Intolerance of Uncertainty as a Cognitive Vulnerability for Obsessive-Compulsive Disorder: A Qualitative Review

Abstract

Intolerance of uncertainty (IU) is the tendency to react negatively to uncertain situations. In this review, we critically evaluate the evidence for IU as a cognitive vulnerability for obsessive-compulsive disorder (OCD), including (a) evidence for a robust association between IU and OCD symptoms, (b) evidence that IU is stable and trait-like, (c) evidence for IU as a causal risk factor that influences the development of OCD directly or indirectly, and (d) evidence that IU is malleable to intervention. The available evidence suggests that IU is likely a candidate cognitive vulnerability factor for OCD, though additional research is needed to determine if increasing tolerance for uncertainty is a mechanism of effective OCD treatment. Future research examining IU across multiple levels of analysis and a more rigorous examination of the nomological network of IU may identify specific pathways contributing to the development and maintenance of OCD.

Obsessive-compulsive disorder (OCD) consists of intrusive, repetitive thoughts, images, or impulses (obsessions) and repetitive behaviors or rituals (compulsions) performed in an effort to relieve obsessional distress. Obsessions and/or compulsions in OCD are distressing and interfere with daily functioning (DSM-5; American Psychiatric Association [APA], 2013). Although the etiology of OCD is complex, a growing body of research suggests that individual differences in intolerance of uncertainty (IU) may confer significant risk (e.g., Gillett et al., 2018; Steketee et al., 1998). IU is defined as “a dispositional characteristic that results from a set of negative beliefs about uncertainty and its implications and involves the tendency to react negatively on an emotional, cognitive, and behavioral level to uncertain situations and events” (Buhr & Dugas, 2009, p. 216). Difficulty in tolerating uncertainty can lead to avoidance of uncertain situations and distress about the potentially negative outcomes of uncertain situations.

Although much remains unknown about the mechanisms that explain how IU contributes to the development of OCD, a theoretical model described in Figure 1 suggests that in uncertain situations, anxious individuals tend to overestimate the likelihood or severity of the potential threat (Einstein, 2014; Morriss et al., 2016; Pepperdine et al., 2018). When personally relevant, perceived threats generate emotional arousal (e.g., Helzer et al., 2009; Riemann & McNally, 1995). If the individual has a high need for predictability, they may experience additional arousal associated with uncertainty (Bar-Anan et al., 2009; Carsten et al., 2022; Einstein, 2014). Individuals with difficulty tolerating this uncertainty arousal may initiate behaviors to reduce the associated arousal and distress, including compulsions or avoidance of the situation and associated thoughts and emotions (Rosen & Knäuper, 2009; Toffolo et al., 2013; Yao et al., 2022). Individuals low in the need for predictability experience less arousal and have the cognitive capacity to recognize that the threat is not yet realized, leading to more adaptive emotional and behavioral responses in service of the individual’s goals (Grupe & Nitschke, 2013; Pushkarskaya et al., 2015; Wild et al., 2014).

Figure 1.

Model of the Role of IU in Psychopathology

Note. Model adapted from “Extension of the transdiagnostic model to focus on intolerance of uncertainty: A review of the literature and implications for treatment,” by D. A. Einstein, 2014, Clinical Psychology: Science and Practice, 21(3), 280–300. https://doi.org/10.1111/cpsp.12077. Copyright 2014 by the American Psychological Association.

The model presented in Figure 1 may be especially useful for conceptualizing the development and maintenance of OCD. Indeed, individuals with OCD frequently overestimate threat, especially when personally relevant (Moritz & Pohl, 2009; Taylor et al., 2010). Desire for predictability also explains unique variance in OCD symptoms (McEvoy & Mahoney, 2012). Compulsions function to reduce uncertainty arousal by increasing the perception of control over future outcomes (i.e., Greco & Roger, 2003). For example, individuals with OCD may engage in excessive checking (“I can’t be sure I locked the door, so I better go back and check”), reassurance-seeking (“Are you sure I didn’t hit someone while I was driving?”), washing and cleaning rituals (“There may be germs on this doorknob, so I need to wipe it down repeatedly throughout the day to prevent anyone from coming into contact with a dangerous disease”), or other mental compulsions (“I’ll repeat the Lord’s Prayer at least one more time through, so I can make sure God heard me and will absolve me of my sin”) in order to reduce the distress associated with uncertainty. Given the proposed role of IU in the development and maintenance of OCD, further investigation into the development of IU, its contributions to OCD, and the ability to modify IU within the treatment of OCD is warranted. As noted by Campbell and Fiske (1959), however, before examining the associations between a specific trait and symptoms of psychopathology, we must first have confidence in the measures of that trait, established through evidence of reliability and convergent and discriminant validity.

Measuring IU

Much of the existing knowledge implicating IU in emotional disorders has been facilitated by the development of the Intolerance of Uncertainty Scale (IUS-27; Freeston et al., 1994). The English version of the IUS-27 (Buhr & Dugas, 2002) demonstrated excellent internal consistency (α = 0.94) and good test-retest reliability (r = 0.74) over five weeks. As evidence of convergent and discriminant validity, the IUS-27 was strongly correlated with measures of worry, anxiety, and depression (rs = 0.55–0.60), with the association between worry and IU remaining significant after controlling for anxiety and depression (partial r = 0.30). An abbreviated version of the IUS-27 has also been developed to facilitate more efficient administration (IUS-12; Carleton et al., 2007). The IUS-12 was highly correlated with the original IUS-27 (r = 0.96) and demonstrated excellent internal consistency (α = 0.91) in an initial investigation of its psychometric properties. The IUS-12 also demonstrated evidence of convergent and discriminant validity, with strong correlations with measures of anxiety, depression, worry, and GAD symptoms (rs = 0.57–0.64) and unique variance in worry and GAD symptoms accounted for by the IUS-12 after controlling for anxiety and depression.

