Attentional bias training in depression: Therapeutic effects depend on depression severity

Abstract

Depressed individuals show maintained attention to negative information and reduced attention for positive information. Selective biases in information processing are considered to have an important role in the origin, maintenance and recurrence of depressive episodes. In two experiments we investigated the effects of attentional bias manipulation on mood and depressive symptoms. In experiment 1 we investigated the effects of attentional bias manipulation compared to a control procedure in a sample of dysphoric students (N = 48) showing mild to severe levels of depressive symptoms. In experiment 2 we investigated the same attentional training procedure in a sample of depressed in- and outpatients (N = 35). Mild improvements on symptom severity were observed in students showing mild depressive symptoms. However, in students showing moderate to severe depressive symptoms, depressive symptoms increased after the training. No beneficial effects of training on top of therapy and/or medication were found in depressed patients. These results indicate that therapeutic effects of attentional bias modification might be dependent on depression severity.

Introduction

Based on cognitive models of depression it has been argued that information processing is an important determining factor of negative mood and depressive symptoms (Beck, 2008, Beevers, 2005). Research based on these models has consistently demonstrated the presence of specific mood-congruent biases in memory and reasoning (Mathews & MacLeod, 2005). Initially, studies investigating attentional bias in depression were less consistent (Williams, Watts, MacLeod, & Mathews, 1997). Yet, over the years, cognitive-experimental tasks have become better tuned towards the distinctive features of attentional bias in depression, showing that attentional biases in depression can only be observed under specific task conditions (Joormann, Yoon, & Zetsche, 2007).

Attentional bias in depression is found to be distinct from attentional bias in anxiety. Whereas anxiety is characterized by early vigilance towards threat (e.g. Koster, Verschuere, Crombez, & Van Damme, 2005), depression would be mainly related to difficulties with disengaging attention away from negative information once attention has been drawn to the depressogenic content (e.g. Bradley, Mogg, & Lee, 1997). A growing body of evidence has demonstrated mood-congruent attentional bias when negative material is self-referent and when the cues are presented for long, but not short durations (≥1s) (e.g., Gotlib et al., 2004). Research recording eye movement corroborate these conclusions by the observation of maintained gaze on negative pictures compared to control pictures, and an absence of initial shifting to negative pictures (e.g. Caseras, Gamer, Bradley, & Mogg, 2007).

Recently, an emotional modification of the spatial cueing task (Posner, 1980) has been developed to investigate maintained attention for negative material in depression. To investigate maintained attention with this paradigm, cues are presented for longer duration (>500 ms), which has been associated with the “inhibition of return” (IOR) effect. IOR has been conceived as a mechanism inhibiting the location of a previously attended stimulus in favour of new locations. Fox et al. (Fox, Russo, & Dutton, 2002) argued that if individuals display maintained attention towards particular cues, the IOR effect will be reduced in magnitude or not be present at all. This effect has been referred to as an “enhanced cue validity effect” (Koster, De Raedt, Goeleven, Franck, & Crombez, 2005). Using this paradigm, attention to negative information has been found in dysphoric individuals (Koster, De Raedt, et al., 2005) as well as in depressed patients (Leyman, De Raedt, Schacht, & Koster, 2007). Besides the observation of maintained attention towards negative material, a number of studies have found reduced attention towards positive information (e.g. Matthews & Antes, 1992).

In cognitive theories, information processing biases are considered to be a possible risk factor for the aetiology, maintenance, and recurrence of depressive episodes (Clark, Beck, & Alford, 1999). However, assumptions on a causal relationship between attentional biases and depression can only be confirmed by research that goes beyond correlational designs. With regard to the contributing role of attentional bias in depression there are a number of research approaches that can provide useful information. One approach is the use of prospective designs to determine whether attentional bias precedes the development of a depressive episode. A number of prospective studies suggest that attentional bias is associated with emotional reactivity and precedes the development of anxiety and depression. For instance, in a seminal study by MacLeod and Hagan (1992), an emotional Stroop task was administered to a group of women before they underwent a gynaecological examination. It was found that an attentional bias for subliminally presented negative information was the best predictor of emotional reactivity to a stressful outcome. In the context of depression, Beevers and Carver (2003) demonstrated that attentional bias interacted with intervening life stress to predict higher scores on depression seven weeks later. Furthermore, several studies have been able to demonstrate a mood-congruent attentional bias in patients following recovery as well as in offspring at risk for the development of depression (Joormann and Gotlib, 2007, Joormann et al., 2007).

Importantly, no strict conclusions concerning the causal hypothesis can be drawn from prospective studies because the influence of a third factor that could account for the established associations cannot be ruled out. It is possible that participants demonstrating an attentional bias at pre-testing differed from the individuals who did not demonstrate an attentional bias on variables that were not controlled for. The only way to adequately address the causal hypothesis is by experimentally manipulating attentional bias in order to test whether variations in the bias influence the emotional disorder. To this end, cognitive bias modification (CBM) procedures have been developed (MacLeod et al., 2002, Mathews and MacLeod, 2002). These procedures intend to train the participant to orient towards or away from specific emotional stimuli through systematic manipulation of contingencies between the valence of the stimulus and the location of a target to which the participant has to respond. One paradigm that has been used is the dot probe task (MacLeod, Mathews, & Tata, 1986). In this task pairs of cues are presented, one emotionally valenced and one neutral, followed by a target (dot probe) that can be presented at the same (congruent trial) or at the opposite location (incongruent trial) of the valenced cue. Participants are asked to respond as quickly as possible to the location of the target without sacrificing accuracy. To induce attentional bias, MacLeod et al. (2002) created a task in which the dot probe appeared at the same location of the negative cue in every trial. This prompts the participant to attend to the negative information. To reduce attentional bias the dot probe was set to always appear on the opposite side of a threatening stimulus, prompting participants to divert their attention away from negative information.

