Theoretical development of the construct of threat appraisal

Carr (Reference Carr1974) was the first author to describe the components of TA as a process of ‘an individual’s evaluation of a situation in terms of its harmful implications’ (Carr, Reference Carr1974; p. 315). According to this definition, the degree of harm implied by a situation is dependent upon not only an individual’s estimates of the subjective probability and unpleasantness of harm occurring, but also the ‘multiplicative’ interaction of these variables. Carr (Reference Carr1974) was informed in the development of this framework by recent psychophysiological research which suggested that obsessive compulsive disorder (OCD) patients made elevated subjective probability judgements of negative outcomes (Carr, Reference Carr1971), and by the stress appraisal model of Lazarus (Reference Lazarus1966).

Carr (Reference Carr1974) suggested that OCD symptomology is an anxiety management response to TA, and proposed that compulsive symptoms in particular appeared to be a reaction to threat. This added explanatory value to these behaviours which, until that point had been considered a very difficult problem for both clinicians and researchers to understand and treat (see review by Rachman, Reference Rachman2015). The framework of TA proposed that in OCD compulsions develop as way to reduce threat, in particular to lower the probability of awful or catastrophic outcomes. The framework acknowledged the limitations of the conceptualization in terms of its inability to explain why people with compulsions actually did over-estimate the probability of future negative events occurring. Nevertheless, proposing these central features of the TA framework had significant clinical implications in terms of opening new avenues for treatment approaches to address threat misappraisals.

Beck’s contributions (Beck, Reference Beck1970; Beck et al., Reference Beck, Rush, Shaw and Emery1979) through the cognitive approach to the treatment of depression also had a profound impact on the developing understanding of anxiety disorders, and led to researchers and clinicians using cognitive models to understand anxiety. Building on his early observations that depression was associated with loss, and anxiety was associated with threat (Beck, Reference Beck1976), now recognised as the cognitive specificity hypothesis, Beck et al. (Reference Beck, Emery and Greenberg1985) detailed a theoretical account of anxiety. This hypothesis proposed that individual cognitive sets (including previous learning experiences, personal memories, personal interests, and predisposition) are influential in determining the manner in which a person extracts information and forms an initial impression, including TA, about a given situation. Beck et al. (Reference Beck, Emery and Greenberg1985) did not only advocate the process whereby appraisal of threat occurs, but they extended the understanding by introducing the idea of the impact of an individual’s idiosyncratic experiences on this cognitive conceptual process. Similar to Carr, they also suggested that the degree of anxiety an individual experiences is dependent not on one individual component of this process, but the overall estimation reached through a process of initial appraisal of severity and probability of potential danger, alongside judgements about abilities to buffer this danger utilizing coping resources. For Beck et al. (Reference Beck, Emery and Greenberg1985), the relationship between TA components was similar to Carr’s (Reference Carr1974) proposal. Threat appraisal was proposed to be dependent on ‘computating’ (Beck et al., Reference Beck, Emery and Greenberg1985; p. 42) the perceived severity of the danger, the probability of it occurring and the individual’s coping resources, which produced a final analysis about the degree of perceived damage expected. Consequently, responses are then proportional and dependent on this overall evaluation.

The work of Beck et al. was influential in emphasizing that it is the meaning that people attach to events, and the impact of their own previous experiences, which are viewed as instrumental in the interpretation of situations, rather than the situation itself. This was significant to the development of the construct of TA, as it emphasizes the complexity of the variables influencing threat appraisal, mainly through the introduction of the interplay between an individual cognitive set, the situation and threat appraisal components.