Factor analytic research has proposed that the IUS-12 consists of both a general IU factor and two subfactors. The first, Prospective IU, refers to the tendency toward active information-seeking to reduce uncertainty. The second, Inhibitory IU, refers to avoidance-oriented responses to uncertainty (Birrell et al., 2011). Several studies have examined the factor structure of the IUS-12, with a bifactor model typically providing the best fit (Hale et al., 2016; Shihata et al., 2018). In a bifactor model, items load onto both a general factor (e.g., general IU) and a specific factor (e.g., either Prospective or Inhibitory IU). For example, Hale et al. (2016) found that all 12 items of the IUS-12 loaded more strongly onto a general IU factor than either specific IU factor in a nonclinical sample. Shihata et al. (2018) found that in a bifactor model, 80% of the variance in a nonclinical sample and 86% of the variance in a clinical sample were attributable to a general IU factor. Within the clinical sample, a bifactor model consisting of general IU and only one lower-level factor, Inhibitory IU, best fit the data. Though presenting a total IUS-12 score is sufficient for most studies, there may be additional utility in reporting subscale scores. For example, Prospective and Inhibitory IU are differentially associated with symptoms of emotional disorders (McEvoy & Mahoney, 2011, 2012), with Prospective IU uniquely associated with OCD and GAD and Inhibitory IU uniquely associated with social anxiety disorder, panic disorder, agoraphobia, and depression. Even within OCD, differential associations with IU have been found; in a study of 974 OCD patients entering intensive treatment, a common IU factor was associated with all four OCD symptom dimensions (contamination, responsibility for harm, unacceptable thoughts, and incompleteness), but Inhibitory IU demonstrated unique associations with contamination and unacceptable thoughts (Pinciotti et al., 2021).

Within OCD samples, the Perfectionism/Certainty subscale of the Obsessive Beliefs Questionnaire (OBQ-PC; Obsessive Compulsive Cognitions Working Group, 2001) has been frequently used to assess IU. This subscale also includes items assessing perfectionism and is part of a larger questionnaire that measures beliefs related to OCD symptoms, while the IUS-27 was originally developed to measure IU associated with worry. Gentes and Ruscio (2011) provide an excellent overview of these two measures and their relations with emotional disorders, concluding that correlations between symptoms of psychopathology and IU are largely consistent between the two measures. However, of the 16 items that make up the OBQ-PC subscale, only four were conceptualized as representing IU, while the remainder relate to perfectionism (Myers et al., 2008). While perfectionism may be implicated in some OCD sub-types, nearly all symptoms of OCD are characterized by an inability to tolerate uncertainty: uncertainty regarding dirt or disease transmission in contamination-related OCD, uncertainty about potential negative outcomes in harm-related OCD and OCD focused on symmetry or exactness, uncertainty over one’s own mental state, character, or morality in OCD involving unacceptable or taboo thoughts. Thus, the OBQ-PC subscale, which is heavily weighted towards perfectionism, may underestimate the significance of the link between IU and OCD symptoms. Unlike the OBQ-PC subscale, items on the IUS-12 do not reference specific symptoms of psychopathology, such as a tendency to worry, experience somatic anxiety, or perform rituals in response to feelings of uncertainty, making the IUS more suitable for use across multiple presentations of psychopathology, including OCD.

IU as a Cognitive Vulnerability for OCD

Although there is a significant body of literature suggesting a robust relation between IU and OCD symptoms, such information alone is not sufficient to determine if IU can be considered a cognitive vulnerability for OCD. Some researchers have proposed that IU is at the core of almost all presentations of OCD (e.g., Grayson, 2010). Despite this assertion, limited research is available examining IU as a vulnerability factor for the development and maintenance of OCD or the role of increasing tolerance of uncertainty as a mechanism for effective OCD treatment. Consistent with the recommendations of Koerner and Dugas (2008), we contend that there must be strong evidence across four domains to determine if IU may be conceptualized as a cognitive vulnerability for OCD. This includes the following:

1. Evidence for a robust association between IU and OCD symptoms,

2. Evidence that IU is stable and trait-like,

3. Evidence for IU as a causal risk factor that influences the development of OCD directly or indirectly, and

4. Evidence that IU is malleable to intervention.

Evidence for an Association Between IU and OCD Symptoms

Tolin et al. (2003) proposed that IU may be a central component of the etiology of OCD. Specifically, Tolin and colleagues found that IU was elevated in OCD patients with checking compulsions compared to nonanxious controls (${\eta }_{\text{p}}^2=0.28$) and hypothesized that this relation was due to the role of heightened pathological doubt, an obsessional lack of confidence in one’s memory of performing an action such as locking a door or turning off the stove. Since then, several studies have found evidence of an association between IU and OCD symptoms in both clinical and nonclinical samples. To summarize such findings, two meta-analyses examined the relation between IU and OCD symptoms throughout the empirical literature. Across 33 studies, Gentes and Ruscio (2011) found mean correlations ranging from 0.42 to 0.50 between IU and OCD, depending on the measure used. Similarly, McEvoy et al. (2019) observed moderate associations between IU and OCD (mean r = 0.42) across 69 studies. Together, these two meta-analyses suggest that IU is moderately to strongly associated with OCD symptoms. Importantly, these meta-analyses included only self-report measures of IU and OCD symptoms and include only cross-sectional data. Thus, additional research examining multimodal and prospective relations between IU and OCD symptoms is reviewed below.

In an examination of IU and obsessive-compulsive symptoms using both self-report measures and behavioral tasks, Sarawgi et al. (2013) administered measures of IU and obsessive-compulsive symptoms and tasks related to either ordering and arranging, checking, washing, contamination avoidance, or neutralization to a large sample of undergraduate students. The findings showed that IU was significantly related to each self-report measure of the obsessive-compulsive symptom domains, even after controlling for anxiety and depressive symptoms, rs = 0.42–0.58. Further, IU predicted performance on all in vivo task domains except neutralization/harm after controlling for anxiety and depressive symptoms, including urge to check in a stove checking task (r = 0.22), avoidance (r = 0.22), urge to wash (r = 0.42), and washing duration (r = 0.30) in a contamination task, and urge to arrange (r = 0.30) and arranging duration (r = 0.17) in an ordering and arranging task. In second nonclinical study of checking behavior, IU was significantly associated with urge to check (r = 0.36), task unpleasantness (r = 0.48), overall checking frequency (r = 0.23), and frequency of checking identical stimuli (r = 0.27) during a visual discrimination task; urge to check (r = 0.37) and task unpleasantness (r = 0.50) were also significantly associated with OCD checking symptoms, though the authors did not control for symptoms of anxiety and depression in these analyses, limiting claims of specificity (Wake et al., 2022). Notably, no studies of the association between IU and performance on tasks related to obsessive-compulsive symptoms within clinical samples were identified, suggesting additional opportunities for future research.

