Effects of interpretation bias modification on unregulated and regulated emotional reactivity

Highlights

• Interpretation Bias Modification (IBM) inconsistently affects stress reactivity.

• We assessed effects of IBM on both regulated and unregulated stress reactivity.

• Scenario IBM resulted in training-congruent changes in interpretation bias.

• IBM had no effects on either regulated or unregulated stress reactivity.

• In line with other null-results, we show that IBM does not affect stress reactivity.

Abstract

Background and objectives

Although induced changes in interpretation bias can lead to reduced levels of stress reactivity, results are often inconsistent. One possible cause of the inconsistencies in the effects of interpretation bias modification (IBM) on stress reactivity is the degree to which participants engaged in emotion regulation while being exposed to stressors. In this study, we distinguished between the effects of IBM on natural, unregulated stress reactivity and the effects of IBM on people's ability to up- or downregulate this stress reactivity.

Method

Both in the context of general anxiety (Experiment 1, N = 59) and social anxiety (Experiment 2, N = 54), we trained participants to interpret ambiguous scenarios in either a positive or a negative manner, and we assessed the effects on unregulated and regulated stress reactivity.

Results

Although we found relatively consistent training-congruent changes in interpretation bias in both experiments, these changes had no effect on either unregulated or regulated stress reactivity.

Limitations

In both experiments, we used healthy student samples and relatively mild emotional stressors.

Conclusions

In line with previous research, our findings suggest that the effects of IBM on unregulated stress reactivity may be small and inconsistent. Differences in the extent to which participants engaged in emotion regulation during stressor exposure are unlikely to account for these inconsistencies.

Introduction

According to cognitive theories, biased cognitive processes are at the core of anxiety problems (e.g., Williams, Watts, MacLeod, & Mathews, 1997). Compared to non-anxious individuals, anxious individuals are more prone to interpret emotionally ambiguous situations or stimuli as threatening or negative. This phenomenon is commonly referred to as interpretation bias (IB), and is considered a relatively consistent finding in both generalized and social anxiety disorder (for recent reviews, see Hirsch, Meeten, Krahé, & Reeder, 2016; Stuijfzand, Creswell, Field, Pearcey, & Dodd, 2018). For instance, Hirsch and Mathews (2000) presented emotionally ambiguous sentences to socially anxious participants and non-anxious controls. Measuring lexical decision reaction times (RTs), they found that socially anxious participants were relatively slow to categorize words that resolved the ambiguity in a positive manner (as compared to words that resolved the ambiguity in a negative manner), while non-anxious controls were relatively fast to categorize such words.

Crucially, IB is considered to be causally involved in the maintenance or exacerbation of anxiety and stress reactivity. This causal relation has been addressed in a number of Interpretation Bias Modification (IBM) studies. In IBM studies, participants are typically exposed to IB training tasks designed to encourage either a positive/safe interpretation or a negative/threatening interpretation or placebo training, followed by an anxiety or stress reactivity measurement. Wilson, MacLeod, Mathews, and Rutherford (2006) were among the first to address the effects of IBM on stress reactivity. They trained participants to interpret homographs in either a threatening or safe manner. Following the training, they presented four distressing video clips, and measured participants’ self-reported levels of anxiety and depression before and after this video stressor. In line with the idea that IB is causally related to stress reactivity, they found that participants in the threat training group showed larger increases in anxiety and depression in response to the video stressor compared to the safe training group.