The threat appraisal equation

Following the work of Carr (Reference Carr1974) and Beck et al. (Reference Beck, Emery and Greenberg1985), a further description of TA by Salkovskis (Reference Salkovskis and Salkovskis1996) emphasized that the appraisal, or interpretation, of threat is the key to our understanding of anxiety (Salkovskis, Reference Salkovskis and Salkovskis1996). In part a synthesis of Beck’s writing on cognition and anxiety (Beck et al., Reference Beck, Emery and Greenberg1985), Salkovskis (Reference Salkovskis and Salkovskis1996) explicitly proposed an interactivity rather than additivity between perceived probability and perceived cost. Informed by clinical observations that if the interpretation of the perceived awfulness or cost of the appraisal could be directly reduced such as through the use of behavioural experiments, the individual’s probability estimate could be vastly reduced without any direct therapeutic intervention specifically for overestimation of probability (P. Salkovskis, personal communication, 13 January 2016). Salkovskis viewed TA as the subjective appraisal of the probability of danger interacting with the subjective perceived awfulness of the potential danger, mitigated by the perception of coping ability and ability to access external resources if required, termed the threat appraisal equation (Fig. 1). As with Beck et al. (Reference Beck, Emery and Greenberg1985), Salkovskis regarded the appraisal of potential personal resources to cope with threat as part of TA rather than a separate process.

Figure 1. Threat appraisal equation. Adapted from Frontiers of Cognitive Therapy by P. Salkovskis (Reference Salkovskis and Salkovskis1996, p. 53), with permission.

Although the components in Salkovskis’ equation had each been described previously in the literature (e.g. Beck et al., Reference Beck, Emery and Greenberg1985; Carr, Reference Carr1974), this framework demonstrated the importance of the multiplicative relationship between primary appraisal components (likelihood of danger × awfulness), rather than an additive relationship. As such it was an important development in our understanding about the relationship between components of threat. Although aware of Carr’s (Reference Carr1974) paper, which also suggests a ‘multiplicative’ relationship between probability and cost components, this did not directly inform writing of the equation (P. Salkovskis, personal communication, 13 January 2016). The coping component of this framework drew conceptually from the works of Lazarus and Folkman (Reference Lazarus1966), and Bandura (Reference Bandura, Flavell and Ross1981), which emphasized that judgements about coping are based on beliefs about the extent to which the individual feels they can change their environment, and how much this capacity is perceived to be within their control.

Few studies have theoretically tested the entire framework as a ‘multiplicative’ equation although some have investigated the different contributions of the individual components. There is a body of evidence which shows a differential role of the two key components in the primary appraisal of threat, with reported perceived probability and cost levels varying across different anxious presentations (e.g. Poulton and Andrews, Reference Poulton and Andrews1996; Uren et al., Reference Uren, Szabo and Lovibond2004; Smits et al., Reference Smits, Rosenfield, McDonald and Telch2006). A smaller number of studies have investigated the role of both components together of the primary appraisal process, and the relationship between them, using specific questionnaire-based measures developed to operationalize the model. An exploratory study by Butler and Mathews (Reference Butler and Mathews1983) was one of the first studies to experimentally examine the proposition that people with anxiety will make higher threat appraisals. Using Carr’s (Reference Carr1974) framework, they developed a measure of cost and probability estimations, of future-orientated aversive events, and computed the interaction between these variables (cost × probability) to test whether anxious patients made higher appraisals of threat. They asked small groups (n = 12) of individuals diagnosed with generalized anxiety disorder (GAD) and major depressive disorder with a benchmarking control group to rate cost estimates of a series of negative events, and to rate the perceived probability of both negative and positive events, in relation to both themselves and also others. They found significantly increased appraisals of probability and cost for negative threatening events amongst the anxious group and the depressed group (who almost met clinical caseness for anxiety), compared with the control group. They further identified a relationship between anxiety and the elevated estimation of personal risk amongst anxious populations only, which differed from the more pervasive appraisals made by the depressed group. For the multiplicative variable of threat appraisal (probability × cost), the overall threat score for the clinical groups was significantly higher than for the control group.