Fewer studies have examined the prospective relation between IU and OCD symptoms. In a retrospective study, individuals reported that IU typically preceded the onset of full OCD by approximately 3 years (Coles et al., 2012). Abramowitz et al. (2006) found that multiple domains of obsessive beliefs, including IU, predicted postpartum OCD symptoms in expectant parents, β = 0.31, but did not examine the relative contributions of IU compared to other common obsessive beliefs, such as overestimation of threat and responsibility for harm or the importance and control of intrusive thoughts. In a prospective study, Pozza et al. (2019) found that, after controlling for depressive symptoms, endorsement of perfectionism and high IU beliefs predicted the severity of OCD symptoms 1 year later in both children and adolescents, β = 0.17, accounting for 7% of the variance in OCD symptom severity. Although these studies provide some evidence suggesting a prospective association between IU and OCD symptoms, the existing literature is relatively limited, as none of the existing prospective studies used a robust measure of IU (such as the IUS) to examine changes in IU over time, potentially conflating IU and perfectionism or other obsessive beliefs.

Evidence That IU Is Stable and Trait-Like

Although the extent to which IU is trait-like has not yet been thoroughly investigated, examinations of the test-retest reliability of commonly used measures may provide some insights into its stability. For example, Buhr and Dugas (2002) reported a five-week test-retest reliability coefficient of r = 0.74 for the original IUS-27. Carleton et al. (2014) also found good test-retest reliability over two weeks for the abbreviated IUS-12, r = 0.77–0.83. Though test-retest reliability is an important aspect of measurement stability, both studies examined reliability over just a few weeks. Hong et al. (2017) did find that IU demonstrated moderate stability (stability coefficient = 0.32) from ages 8 to 11, providing some evidence that IU remains a stable trait over a longer time interval through at least early adolescence. However, examination of the stability of measures of IU is an incomplete test of whether the construct is stable and trait-like, given that such studies typically only report correlations on a measure at two timepoints. In addition, test-retest coefficients underestimate the true stability of a given trait as they are attenuated by measurement error (Costa & McCrae, 1988); models that can account for measurement error may provide a more accurate examination of stability (Cole et al., 2005; Kenny & Zautra, 1995). In a recent study, Knowles et al. (2022) examined the stability of IU within a community sample over a 5-month period by partitioning the IUS-12 into time-invariant (trait) and time-varying (state) components, similar to previous studies that have differentiated between trait and state components of anxiety (e.g., Eysenck, 1983). Time-invariant IU accounted for 82% of the variance in total IU, 80% of the total variance in Prospective IU, and 76% of the total variance in Inhibitory IU. In addition, contributing to the conceptualization of IU as a stable, trait-like cognitive vulnerability for OCD, the time-invariant component of total IU, as well as both Prospective and Inhibitory IU, predicted obsessive-compulsive symptoms over time. Furthermore, the time-invariant component of IU was strongly associated with OCD symptoms when controlling for depression. This study provides important new evidence suggesting that it is the stable, time-invariant aspect of IU that has a significant influence on OCD symptoms, rather than the time-varying (state) components of IU (daily fluctuations in one’s ability to tolerate uncertainty) that may shift based on external factors, such as stress. In addition, similar patterns in the relationship between time-invariant IU and OCD symptoms were found regardless of whether total IU or individual subfactors were considered.

Evidence for IU as a Causal Risk Factor in the Etiology of OCD

Kraemer et al. (1997) operationalize a causal risk factor in a manner that is useful for examining the extent to which IU may be implicated in the development of OCD. This operationalization highlights the following criteria for which the available evidence is critically examined: (a) temporal precedence, such that the risk factor (IU) must precede the hypothesized outcome (OCD); (b) the ability to be manipulated; and (c) when manipulated, the risk factor (IU) must change the risk of the outcome (OCD).

Temporal Precedence

Temporal Precedence of IU in the Development of OCD Symptoms.

Evidence that IU temporally precedes OCD symptoms is mixed. Using the IUS-12, Shapiro et al. (2020) did not find evidence of temporal precedence of IU over a 1-year period; baseline IU did not predict the incidence of new internalizing diagnoses or changes in symptoms of emotional disorders, including OCD, among undergraduates with elevated IU. In addition, although changes in IU over 1 year were associated with changes in symptoms of other emotional disorders, changes in IU did not predict OCD symptoms. However, there was a high rate of attrition in this study (from n = 138 at baseline to n = 42 at 1-year follow-up), and the use of a restricted range of IU scores may have weakened the prospective link between IU and OCD symptoms.

Temporal Precedence of IU During OCD Treatment.

In a study of cognitive therapy for OCD, individuals with above-average decreases in IU during treatment had significantly greater decreases in obsessions and compulsions compared to individuals with below-average changes in IU, d = 0.77, with changes in IU preceding changes in OCD symptoms (Wilhelm et al., 2015). In a study of exposure and response prevention for OCD, changes in IU did not precede changes in OCD symptoms (Su et al., 2016). However, both studies used the OBQ-PC subscale and did not isolate changes in IU from changes in perfectionism.

Manipulating Contextual Uncertainty and Beliefs About Uncertainty

Several studies have examined the extent to which both state uncertainty and uncertainty-related beliefs can be manipulated. Below, we review these experimental studies and discuss differences in their design that may impact the interpretation of the results.

Manipulations of Contextual Uncertainty.

In the first study that attempted to modify “intolerance of uncertainty,” Ladouceur, Gosselin, and Dugas (2000) manipulated participants’ state uncertainty during a roulette game by telling them that they had either a worse chance of winning compared to participants in a previous study (higher uncertainty) or a better chance compared to past participants (lower uncertainty). The stakes of the study were raised in that participants were told that $100 would be donated to a fictitious foundation if and only if they drew even or won the game. Participants in the increased uncertainty group reported higher levels of worry about the fictional foundation compared to participants in the decreased uncertainty group, d = 1.70. However, the experimental induction of uncertainty in this study, as well as the perceived consequence, was not personally relevant to participants; uncertainty inductions may be more ecologically valid when participants are personally invested in the outcome (Mosca et al., 2016). In addition, though this study was designed to examine differences in experimentally induced IU, the manipulation did not target enduring beliefs about uncertainty. Follow-up studies have included experimental manipulations designed to change participants’ beliefs about uncertainty (e.g., Grenier & Ladouceur, 2004), though the durability of such changes outside the experimental context is not yet known.

Manipulating Beliefs About Uncertainty.

