Flipbook

1 INTRODUCTION 15 effects of the incorporation or removal of safety behaviors during exposure. Meulders et al. (2016) argued that the inconsistency in empirical findings may be caused by variability in the conceptualization and operationalization of safety behavior. This argument is in line with an earlier proposal of Rachman et al. (2008). They argued that there is no evidence that all types of safety behaviors are detrimental to the beneficial effects of exposure, and called for a reconsideration of the categorical rejection of safety behaviors during treatment. The incorporation of safety behavior in exposure may be beneficial, because it could make the therapy less demanding and less threatening to patients. This may facilitate treatment and may reduce drop- out and refusal. Several studies suggest that adding safety behavior to exposure can indeed enhance treatment acceptability (Levy & Radomsky, 2014; Levy, Senn & Radomsky, 2014; Milosevic & Radomsky, 2013b). However, other studies did not find differences in the acceptability of exposure with and without safety behaviors (see, for example, Deacon et al., 2010; Milosevic & Radomsky, 2013a). The crucial question is under which conditions are safety behaviors harmful or innocuous for the effects of exposure therapy? Answering this question is not only theoretically, but also clinically relevant, because it has consequences for the psychological treatment of pathological anxiety. A hypothesis follows from learning theory: under some conditions, safety behaviors may interfere with inhibitory learning, because they preclude the occurrence of the expected catastrophe. An example is sitting down when one feels dizzy to avoid fainting: this will prevent learning that dizziness does not predict passing out. Under other conditions, however, safety behaviors may not interfere with this learning, because they do not preclude the occurrence of threat. An example would be wearing gloves, goggles, and an apron when approaching a spider (Milosevic & Radomsky, 2013a). These safety behaviors may not interfere with learning that the spider (CS) does not predict the US (e.g., getting attacked). To our knowledge, the hypothesis that the negative