Several studies investigating this type of attentional training procedure have been able to successfully modify the cognitive bias of interest and found significant effects on emotional reactivity to induced or to real life stressors (for a review see MacLeod, Koster, & Fox, 2009). These results confirm the hypothesized causal relationship between attentional bias and vulnerability factors associated with the development of emotional disorders. For example, in the seminal study of MacLeod et al. (2002), inducing versus reducing attentional bias led to a differential effect on emotional reactivity to a laboratory stressor. Self-reported emotional responses indicated that participants in the reduction training reacted less emotionally to this stressful situation than participants in the induction condition.

Besides being of theoretical interest, CBM research may also hold promise for therapeutic practice. There are several reasons to examine the potential of CBM in depression. First, several outcome studies report that the rate of depression relapse in recovered patients runs up to 72% depending on the number of past episodes. (e.g. Kessler, Zhao, Blazer, & Swartz, 1997). This could indicate that cognitive vulnerability factors remain unchanged despite therapy. Based on recent studies showing that after recovery depressed patients are still characterized by selective attention to emotional faces (Joormann & Gotlib, 2007), attentional bias can be conceived of as a stable cognitive vulnerability factor. It has been argued that this attentional bias can enhance the risk for relapse through its' effects on emotional reactivity and thought processes such as rumination (for a review, De Raedt & Koster, 2010). Therefore, it seems important to address whether CBM procedures could be used as an intervention to tackle these biases. Second, in anxiety disorders promising findings have been obtained with extended attentional training programs. In a sample of patients with generalized anxiety disorder (GAD), Amir, Beard, Cobb, and Bomyea (2009) conducted a training study using the modified dot probe task with words for 8 sessions within four weeks. Diagnostic assessment by means of a structural clinical interview under double-blind conditions immediately after the final training showed that 58% of participants in the training condition no longer met diagnostic criteria for GAD, compared to 17% in the control condition. Schmidt et al. (Schmidt, Richey, Buckner, & Timpano, 2009) delivered a highly similar training program, a dot probe task with emotional facial expressions, to individuals with generalized social anxiety disorder (SAD). After the training, 72% of participants in the training condition no longer met diagnostic criteria for SAD, compared to only 11% in the control condition. Furthermore, this effect sustained at follow-up four months later, with 64% of participants who received the training still being in remission, compared to 25% of participants in the control condition.

Given the problem of relapse and the promising findings of attentional retraining in anxiety, it would be interesting to explore the effect of attentional training interventions in depression. To date, only one study has been published investigating attentional retraining in dysphoria (Wells & Beevers, in press). During a period of 2 weeks 16 dysphoric students with mild to moderate symptom severity completed 4 sessions of attentional training. The training program as introduced by MacLeod et al. (2002) was applied and adapted for depression. In their version training materials were dysphoric scenes as well as sad faces presented for a long duration (faces: 3 s; scenes: 4.5 s). Participants completed 4 training sessions (of 196 trials each) in two weeks. Compared to a no-training condition, the participants in the training condition reported significantly less depressive symptoms immediately after training (effect size: d = 0.52) as well as at follow-up 2 weeks later (effect size: d = 1.04). Improvement in the training condition was mediated by change in attentional bias. These results are in line with the assumption of causality ascribed to biased attention in the course of depressive symptoms. These promising findings in a dysphoric student sample justify more extensive research in severely depressed samples.

We wanted to investigate the effects of attentional bias modification in a large sample of dysphoric students (experiment 1) as well as in a sample of depressed in- and outpatients (experiment 2). Provided that there are similarities as well as important differences between subclinical and clinical samples (see Ingram & Siegle, 2009), these studies were run in parallel to allow a broad picture of the effectiveness of attentional retraining in depression. Moreover, young people with depressive symptoms have a high risk for developing clinical depression prospectively (Fergusson, Horwood, Ridder, & Beautrais, 2005). We optimized the training procedure to fit the specific characteristics of attentional bias in depression. Accordingly, we implemented the spatial cueing task to measure and modify maintained attention to negative information. First, all stimuli were self-referent adjectives rated for valence, familiarity and depression specificity. Second, all stimuli were presented during 1500 ms, in order to train the later, elaborative stages of attention. Third, the training aimed to decrease attention for negative information as well as to increase attention towards positive information. We compared a training condition to a no-training control condition in which the same stimuli are presented but without contingency between the emotional stimuli and location of the target.

In sum, the present study tests the hypothesis that attentional bias is causally involved in depression by experimentally manipulating attentional bias and examining the influence on depressive symptoms. If attentional bias indeed is causally related to depression, reducing attentional bias should reduce the presence of depressive symptoms.