Grenier and Ladouceur (2004) asked participants to visualize and pretend they had ingested a medication that caused an unpredictable effect, then read a script that either emphasized difficulty tolerating uncertainty (e.g., “It’s frustrating that I don’t know what’s going to happen to me. It’s out of my control”) or acceptance of uncertainty (e.g., “I have to live one day at a time. No one is able to plan and organize everything in advance”). Participants low in state anxiety at baseline reported higher worry after the difficulty tolerating uncertainty manipulation, d = 1.30, and lower worry after the acceptance of uncertainty manipulation, d = −0.84, suggesting that the effects of experimentally manipulated uncertainty beliefs may be limited to individuals in a nonanxious state. While this may suggest a potential ceiling effect of uncertainty manipulation, it is also possible that individuals who are anxious at baseline may already feel somewhat uncertain, with additional uncertainty not contributing to their anxiety. In other words, individuals may have a threshold for the total degree of uncertainty they are willing to tolerate and may not experience a direct dose-response effect of uncertainty on anxiety during uncertainty inductions.

Subsequent research by Mosca et al. (2016) asked participants to select a personally relevant potential negative event before reading a list of uncertainty-related beliefs. Participants in one condition read statements that emphasized negative beliefs about uncertainty, such as “Concerning the negative event it’s difficult to not know what will happen;” these participants reported greater state IU, worry, and negative affect, ${\eta }_{\text{p}}^2=0.09-0.13$, than those in the acceptance of uncertainty condition, who read statements such as “It doesn’t bother me to not know what will happen to me.” Individuals in the acceptance of uncertainty condition did not differ from those in a control condition who did not read statements about uncertainty beliefs, suggesting that it may be easier to experimentally increase state IU than it is to increase acceptance of uncertainty.

In another study that manipulated individual beliefs about uncertainty using a personally relevant event, participants exposed to negative beliefs about uncertainty reported a greater likelihood of feared consequences to personal worries, ${\eta }_{\text{p}}^2=0.08$ (Deschenes et al., 2010), suggesting a potential cognitive bias as a consequence of experimentally modified beliefs about uncertainty. Finally, Rosen and Knäuper (2009) provided individuals with false feedback regarding their level of IU after the administration of a doctored IU questionnaire, which increased or decreased the probability that an item would be endorsed. Individuals who received high IU feedback in a high situational uncertainty condition reported increased worry, d = 1.05, and sought more information about a fictional sexually transmitted infection, d = 0.80, compared to individuals who received low IU feedback in a low situational uncertainty condition, demonstrating a potential behavioral consequence of negative beliefs about uncertainty.

Effects of Experimentally-Modified Beliefs About Uncertainty on OCD Symptoms

Though the existing evidence suggests that beliefs about uncertainty can be modified within an experimental context, only two studies have examined the effects of changes in uncertainty-related beliefs on obsessive-compulsive symptoms. First, a recent study expanded upon Rosen and Knäuper’s experimental manipulation of uncertainty beliefs (Geok et al., 2022). In addition to providing undergraduate participants with false feedback regarding their ability to tolerate uncertainty, participants were asked to keep a daily diary of situations where they tolerated uncertainty well (low IU condition) or poorly (high IU condition) over the course of a week. Geok and colleagues found a significant interaction between IU condition and baseline OCD symptom severity, β = 0.64, such that individuals with high baseline OCD symptoms demonstrated a greater decrease in OCD symptoms after the low IU induction compared to individuals in the high IU induction, suggesting a beneficial effect of tolerating uncertainty on OCD symptoms. Next, in an experimental false-feedback study, undergraduate students who were told they were highly intolerant of uncertainty reported significantly higher threat perceptions about a personally relevant intrusive thought compared to individuals who were told they were more tolerant of uncertainty, d = 0.73, though this manipulation did not affect performance in a checking task (Faleer et al., 2017). Thus, while modifying uncertainty beliefs in an experimental context may impact OCD symptoms, the effects may be limited to self-reported OCD symptoms or specific OCD symptom domains, and additional research is needed to determine if these effects are as replicable as the effects of experimentally induced IU on worry.

Evidence That IU Is Malleable During Treatment

Modifying beliefs about one’s ability to manage uncertainty may contribute to better OCD treatment outcomes (Grayson, 2010; Jacoby & Abramowitz, 2017). Several studies have examined changes in IU within a treatment context. In most studies, IU is examined as one potential treatment outcome, often as a correlate of symptom improvement. In a few studies, the ability to tolerate uncertainty is a specific treatment target.

Effects of OCD Treatment on IU

Several studies have found improvements in IU over the course of treatment for OCD. For example, Belloch et al. (2010) reported that IU decreased significantly after 18 sessions of cognitive therapy for OCD, with IU decreasing more for individuals with autogenous (ego-dystonic; personally unacceptable and unrealistic thoughts) obsessions, d = 2.32, compared to individuals with reactive (realistic, aversive thoughts) obsessions, d = 0.95. In both groups, changes in IU were maintained at 1-year follow-up. Su et al. (2016) reported that perfectionism and IU decreased after 8 weeks of twice-weekly exposure and response prevention for OCD, d = 0.60, and continued to decrease from baseline to 24-week follow-up, d = 0.91. Although most studies of OCD that demonstrate effects on IU focus on cognitive-behavioral treatments, Mathur et al. (2021) found that both perfectionism and IU significantly decreased among OCD patients undergoing mindfulness-based therapy (Cohen’s f² = 0.45).

The evidence demonstrating that IU decreases after OCD treatment provides an initial test of the malleability of IU. However, fewer studies report an association between changes in IU and changes in OCD symptoms during treatment. In the first such study, Overton and Menzies (2005) assessed idiosyncratic cognitive beliefs among individuals with primary checking-type OCD. To measure IU, individuals rated distress about uncertainty related to their most common checking compulsion, such as “How distressing is it for you if you cannot be certain that your front door is locked?” After 12 sessions of exposure and response prevention, individuals demonstrated a large change in IU related to their target compulsion, d = 2.43, and changes in uncertainty beliefs were strongly correlated (r = 0.75) with changes in OCD symptoms. While this study provides compelling evidence that changes in IU and OCD symptoms are strongly linked during treatment for compulsive checking, only a specific form of IU was assessed which limits broader inferences.

