Table 1
Common definitions of IU and the implied psychological mechanism.
| PSYCHOLOGICAL MECHANISM | EXAMPLE DEFINITION | RELEVANT CITATIONS | |
|---|---|---|---|
| 1 | Negativity overweighting | “[…] tendency of an individual to consider it unacceptable that a negative event may occur, however small the probability of its occurrence” (Dugas, Gosselin & Ladouceur, 2001) | (Arditte Hall & Arditte, 2024; Bredemeier & Berenbaum, 2008; Deschenes et al., 2010; Dugas etal., 2005; Morriss et al., 2022) |
| 2 | Probability distortion | “[…] tendency to overestimate the chance of and be unwilling to accept potential, but unlikely, negative outcomes in uncertain situations […]” (Donthula et al., 2020) | (Deschenes et al., 2010; Dugas, Buhr & Ladouceur, 2004; Gosselin et al., 2008; Luhmann, Ishida & Hajcak, 2011; Milne, Lomax& Freeston, 2019; Reuman et al., 2015) |
| 3 | Information deficit aversion | “[…] individual’s dispositional incapacity to endure the aversive response triggered by the perceived absence of salient, key, or sufficient information, and sustained by the associated perception of uncertainty” (Carleton, 2016b) | (Carleton, 2016a; Freeston & Komes, 2023; Grenier, Barrette & Ladouceur, 2005; Ladouceur, Talbot & Dugas, 1997; Milne, Lomax & Freeston, 2019; Pepperdine, Lomax & Freeston, 2018) |
Figure 1
Cumulative prospect theory and the psychological mechanisms underlying Intolerance of Uncertainty. A) Negativity overweighting can be captured in CPT by loss aversion, measured with the λ parameter, B) probability distortion can be captured in CPT by the shape of the nonlinear probability weighting function, measured with the γ parameter, C) information deficit aversion aligns with the Description–Experience gap (DE gap), which can be captured in CPT by the difference in probability weighting between decisions from description and decisions from experience. Depending on task structure and analysis strategy, probability weighting in decisions from experience can over- or underweight probabilities.
Figure 2
Affective decision task. A) Schematic screen in the description condition. Participants choose medication A or B, based on the provided descriptions of consequences, by clicking the respective button. B) Schematic screen in the experience condition. Participants click on the boxes to “experience” possible the consequences (i.e. sample from) for each medication. In this example, the participant sampled thrice, twice from medication A (with different outcomes), and once from medication B. Participants could sample as much as they wanted before making the decision. Note that no feedback was provided after choices in both conditions; each participant completed both versions of the task. C) After both tasks, participants rated pains and side effects for distress according to a ten-point Likert-scale.
Figure 3
Summary of model fit. A) Posterior predictive checks for the CPT model. We defined several decision strategies participants may have used. For each strategy, we coded whether a participant’s choice aligned with the strategy, excluding trials where no optimal option could be determined. The proportion of observed choices consistent with each strategy is shown in black, and the model-predicted probability of choosing the rule-consistent option is shown in blue (description) or red (experience). Maximize expected value = Defined as deciding for the option with the highest sum of probability-weighted benefit and side effect; Minimize Side Effect = Defined as taking the option with the least intense side effect; Minimize Side Effect Prob. = Defined as taking the option with the least probable side effect; Maximize Treatment = Defined as taking the option with the best treatment probability. B) Model performance, quantified via LOO-based predictive accuracy. Using leave-one-out cross-validation log-likelihoods, we computed predicted choice probabilities for each trial, converted these to binary predictions using a 0.5 threshold, and calculated accuracy as the proportion of correctly predicted choices per participant. Horizontal lines show mean accuracy for conditions. C) Parameter recovery of the model. Correlations between simulated and recovered parameters for description were r = 0.92 and r = 0.98 for λ and γ respectively. For experience, correlations were r = 0.93 and r = 0.93 for λ and γ. For details, refer to section 1.2 in the supplement.
Table 2
Descriptive statistics of the Intolerance of Uncertainty Scale (IUS), its subscales and the Difficulties in Emotion Regulation Scale (DERS).
| M | SD | Min | 25% | 50% MEDIAN | 75% | MAX | |
|---|---|---|---|---|---|---|---|
| IUS Total Score | 50.55 | 13.89 | 22 | 42.75 | 50.00 | 61.25 | 83 |
| IUS Impaired Ability | 14.95 | 5.38 | 6 | 12.00 | 14.00 | 19.00 | 27 |
| IUS Distress under IU | 17.50 | 5.44 | 8 | 13.75 | 17.50 | 21.00 | 30 |
| IUS Vigilance | 18.10 | 5.11 | 6 | 15.00 | 18.00 | 21.00 | 29 |
| DERS Total Score | 93.34 | 23.53 | 42 | 77.75 | 94.00 | 113.00 | 142 |
Figure 4
Overview of modeling parameters. A) Comparison of individual estimates of loss aversion and nonlinear probability weighting (NPW) between the description and experience conditions, and the individual description–experience gap (DE gap). B) Individual estimates of the value function and the probability weighting functions. The black curves represent the function of the median of the posterior estimates of the population-level parameters. C) Population-level probability weighting functions (i.e., medians of μγ,e and μγ,d estimates). D) Population-level posterior distribution of the DE gap. The individual estimates in A and B are the medians of the individual-level posterior distributions; the black lines show medians of corresponding population-level distributions.
Figure 5
Correlation Matrix of CPT Parameters and behavioral variables with the intolerance of uncertainty scale (IUS), its subscales and the DERS. Numbers represent the Spearman correlation, numbers in brackets BF10 for undirected Bayesian tests of these correlations. The λ gap is plotted for completeness. The discrete DE gap uses the discrete operationalization (i.e., proportion of preference reversals) from Wulff, Mergenthaler-Canseco & Hertwig (2018).