1 - (IOP 37) Integrating Computational and Clinical Assessments of Intolerance of Uncertainty: A Multimodal Approach

Uncertainty is a natural part of daily life and often causes discomfort, though individuals vary greatly in their tolerance of it. Intolerance of uncertainty (IU)—the tendency to respond negatively to uncertainty across emotional, cognitive, and behavioral domains—is one risk factor that has been implicated in the development of anxiety and related disorders. Critically, previous research on IU has been hindered by inconsistent definitions of IU across disciplines. While clinical studies primarily rely on retrospective self-report measures of IU, the neurocomputational decision-making field has developed more sophisticated behavioral operationalizations of IU that are grounded in models of risk and ambiguity aversion. These two approaches remain largely disconnected, limiting our understanding of IU in the clinical literature, including how it has been quantified and its predictive utility. The current study used a multimodal assessment of IU to explore how self-report and computational measures of IU relate to one another, as well as to examine how these assessments predicted later anxiety symptoms. Participants (N = 236) were adults from the South Florida community (70.8% Female, 65.7% White, 31.4% Hispanic) who completed two study visits approximately one month apart. At Visit 1, participants completed a self-report measure of IU and the Risk and Ambiguity Task, a computerized monetary decision-making task that captures choice behavior for gain and loss decisions under uncertainty. We used computational modeling to estimate behavioral risk and ambiguity aversion for each participant based on their task choices. At Visit 2, participants completed a self-report measure of anxiety symptoms. Greater self-reported IU was significantly associated with greater behavioral risk aversion for both gains (r = -.14, p = .04) and losses (r = .17, p = .02). Surprisingly, greater self-reported IU was associated with lower behavioral ambiguity aversion for gains (r = -.16, p = .02) but was not associated with ambiguity aversion for losses (r = .01, p = .93). Additionally, self-reported IU was the only significant predictor of anxiety symptoms one month later (β = .38, p < .001); none of the task-based estimates significantly predicted anxiety (all p’s > .05). This study was the first to integrate these distinct operationalizations of IU and compare associations between self-report and task-based measures of this construct. Individuals with higher self-reported IU were more likely to make less risky decisions for both gains and losses in a monetary gambling task. Surprisingly, this relationship was reversed for behavioral ambiguity aversion, highlighting important considerations for future research. Continuing to bridge the disconnect between clinical measures and more computational, behavioral approaches is necessary to advance our understanding of IU and related constructs in order to improve treatment and prevention efforts. Our findings also suggest that an individual’s perception of their uncertainty tolerance may be most predictive of their later anxiety symptoms, identifying an important treatment target to consider clinically. Additional clinical implications and future research directions will be discussed.