Using the OBQ-PC subscale, Wilhelm et al. (2015) found that individuals who reported above-average decreases in perfectionism and IU during cognitive therapy for OCD reported significantly greater decreases in obsessions and compulsions compared to individuals who reported below-average changes in perfectionism and IU during treatment, d = 0.77. Similarly, after 16 sessions of cognitive-behavioral therapy for OCD, Kyrios et al. (2015) observed significant changes in perfectionism and IU (Hedges’s g = 1.10), which were maintained at 6-month follow-up. In addition, greater changes in perfectionism and IU predicted better post-treatment OCD symptoms, β = −0.57. Pinciotti et al. (2020) found that improvements in IU partially explained symptom improvement (β = 0.06) among patients undergoing residential treatment for OCD, suggesting that changes in IU are associated with symptom changes in individuals with the most severe OCD symptoms. Similar associations were found for children and adolescents with OCD and other anxiety-related disorders who attended intensive outpatient treatment, with greater reductions in IU from admission to discharge associated with lower levels of anxiety, β = 0.32, and functional impairment, β = 0.34, at discharge (Sperling, 2023).

Although several studies have identified a link between changes in IU and changes in OCD symptoms during treatment, the directionality of this association is not yet known. In addition, given the lack of experimental control in these treatment studies, it is unclear how changes in IU lead to changes in OCD symptoms or vice versa. A third, unidentified variable, such as changes in distress tolerance or specific behavioral changes, may yet explain this link and will require further exploration.

Direct Interventions Targeting IU

A few studies have attempted to directly reduce IU during treatment. Ladouceur, Dugas, et al. (2000) designed a 16-week therapeutic treatment specifically targeting IU for individuals with GAD. Treatment elements included awareness training, cognitive modification of maladaptive beliefs about worry, imaginal exposure, and problem-orientation training; throughout all treatment components, the ability to tolerate uncertainty was emphasized. Individuals in the treatment condition demonstrated substantial reductions in IU from pre-treatment to post-treatment, d = 1.83, with gains maintained through 12-month follow-up, suggesting substantial malleability in IU as a result of the intervention. In addition, 77% of participants no longer met criteria for GAD at post-treatment. However, the extent to which similar effects may be observed for patients with OCD remains unclear. Whittal and McLean (2002) do describe a group CBT intervention for OCD that includes challenging IU. Specifically, patients are asked to normalize uncertainty by surveying 10 friends or co-workers about whether they remember locking the door the last time they left the house, and how certain they are that the door is locked. However, studies have yet to systematically examine the efficacy of this approach. In another study, a group-based treatment designed to reduce IU (“Making Friends with Uncertainty”; Mofrad et al., 2020) was piloted with 24 individuals with emotional disorders, including OCD, GAD, and depression. The intervention included psychoeducation about uncertainty, encouragement to experiment with uncertainty in low and higher stakes situations, and discussion to help participants apply this learning to threatening scenarios. Overall, 45% of participants showed reliable change in IU, and 80% demonstrated reliable change in anxiety or depression. Although the group included individuals with OCD symptoms, the study did not examine the impact of the intervention on OCD.

Finally, in a single-session intervention, Oglesby et al. (2017) implemented a cognitive bias modification intervention for individuals who demonstrated a high IU interpretation bias, such that they were more likely to endorse a negative interpretation (e.g., “I have a terrible disease”) as related to an ambiguous prime (e.g., “doctor called”) compared to a more neutral interpretation (e.g., “appointment reminder”). During the intervention, individuals were provided with feedback suggesting that neutral interpretations were “correct,” while negative interpretations were “incorrect.” Compared to individuals in a non-IU-relevant control condition, there was a significantly greater decrease in IU among individuals who received the cognitive bias modification intervention, B = 5.10. As evidence of a specific mechanism, there was a significant indirect effect of the IU cognitive bias modification condition on change in IU from baseline to one-month follow-up through change in IU interpretation bias. Thus, even a brief, single-session intervention can reduce a cognitive bias associated with IU, which reduced self-reported IU. However, the effect of this intervention on OCD symptoms is unclear.

Redux: Is IU a Cognitive Vulnerability for OCD?

The existing evidence suggests that IU may function as a cognitive vulnerability for OCD. First, there is evidence of a strong association between IU and OCD symptoms in both cross-sectional and prospective studies. Second, there is evidence that IU is stable and trait-like, with existing studies examining stability over periods of weeks or months. Most of the variance in IU appears to be attributable to a stable, time-invariant (“trait”) component, which consistently predicted OCD symptoms over five months in a community sample (Knowles et al., 2022). Third, the evidence suggests that IU may be a causal risk factor for OCD, though the evidence is mixed. Although one study suggests that changes in IU temporally precede changes in OCD symptoms during treatment (Wilhelm et al., 2015), two others did not find evidence of temporal precedence (Shapiro et al., 2020; Su et al., 2016). However, the question of temporal precedence does require some consideration of when IU would be predicted to have a causal effect, and most studies are not designed for this kind of longitudinal examination. Across experimental studies, there was evidence suggesting that IU can be manipulated via induction, though the duration of such changes in IU is unclear. Two experimental studies find evidence for an effect of experimentally inducing IU on OCD symptoms (Faleer et al., 2017; Geok et al., 2022), though this effect may only be evident in self-reported changes in OCD symptoms and may be specific to some domains of OCD symptoms and not others. Fourth, there is significant evidence that IU improves during treatment for OCD and is associated with change in OCD symptoms during treatment. The magnitude of change in IU during treatment varies, with effect sizes ranging from moderate to vary large depending on the type of intervention and whether the target is general IU or the ability to tolerate idiosyncratic uncertainty. Finally, while a few interventions targeting IU have been tested, none of these studies have specifically examined the association between changes in IU and OCD symptoms. Based on the evidence to date summarized in Table 1, IU is most likely a cognitive vulnerability for OCD. However, it remains unclear if improving tolerance of uncertainty is a necessary and sufficient mechanism of effective OCD treatment (e.g., Grayson, 2010).

Table 1.