A more direct test of the multiplicative relationship of cost and probability was undertaken by Freeston et al. (Reference Freeston, Gagnon, Ladouceur, Thibodeau, Letarte and Rheaume1994) who recruited an analogue sample from general hospital waiting rooms, all of whom reported experiencing at least one health-related intrusive thought over the past month. The threat appraisal components, although based on Carr’s (Reference Carr1974) framework, were adapted to the appraisal of health-related intrusive thoughts. This study showed that both the specific probability and cost appraisal ratings accounted significantly for variance in worry, and related to difficulties in removing intrusive worrying thoughts. When the multiplicative components were added later in the hierarchical regression after the singular components of threat appraisal (probability and cost), the multiplicative or interaction threat term still accounted for significant additional variance in worry. The findings support the position that the singular TA terms and the interaction between them are closely linked to anxiety, and they suggest that adding the interaction term may improve the strength of the appraisal model. In a further study on worry by Berenbaum et al. (Reference Berenbaum, Thompson and Pomerantz2007), these authors found that the interaction term accounted for 2% above the independent effects of the individual components (cost and probability). Worry was associated with the interaction of appraisal components, and this interaction added a unique contribution beyond the independent components.

As noted previously, a theoretical elaboration by Beck et al. (Reference Beck, Emery and Greenberg1985) and synthesis by Salkovskis (Reference Salkovskis and Salkovskis1996) proposed that TA includes secondary, coping appraisal components. Partial support for the proposal that lower estimates of coping resources impacts upon or mitigates the critical response to the appraisal of likelihood of severity and costly negative outcomes was found in a study by Woods et al. (Reference Woods, Frist and Steketee2002) who attempted to operationalize Salkovskis’ (Reference Salkovskis and Salkovskis1996) equation, measuring and computing all components except the rescue variable. They utilized a semi-idiosyncratic measure to examine TA components in a group of individuals diagnosed with OCD (n = 18) and a non-clinical benchmarking group (n = 72). In the sample of OCD patients, no relationship between threat appraisal components and subsequent OCD symptoms were identified. The only variable that contributed significantly to the prediction of anxiety symptoms in the clinical sample was a low estimation of their ability to cope. In comparison, only the probability estimation of a negative outcome contributed to the prediction of symptoms of anxiety in the analogue sample. Whilst acknowledging the small sample sizes in this study, the results provide some support for the claim that lower perceived ability to cope may impact on the overall anxiety response during the appraisal processes of a future possible threat.

Treatment models incorporating the threat appraisal equation

Cognitive models of anxiety propose that the focus of CBT should be the re-appraisal of threat components in order to allow the re-evaluation of beliefs and hence ultimately reduce anxiety symptoms (e.g. Clark, Reference Clark1986; Salkovskis, Reference Salkovskis and Salkovskis1996). In support of this focus of treatment, there are studies which have found that threat re-appraisal is associated with a reduction in anxiety symptoms after CBT treatment (e.g. McNally and Foa, Reference McNally and Foa1987; Smits et al., Reference Smits, Rosenfield, McDonald and Telch2006). However, the inconsistencies across studies to test necessary criteria for causality have made it difficult to conclude definitively that threat re-appraisal is the main driver or active ingredient of symptom improvements (e.g. see systematic review by Smits et al., Reference Smits, Julian, Rosenfield and Powers2012). Thus although a reduction in the appraisal of probability and cost estimates infers a relationship between threat appraisal and anxiety symptoms, further and more rigorous testing is needed to determine if threat re-appraisal acts as the key change mechanism.

While the role of threat appraisal components (cost and probability) in anxiety have been identified, it is unclear if these variables act as triggers, or whether they act to perpetuate or maintain anxiety alongside other appraisal variables. The evidence to date does not eliminate the idea that these components could act in a causal way to lead to anxiety; however, further stringent tests of this model are needed to provide better justification for a rationale which reaches beyond the identified significant correlational relationship of probability and their interaction, to one of causation. The main findings suggest that the interactivity between probability and cost estimates may add additional predictive value in understanding anxiety rather than simply considering the singular or additive role of these processes. There is some evidence to suggest that coping efficacy (or lack thereof) is related to the level of anxiety; this is an important consideration within the cognitive appraisal model. However, research investigating the role of the coping appraisal process remains limited.