Although there are several studies with similar results (e.g., see Hayes, Hirsch, Krebs, & Mathews, 2010; Lang, Moulds, & Holmes, 2009; Mackintosh, Mathews, Yiend, Ridgeway, & Cook, 2006; Tran, Siemer, & Joormann, 2011), these positive effects have not been replicated consistently. Salemink, van den Hout, and Kindt (2007a) used the scenario paradigm developed by Mathews and Mackintosh (2000) to train participants’ IB. In this paradigm, ambiguous scenarios are presented, with a crucial word in the last sentence of these scenarios consisting only of a few letters. Participants are required to complete these word fragments. In positive interpretation training groups, the correct solutions of the word fragments disambiguate the entire scenario in a positive or safe manner, while in negative interpretation training groups, the correct solution of the word fragment disambiguates the scenario in a negative or threatening manner. Salemink et al. (2007a) found that such training had the intended effect on IB, with participants in the positive training group subsequently more readily making positive interpretations and participants in the negative training group more readily making negative interpretations. However, these effects on IB did not translate to effects in stress reactivity, as there were no group differences in anxiety or depression following a stress induction (e.g., see also Salemink, van den Hout, & Kindt, 2009). More recently, the results of a meta-analysis confirmed that IBM does not consistently affect emotional reactivity in response to stressors, although there was significant heterogeneity between studies (Menne-Lothmann et al., 2014; but see Krebs et al., 2018).

One possible cause of the inconsistencies in the effects of IBM on stress reactivity is the degree to which participants engaged in emotion regulation while being exposed to stressors. Emotion regulation is commonly defined as “the processes by which individuals influence which emotions they have, when they have them, and how they experience and express these emotions” (Gross, 1998, p. 275). A central emotion regulation strategy is reappraisal, which involves changing the interpretation of emotion-evoking stimuli or situations. As such, changing the meaning or interpretation that is assigned to emotionally relevant stimuli or situations is central to both IBM and reappraisal. In the context of depression, Joormann and D'Avanzato (2010) speculated that IB could lead to automatic appraisals of emotion-eliciting situations, thus hampering emotion regulation through reappraisal. Corroborating this idea, Everaert et al. (2017) recently found a negative correlation between negative IB and self-reported use of positive reappraisal. In other words, people with a strong tendency to interpret ambiguity in a negative manner were less likely use positive reappraisal in daily life, and vice versa.

Both the conceptual similarities and the correlation between IB and reappraisal use suggest that there may be common processes involved in both. For instance, both IBM and reappraisal involve the ability to generate outcome exemplars. It is possible that IBM trains people to become better at generating alternative outcomes, corresponding with the valence of the training condition. When confronted with a negative situation after positive IBM, people's increased ability to generate positive outcomes could lead them to reappraise this situation in a more positive way. In contrast, after negative IBM, the increased vulnerability to generate negative outcomes could lead to more persistent negative interpretations of negative situations, thus hampering positive reappraisal. As such, experimentally induced reductions in negative IB could lead to more efficient use of reappraisal as a strategy to downregulate negative emotions, while induced increases in negative IB may lead to less efficient use of reappraisal to downregulate negative affect.

If IBM does indeed increase people's ability to regulate emotions through reappraisal, inconsistencies in the effects of IBM on stress reactivity could be explained by differences in the efficiency of emotion regulation during the stress inducing tasks. IBM studies have focused exclusively on the intensity of stress reactivity as it occurs naturally, as participants are typically not instructed to regulate their emotions and they are only asked to report on the self-observed intensity of negative affect in response to a stressor. However, no studies to date have examined the impact of IBM on emotion when participants attempted to regulate their negative affect. Hence, in past studies, it is possible that inconsistencies in the effects of IBM on stress reactivity were caused by participants' attempts to reduce or perhaps even increase their levels of distress without being explicitly asked to do so.

In our present study, we set out to dissociate the effects of IBM on natural, unregulated stress reactivity and the effects of IBM on people's ability to regulate this stress reactivity. In a first experiment, we trained participants to interpret ambiguous scenarios in either a negative/threatening or a positive/safe manner, and we assessed the effects of this training on self-reported negative emotion intensity while either viewing threatening film clips without emotion regulation instructions versus with instructions to upregulate versus to downregulate negative emotions. In line with Wilson et al. (2006), we expected participants in the positive training group to respond with smaller increases in negative mood than participants in the negative training group when they received no explicit emotion regulation instructions. In addition, we hypothesized that positive IBM would improve the downregulation but hamper the upregulation of negative affect in response to stressors when they were explicitly instructed to do so. Inversely, we hypothesized that negative IBM would hamper the downregulation but improve the upregulation of negative affect.