Evaluating the Evidence for IU as a Cognitive Vulnerability for OCD

Criteria	Type of evidence	Example citations	Limitations
Robust association between IU and OCD symptoms	Strong association across multiple studies (meta-analysis) Specific association after accounting for general factors (e.g., neuroticism) Prospective associations Associations across multiple levels of analysis	Gentes and Ruscio (2011); McEvoy et al. (2019) Boelen and Reijntjes (2009) Pozza et al. (2019) Sarawgi et al. (2013)	Majority of studies are cross-sectional Few studies examine associations across multiple levels of analysis
IU as a causal risk factor that influences the development of OCD	Temporal precedence (changes in IU precede changes in OC symptoms within an intervention context) Experimental manipulation of beliefs about uncertainty leads to changes in OC symptoms	Wilhelm et al. (2015) Faleer et al. (2017); Geok et al. (2022)	Mixed evidence of temporal precedence Unclear duration of effects of manipulated uncertainty beliefs Longitudinal studies needed over long durations (e.g., early childhood to first manifestation of symptoms)
IU is stable and trait-like	Longitudinal studies in children/adolescents Longitudinal research partitioning variance into trait and state components	Hawes et al. (2021); Zdebik et al. (2018) Knowles et al. (2022)	Association between IU and OC symptoms examined in a community sample; needs replication in clinical samples Limited to the IUS-12 and IUSC; further studies across multiple levels of analysis are needed
IU is malleable to intervention	Direct interventions modifying beliefs about uncertainty Changes in IU after OCD treatment Changes in IU associated with changes in OC symptoms after treatment	Ladouceur, Dugas et al. (2000); Mofrad et al. (2020); Oglesby et al. (2017) Belloch et al. (2010); Mathur et al. (2021); Su et al. (2016) Pinciotti et al. (2020); Wilhelm et al. (2015)	No IU-specific interventions with measured effects on OC symptoms; changes in IU and OC symptoms may be attributed to an unspecified third variable (e.g., distress tolerance)

Note. IU = Intolerance of Uncertainty; OCD = Obsessive-Compulsive Disorder; OC = Obsessive-Compulsive.

Future Directions in the Study of IU and OCD

The Developmental Origins of IU

An important aim in better understanding the role of IU in the etiology of OCD is a better understanding of the factors that contribute to the development of IU. Clarifying the processes that contribute to the development of IU may have direct implications for prevention efforts. One longitudinal study found that mothers’ ratings of children’s shyness (B = 0.84) and observed dysphoria during a laboratory behavioral task (B = 0.87) at age 3 predicted higher self-reported IU in early to mid-adolescence, suggesting a potential association between observed negative emotionality in young children and later IU (Hawes et al., 2021). Similarly, behavioral inhibition (β = 0.24) and insecure attachment (β = 0.32) at age 6 predicted high IU above and beyond neuroticism in young adults after 15 years (Zdebik et al., 2018). Behavioral inhibition and IU are highly related concepts, and behavioral inhibition is a stable characteristic of childhood temperament with strong predictive value for the development of anxiety-related disorders in childhood, adolescence, and adulthood (Fox et al., 2005). Given that IU likely requires an understanding of the unknown and future potential threats, it may be the product of behavioral inhibition, which describes behavior around novel or unfamiliar situations or people. Measures of IU have been developed for use in youth ages 7–17 (Comer et al., 2009), and a parental report adaptation has been used to assess IU in children ages 3–10 (Sanchez et al., 2017). These will be important tools in future research examining the extent to which behavioral inhibition and other traits may function as building blocks for IU. Importantly, traits such as behavioral inhibition alone may not predispose a child to high IU or subsequent psychopathology; environmental considerations also play a role. For example, one study demonstrated that IU partially mediated the association between early adversity and psychopathology (Hayward et al., 2020), though the study was cross-sectional in nature. Future studies should examine the role of adverse childhood experiences in the development of IU and psychopathology, including OCD.

Longitudinal Research on IU and OCD

Though prospective research on IU and OCD is increasing, there are still unanswered questions regarding the causal nature of the association between IU and OCD. In particular, establishing the temporal precedence of IU in predicting OCD symptoms has been difficult, with mixed results in two treatment studies (Su et al., 2016; Wilhelm et al., 2015) and null results in an observational “high-risk” sample (Shapiro et al., 2020). Establishing evidence of temporal precedence within the treatment context is one way to determine if IU is a mechanism for effective treatment for OCD, but more methodologically rigorous long-term studies are needed to examine the developmental trajectories of IU and OCD. Although mean age of onset of OCD occurs in late adolescence (Brakoulias et al., 2017; Ruscio et al., 2010), a distinct early onset phenotype has been reported, demonstrating more severe symptoms and a faster progression from the first appearance of symptoms to full OCD (Anholt et al., 2014; Sobin et al., 2000). Longitudinal studies should ideally measure IU prior to the (early) onset of OCD with extended follow-up across development. This highlights the importance of programmatic measurement research dedicated to the reliable and valid assessment of IU early in development.

Experimental Research on IU and OCD

Research has shown that beliefs about uncertainty can be experimentally modified to influence OCD symptoms (Faleer et al., 2017; Geok et al., 2022), which is an important step in understanding how IU confers risk for OCD. However, it is not yet known how long the effects of changes in uncertainty beliefs last, which has important implications for research on interventions aiming to change IU; we suspect that these experimental inductions affect state (time-varying) IU but may not change trait (time-invariant) IU. Future studies should include follow-up measurement to determine if the effects of changes in uncertainty-related beliefs are enduring, whether in the short-term (days) or long-term (weeks to months). It is unclear if studies demonstrating that IU can be reduced within a brief experimental context can directly inform therapeutic interventions, where lasting changes in trait IU are a specific goal. In addition, only one study of experimentally modified beliefs about uncertainty included a control condition in which beliefs about uncertainty were not targeted (Mosca et al., 2016); this study only found an effect of exposure to negative beliefs about uncertainty, but not acceptance of uncertainty, compared to individuals in the control condition. It is possible that state IU is easier to increase than decrease within a brief experimental context, which may have consequences on IU-focused interventions and suggests that single-session interventions may be insufficient to decrease IU. Though associations between changes in IU and changes in OCD symptoms within treatment studies have been reported (Pinciotti et al., 2020; Su et al., 2016; Wilhelm et al., 2015), the effects of specific interventions targeting IU have not been examined in clinical OCD samples.

Multimodal Measurement of IU

IU is typically measured by self-report. However, self-report measures are often biased toward socially desirable responding and can only capture aspects of behavior that individuals are consciously aware of and choose to report. Thus, researchers have examined correlates of IU using behavioral tasks and psychophysiological measurement to establish a broader nomological network that helps define the construct of IU. According to Cronbach and Meehl (1955), “A necessary condition for a construct to be scientifically admissible is that it occur in a nomological net, at least some of whose laws involve observables.” Given that IU, like many psychological traits, is not directly observable, researchers must establish reliable correlates that can be directly observed. Examining IU across multiple levels of analysis also allows for better understanding of underlying mechanisms (Bilder et al., 2013; Cicchetti & Dawson, 2002).