Clinically, one to two key dysfunctional cognitions or beliefs that are believed to be maintaining the unhelpful or incorrect assumptions about the world, the self or others are often targeted in CBT treatment for anxiety. The threat equation detailing subjective probability × subjective cost suggests that the level of harm implied in a situation is a result of the interaction of these threat elements. Carr et al. (Reference Carr1974) outlined how individuals struggling with OCD often make high subjective estimates of the probability of undesired outcomes and therefore utilize compulsive behaviours to reduce the perceived likelihood of the threatening outcomes occurring. However, individuals with OCD do not use compulsive behaviours in relation to every probable negative situation. Rather, it is suggested that where high cost estimates interact with pre-existing high probability estimates that this produces a high level of threat, necessitating threat reducing behaviour, e.g. compulsions (Carr et al., Reference Carr1974). In this framework the compulsive behaviours in OCD are perceived (incorrectly) as the most effective way to reduce the probability of an awful outcome happening. For example, where there is a belief that one is very likely to pass on a terrible infection to a young child through touch (high probability estimate and high evaluation of a negative outcome), extensive hand-washing rituals may be employed (to reduce the probability of the awful event occurring). The chosen strategy lowers the perceived threat level, which lowers the individual’s anxiety and thus reinforces the hand-washing behaviour. In such a case, interventions such as exposure and response prevention may be promoted to enable a client to tolerate reducing rituals in order to demonstrate that the feared event does not happen and/or that the individual can tolerate the discomfort of not responding to the threat through rituals. This model proposes that clinicians treating anxiety should not only be curious about subjective probability estimates, but also ensure specific predictions are elicited about the consequences of a situation, if they are to be successful in targeting the most acute symptoms.

Theoretical development of the construct of intolerance of uncertainty

The construct of IU originated in clinical observations of patients with GAD, as Freeston et al. (Reference Freeston, Gagnon, Ladouceur, Thibodeau, Letarte and Rheaume1994) noted that some individuals appeared to perceive uncertain elements of situations as threatening, and it seemed that it was the ‘not knowing’ with regard to uncertain events which underlined experiences of worry in clients. It was noted that, for some individuals who were highly intolerant of uncertainty, their appraisal or perception of threat was increased even in the absence of objective risk, difficulties or danger. This led to a proposed conceptual model of GAD which placed IU as a core feature underlining and intensifying features of GAD (Dugas et al., Reference Dugas, Gagnon, Ladouceur and Freeston1998).

Although the development of IU was originally associated with GAD (Freeston et al., Reference Freeston, Gagnon, Ladouceur, Thibodeau, Letarte and Rheaume1994), subsequent research has demonstrated that it is an important construct beyond this disorder implicated in many clinical presentations. Studies have found IU to be associated with symptoms of other emotional disorders, including various anxiety disorders such as OCD, social anxiety, health anxiety Boelen and Reijntjes, Reference Boelen and Reijntjes2009; Gentes and Ruscio, Reference Gentes and Ruscio2011; (Steketee et al., Reference Steketee, Frost and Cohen1998), as well as other disorders such as depression (Miranda et al., Reference Miranda, Fontes and Marroquin2008) and eating disorders (Brown et al., Reference Brown, Robinson, Campione, Wuesnch, Hildebrandt and Micali2017). Therefore, IU can be considered to be impacting on a range of transdiagnostic difficulties rather than being a disorder-specific construct.

While earlier research tended to examine levels of IU in clinically defined groups, later emerging studies sought to establish the composition and nature of this construct. Although original definitions of IU had been vague and evolving, the study by Birrell et al. (Reference Birrell, Meares, Wilkinson and Freeston2011) provided evidence that IU is one construct with distinct components. The systematic review established that IU consists of two distinct and stable factors: the desire for predictability, and paralysis of cognition and action in the face of uncertainty (Birrell et al., Reference Birrell, Meares, Wilkinson and Freeston2011; p. 1998). Further to this, a taxometric analysis of IU by Carleton et al. (Reference Carleton, Weeks, Howell, Asmundson, Antony and McCabe2012) concluded that IU acts more like a dimensional than a categorical construct. Various studies support this conclusion. For example, data from undergraduate samples indicate a strong association between IU and social anxiety (Carleton et al., Reference Carleton, Collimore and Asmundson2010). This evidence indicates that IU may be conceptualized as a continuous construct, varying in degrees from non-clinical to clinical populations and therefore is applicable to people in everyday life situations.

