Safety behavior prevents the occurrence of threat, thus it is typically considered adaptive. However, safety behavior in anxiety-related disorders is often costly, and persists even the situation does not entail realistic threat. Individuals can engage in safety behavior to varying extents, however, these behaviors are typically measured dichotomously (i.e., to execute or not). To better understand the nuances of safety behavior, this study developed a dimensional measure of safety behavior that had a negative linear relationship with the admission of an aversive outcome. In two experiments, a Reward group receiving fixed or individually calibrated incentives competing with safety behavior showed reduced safety behavior than a Control group receiving no incentives. This allowed extinction learning to a previously learnt warning signal in the Reward group (i.e., updating the belief that this stimulus no longer signals threat). Despite the Reward group exhibited extinction learning, both groups showed a similar increase in fear to the warning signal once safety behavior was no longer available. This null group difference was due to some participants in the Reward group not incentivized enough to disengage from safety behavior. Dimensional assessment revealed a dissociation between low fear but substantial safety behavior to a safety signal in the Control group. This suggests that low-cost safety behavior does not accurately reflect the fear-driven processes, but also other non-fear-driven processes, such as cost (i.e., engage in safety behavior merely because it bears little to no cost). Pinpointing both processes is important for furthering the understanding of safety behavior.

安全行为可以防止威胁的发生，因此它通常被认为是适应性的。然而，焦虑相关障碍中的安全行为通常代价高昂，并且即使情况并不存在现实威胁，也会持续存在。个人可以在不同程度上参与安全行为，然而，这些行为通常是二元测量的（即，执行与否）。为了更好地理解安全行为的细微差别，本研究开发了一种安全行为的维度测量，该测量与接受厌恶结果呈负线性关系。在两个实验中，接受与安全行为竞争的固定或单独校准激励的奖励组显示出比没有接受激励的对照组减少的安全行为。这允许灭绝学习到奖励组中先前学习的警告信号（即，更新这种刺激不再发出威胁信号的信念）。尽管奖励组表现出灭绝学习，但一旦安全行为不再可用，两组都表现出对警告信号的恐惧增加相似。这种零组差异是由于奖励组中的一些参与者没有足够的激励来脱离安全行为。维度评估揭示了对照组中低恐惧但大量安全行为与安全信号之间的分离。这表明低成本的安全行为并不能准确地反映恐惧驱动的过程，但也反映了其他非恐惧驱动的过程，例如成本（即仅仅因为它几乎不承担成本而从事安全行为）。