One example of a framework for researching psychological constructs across multiple levels of analysis is the National Institute of Mental Health’s Research Domain Criteria Initiative (RDoC; Insel et al., 2010; Sanislow et al., 2010). Within this framework, the Intolerance of Uncertainty Scale has been identified as a self-report measure of Potential Threat, defined by NIMH as “activation of a brain system in which harm may potentially occur but is distant, ambiguous, or low/uncertain in probability, characterized by a pattern of responses such as enhanced risk assessment (vigilance).” The current RDoC matrix (https://www.nimh.nih.gov/research/research-funded-by-nimh/rdoc/constructs/rdoc-matrix) suggests a number of other tasks, measures, and circuits implicated in Potential Threat, including physiological measures such as potentiated startle and measures of adrenocorticotropic hormone, cortisol, and corticotropin-releasing factor; neural circuits including the bed nucleus of the stria terminalis; and paradigms such as the no, predictable, and unpredictable threat task (NPU-threat task; Schmitz & Grillon, 2012), in which participants’ responses to predictable and unpredictable threat can be directly compared. The RDoC approach provides a framework by which IU can be modeled across multiple levels of analysis in future research efforts aimed at identifying etiological mechanisms that can be translated into clinical interventions for OCD (e.g., Fineberg et al., 2011). Although measures that capture IU at the self-report level of analysis have been established, much less is known about the utility of measures of IU that correspond to the behavioral and physiological levels of analysis.

Behavioral Measurement of IU

Behavioral measures of IU have been developed to complement self-report measures. Ladouceur, Talbot, and Dugas (1997) proposed a probabilistic inference task in which participants were asked to draw as many marbles as they choose from a bag until they wanted to guess whether the bag contained mostly black or mostly white marbles. Participants also rated their level of certainty associated with their response. Within a moderately ambiguous condition (with a ratio of 85:15 white to black marbles or vice versa), the number of marbles drawn by a participant before they reached a decision was moderately correlated with self-reported IU (Spearman’s ρ = 0.43). Jacoby et al. (2014) referred to this task as the Beads Task and found that, among individuals with anxiety disorders, self-reported IU was associated with the number of draws to reach a decision as well as distress experienced during the task, but only as measured by the OBQ-PC (rs = 0.34–0.40); correlations with the IUS-12 were not significant. Given that individuals with OCD often repeat compulsions to achieve feelings of certainty (Rachman, 2002; Wahl et al., 2008), number of draws to a decision in this task could serve as a useful analog to compulsive behavior. Importantly, however, the association between number of draws to a decision on the Beads Task, OCD symptoms, and IU has not been consistent across studies (Morein-Zamir et al., 2020).

Further attempts to increase the ecological validity of the Beads Task have been conducted by incorporating a cold-pressor task as a threat for incorrect responses (Jacoby et al., 2016) and recruiting a confederate who would be “punished” if the participant made an incorrect response (Jacoby et al., 2019). Results demonstrated a significant association (r = 0.48) between Prospective IU and task-related distress in the solo version of the task, and a significant association (r = 0.31) between Inhibitory IU and task-related distress when using a confederate, suggesting different potential processes by which IU impacts distress related to responsibility for harm to oneself or others. Responsibility for harm is a key belief underlying some presentations of OCD (e.g., Salkovskis, 1999; Wheaton et al., 2010); indeed, Jacoby and colleagues found that the number of draws to a decision in the partnered version of the Beads Task was moderately associated (r = 0.38) with the unacceptable thoughts dimension of OCD symptoms. The Beads Task has also been validated for use with children, without a threatened punishment for incorrect responses (Osmanağaoğlu et al., 2021). However, even with task modifications that increase the threat associated with incorrect responses, the Beads Task is not personally relevant to many anxious individuals, especially those without harm-related concerns. Further development of behavioral IU tasks that include personally relevant stimuli may increase their utility.

A related behavioral IU task is the PACT Anagram Task (Beadel et al., 2014; O’Bryan et al., 2021). Participants are briefly presented with anagrams and told that their ability to solve them is a measure of verbal intelligence. Participants select one of five potential answers, then rate their confidence in their response as well as their level of distress during the task. Participant distress was moderately associated with self-reported IU, r = 0.36, and obsessive-compulsive symptoms, r = 0.30 (O’Bryan et al., 2021). Though both the Beads Task and PACT Anagram Task demonstrate significant correlations with self-reported IU, the associations were moderate in strength and were based on participant-reported distress, rather than specific behavioral indicators. A similar task was used by Bartoszek et al. (2022), in which individuals were given the opportunity to ask questions about a purported aptitude test. Individuals with high IU were more likely to seek additional information even when they were told that the test had a poor ability to predict future outcomes (individuals with high and low IU did not differ in information seeking when they were told that test had excellent predictive power). As noted by O’Bryan et al. (2021), finding behavioral indicators of IU has been challenging, especially ones that can be standardized and performed in a laboratory context.

Physiological Correlates of IU

The NPU-threat task (Schmitz & Grillon, 2012) can be used to assess physiological responses to uncertainty. During this task, participants see various cues that indicate no threat (no shock will be given; “safety”), predictable threat (shock will be given only when a cue appears), or unpredictable threat (shock could occur at any time during this block of the trial). Variations of the task use a startle probe such as a loud noise or aversive scream instead of a physical shock. Nelson and Shankman (2011) reported that individuals high in IU had an attenuated startle response during the unpredictable threat condition, but not during the predictable threat condition, with the results largely driven by Inhibitory IU. In a study using facial electromyography, Morriss et al. (2022) found that higher IU was associated with greater corrugator supercilii activity to both unpredictable and predictable threat of shock compared to safety, as well as poorer discriminatory skin conductance response between unpredictable threat of shock and safety.

Other variations of the NPU-threat task have manipulated the temporal predictability of the threat probe in addition to whether the threat is predicted by a cue or not. For example, Carsten et al. (2022) found that IU predicted increased startle to lower shock probability (33% vs. 100%), regardless of whether the shock occurred in the unpredictable or predictable threat condition; overall, IU predicted higher anxiety ratings in both the unpredictable and predictable threat conditions. Thus, although IU appears to predict threat responses, some studies have found that high IU is correlated with increased startle in unpredictable compared to predictable threat, whereas others find that IU is predictive of general increased reactivity to threat regardless of whether the threat is cued or uncued. Importantly, no version of the NPU-threat task has been used in an analog or clinical OCD sample, an important direction for future research.

Other physiological correlates of IU during unpredictable threat contexts include event-related potentials, including an enhanced tactile P300 response (Ferry & Nelson, 2021) and error-related negativity (ERN; Jackson et al., 2016). Interestingly, Jackson and colleagues identified different electrophysiological responses associated with the Inhibitory and Prospective components of IU, such that Prospective IU was associated with a larger ERN response and Inhibitory IU was associated with smaller ERN response. This finding suggests additional utility in exploring differences among approach and avoidance strategies in response to uncertainty. Although Jackson and colleagues note that an enhanced ERN has been found in OCD, no studies to date examine the associations between ERN, IU, and OCD symptoms, or the association between ERN and IU in an OCD sample.