A recent literature synthesis by Carleton (Reference Carleton2016) suggested that IU may underpin anxiety, describing the ‘fear of the unknown maybe a, or possibly the, fundamental fear’ (p. 39). Previous studies have shown that IU robustly contributes to anxiety even after controlling for other fundamental fears which are significant predictors of anxiety, for example neuroticism, anxiety sensitivity, and positive and negative affect (Carleton et al., Reference Carleton, Norton and Asmundson2007a, b; Carleton et al., Reference Carleton, Thibodeau, Osborne, Taylor and Asmundson2014; McEvoy and Mahoney, Reference McEvoy and Mahoney2011). This conceptualization suggests that IU may underlie anxiety, and the evidence indicates that IU’s contribution is independent of other core fear-related constructs.

However, currently the understanding of the pathways through which IU both elevates and maintains anxiety is as yet a less well understood area. In contrast to the longstanding and well-established accepted models of threat appraisal in anxiety (e.g. Clark, Reference Clark1999), there are only a small number of papers which have modelled IU as a feature in anxiety through taking a transdiagnostic view of anxiety rather than disorder-specific (Einstein, Reference Einstein2014; Grupe and Nitsche, Reference Grupe and Nitsche2013). Grupe and Nitsche (Reference Grupe and Nitsche2013) proposed an integrated neurobiological and psychological model of anxiety with a focus on anticipatory responding to uncertainty. They identify five processes which under maladaptive control account for anxiety, including inflated estimates of threat cost and probability, increased attention to threat and hyper-vigilance, deficient safety learning, behavioural and cognitive avoidance, and heightened reactivity to threat and uncertainty. The presented evidence suggests that unhelpful or dysfunctional anxiety associated with uncertainty around a potential threat is central to anxiety at both a clinical and sub-clinical level. Within this model the neural basis of heightened reactivity to uncertainty is seen as separate but related to processes to inflated estimates of threat cost and probability. Einstein’s (Reference Einstein2014) transdiagnostic model, in contrast, suggests that IU only contributes to anxiety in the presence of elevated threat appraisal. This model proposes that where individuals are faced with uncertainty they make elevated threat estimations linked with aversive possible outcomes. Within this model, uncertainty in the context of threat estimations is aversive, and not uncertainty in and of itself. This view sits in opposition to the conceptualization of IU as ‘fear of the unknown maybe a, or possibly the, fundamental fear’ (Carleton, Reference Carleton2016) and the evidence which suggests that IU may contribute independently to anxiety. Although there has been increasing interest in the role of IU in the development and maintenance of anxiety symptoms there are therefore discrepant conceptual views around the ability of the concept of IU to account for the onset and maintenance of anxiety independently of threat estimations.

Theoretically, higher levels of dispositional or trait IU are proposed to impact negatively on cognitive processing under conditions of uncertainty (Buhr and Dugas, Reference Buhr and Dugas2002). There is evidence from correlational studies showing that clinical populations and non-clinical populations who score higher on measures of trait IU also score higher on symptom measures of anxiety (e.g. Freeston et al., Reference Freeston, Gagnon, Ladouceur, Thibodeau, Letarte and Rheaume1994; Mahoney & McEvoy, Reference Mahoney and McEvoy2012a; Tolin et al., Reference Tolin, Abramowitz, Brigidi and Foa2003). Given the early stages of the literature in modelling the contribution of IU in anxiety there is a requirement for further efforts to disentangle the relationship between TA and IU in anxiety which would help to clarify IU’s contribution to anxiety independently of TA.

