S. (2002). Age of acquisition effects in causal mediation components of the hopelessness theory of word reading and other tasks. Journal of Memory and Language, depression. Journal of Personality and Social Psychology, 63, 47, 1–29. 667–675.

123