Other studies have examined the relation between IU and psychophysiological responding within the context of threat conditioning and extinction paradigms. In a recent meta-analysis, Morriss et al. (2021) found a consistent small to medium association (Hedges’ g = 0.28–0.29) between IU and delayed extinction to threat cues as indexed by skin conductance responses, suggesting that individuals high in IU had difficulty updating learned threat associations (“This cue no longer means there will be a shock, but I can’t be too careful”). In addition, this association was robust after controlling for trait anxiety, suggesting evidence of a specific association between delayed threat extinction and IU. Thus, within multiple threat paradigms, there is evidence for physiological response patterns associated with IU. Similar studies have found evidence of impaired updating of learned threat associations, as measured by skin conductance, between individuals with OCD and healthy individuals (Apergis-Schoute et al., 2017); again, however, studies examining associations between IU and physiological responses to threat cues have not been conducted within OCD samples.

Expanding the Nomological Network of IU and OCD Symptoms

A better understanding of the mechanism(s) by which IU has its effects on OCD may be informed by a more rigorous examination of the nomological network of IU (Cronbach & Meehl, 1955). In other words, understanding IU as a cognitive vulnerability for OCD requires situating IU along with other relevant constructs to clarify points of convergence and divergence. Several conceptual approaches have described IU as a specific component of a broader personality trait. For example, IU in adulthood may originate from a behaviorally inhibited temperament in childhood. IU may also be a specific component of neuroticism, the tendency to experience negative emotions, especially in response to stressors (Barlow et al., 2014). This view is supported by research that finds a direct effect of neuroticism on IU as well as specific contributions of IU on anxiety symptoms in both clinical and nonclinical samples (McEvoy & Mahoney, 2012; Sexton et al., 2003), and a meta-analytic study that found that IU, among other cognitive vulnerabilities for emotional disorders, had the strongest factor loading onto a core neuroticism factor (Hong & Cheung, 2015).

The view that IU is highly related to neuroticism also converges with the view of IU as a transdiagnostic factor for emotional disorders more broadly (McEvoy et al., 2019; Rosser, 2019). Indeed, a recent meta-analysis conducted by McEvoy et al. (2019) demonstrated effect sizes ranging from r = 0.40 to 0.57 between IU and symptoms of eating disorders, depression, and anxiety-related disorders, including generalized anxiety disorder (GAD), social anxiety disorder, panic disorder, agoraphobia, and OCD. Consistent with previous studies (Gentes & Ruscio, 2011; McEvoy et al., 2019) and conceptualizations of IU as a transdiagnostic cognitive vulnerability, a recent study also found that the time-invariant component of IU was strongly associated with depressive symptoms even when controlling for symptoms of OCD (Knowles et al., 2022). One interpretation is that the link between IU and various emotional disorders may be an artifact of heightened associations with neuroticism. However, McEvoy and Mahoney (2012) found that IU was significantly associated with symptoms of social phobia, panic disorder and agoraphobia, GAD, depression, and OCD when controlling for neuroticism.

Consideration of the domains of IU may also provide further insights into the nomological network of IU beyond neuroticism. For example, Hong and Lee (2015) found that inhibitory IU demonstrated stronger associations with fear of negative evaluation, anxiety sensitivity, looming cognitive style, and rumination compared to prospective IU. From a nomological network perspective, this distinction between inhibitory and prospective IU suggests potentially distinct etiological pathways that may confer risk for different disorders. Although Hong and Lee (2015) did not include a measure of OCD symptoms, they speculate that, given differential relations between prospective and inhibitory IU and various psychological disorders, individuals with OCD may benefit most from interventions that target prospective IU, such as cognitive restructuring to change threat perceptions, while individuals with social anxiety, panic disorder, and depression with greater inhibitory IU may benefit more from active engagement with uncertainty, perhaps through exposure-based interventions that encourage active approach. Such recommendations will need to be tested empirically. Though cognitive approaches to reducing IU may be useful, behavioral changes may be equally effective; belief change is a mechanism of effective exposure interventions, which target behavioral change (Knowles & Tolin, 2022). For instance, distress-related expectancy violation during exposure predicted OCD remission in a recent study (Elsner et al., 2022); similar violation in expectancies in one’s ability to tolerate uncertainty may also mediate OCD symptom reduction during exposure. Understanding how established OCD risk factors (i.e., obsessive beliefs) fit into the nomological network of IU will be an important next step in understanding the broader construct of IU and the extent to which its association with OCD symptoms is distinct or a function of other related constructs, as well as inform OCD treatment.

Conclusions

There is consistent evidence of a strong association between changes in IU and changes in OCD symptoms. However, the precise causal nature of IU in OCD remains unclear. In fact, recent prospective evidence from a veteran sample suggests that IU may be a significant factor in the maintenance, but not development, of anxiety-related pathology, including OCD (Hunt et al., 2022). Further research is clearly needed to disentangle the role of IU in both the development and maintenance of OCD. The available literature also suggests associations between cognitive-behavioral treatments for OCD and changes in IU. However, evidence for a specific association (e.g., other proposed mechanisms do not account for therapeutic change after controlling for changes in IU), has not yet been established. Furthermore, to establish that increased tolerance of uncertainty is a mechanism within effective treatment for OCD, additional evidence is needed. For example, Kazdin (2007) outlines the following criteria to identify a purported mechanism of a particular intervention: a strong, specific association; consistency across multiple studies and samples; direct manipulation of the mechanism; evidence of temporal precedence within the intervention context; evidence of a gradient or dose-response effect; and plausibility or coherence with the broader scientific literature. Given the evidence for IU as a cognitive vulnerability for OCD, as well as clinical observations of the importance of learning to tolerate uncertainty for individuals with OCD, increased tolerance of uncertainty may be an important mechanism of effective OCD treatment. Future research that uses longitudinal and multimodal measurement and that directly manipulates IU will be well-positioned to advance current knowledge on IU and will inform the development of evidence-based treatment for OCD.

Public Health Significance Statement.

This review details important evidence suggesting that difficulty tolerating uncertainty may be a risk factor for obsessive-compulsive disorder (OCD). It also highlights areas for future research that could lead to better treatments for OCD.