Evidence for the role of intolerance of uncertainty in anxiety

Experimental tasks have investigated the association between an individual’s threshold of ability to tolerate uncertainty and the evaluation of aspects of an uncertain event. Ladouceur et al. (Reference Ladouceur, Gosselin and Dugas2000) developed a gambling task to experimentally manipulate IU in a non-clinical group. The group in the higher IU condition demonstrated higher levels of worry compared with the group in the lower IU condition on this task. This study was the first to demonstrate the negative impact of IU on cognitive processes (increased worry) and provided evidence consistent with the proposed IU model of GAD. A further series of studies has demonstrated that higher levels of uncertainty guides processing and contributes to the manifestation of worry (Koerner and Dugas, Reference Koerner and Dugas2008; Ladouceur et al., Reference Ladouceur, Talbot and Dugas1997). Given the strength of the relationship with worry, studies investigated a possible casual role in social anxiety and health anxiety-provoking tasks, with results demonstrating a similar trend (Carleton et al., Reference Carleton, Collimore and Asmundson2010; Rosen and Knauper, Reference Rosen and Knauper2009). This evidence supported a transdiagnostic view of IU as the studies demonstrate a link between levels of IU, changes in cognitive processes and experiences of anxiety relative to levels of uncertainty.

The described studies demonstrate evidence of an association between the manipulation of IU and changes in anxiety symptoms. In the recent literature, efforts have been made to investigate the mechanisms by which IU may contribute to anxiety. Mahoney and McEvoy (Reference Mahoney and McEvoy2012b) carried out a study to explore the indirect effects of IU across a range of clinical anxiety disorders. IU was found to account for unique variance in all specific symptom measures, and partially mediated the relationship between neuroticism and symptom measures. This evidence supports IU as a common factor across anxiety disorders, and in terms of its positioning in theoretical frameworks, the authors suggest that IU may be a lower-order cognitive factor influencing the relationship between personality variables and symptoms of anxiety. Although other anxiety constructs (e.g. anxiety sensitivity) were not included in the analysis, there is evidence to suggest that IU may contribute to anxiety beyond other core fear constructs (Carleton et al., Reference Carleton, Thibodeau, Osborne, Taylor and Asmundson2014). The authors suggest IU as a construct is perhaps contributing to anxiety symptoms through its relationship with high-order constructs (e.g. neuroticism).

Further enquiry into the mechanisms by which IU may have an impact on anxiety has considered the circumstances likely to activate this construct. Studies show that individuals high in IU may interpret information under conditions of uncertainty and ambiguity in a more threatening manner. In clinical samples, it has been shown that anxious populations rate ambiguous situations (e.g. using vignettes such as ‘you wake in the middle of the night, you think you have heard something but all is quiet’) as highly threatening (Butler and Mathews, Reference Butler and Mathews1983) and more concerning in comparison with non-ambiguous positive and negative situations (Anderson et al., Reference Anderson, Dugas, Koerner, Radomsky, Savard and Turcotte2012). These negative interpretation biases towards ambiguous situations have been shown in samples of individuals higher in trait levels of IU than those comparatively lower in trait IU (Dugas et al., Reference Dugas, Hedayati, Karavidas, Buhr, Francis and Phillips2005). A further study by Koerner and Dugas (Reference Koerner and Dugas2008) found a similar relationship between level of trait IU (high and low groups) and rating on ambiguous situations in an analogue sample. In this study, the high trait IU group rated all situations more negatively; however, between the low and high trait IU groups, there was only a significant group effect for the ambiguous situations. A more recent vignette-based study by Reuman et al. (Reference Reuman, Ryan, Fabricant, Herring and Jonathan2015) found that uncertain and threatening scenarios that had more explicit uncertainty provoked a higher rating of anxiety than situations in which uncertainty was more implicit. These results demonstrate that tolerance levels for uncertainty play an instrumental role in the negative interpretation of ambiguous situations, rather than unambiguous situations. Thus, IU may play a role in elevating threatening interpretations of uncertain situations and increasing negative reactions to uncertain outcomes.

However, a limitation across these studies is the lack of clear differentiation between ambiguity and uncertainty measures. It is proposed that IU is future orientated, whereas intolerance of ambiguity (IA) is concerned with unclear situations interpretable in the ‘here and now’ (Grenier et al., Reference Grenier, Barrette and Ladouceur2005). IU has been shown to continue to predict variance in worry after other constructs, including IA, have been added to regression analysis. It may be concluded from the evidence that individuals higher in dispositional IU are more likely to experience negative anxious affect in response to ambiguity or uncertainty (Koerner and Dugas, Reference Koerner and Dugas2008). These findings suggest that IU leads to a perception of threat, yet whether IU is contributing to anxiety over and above threat appraisals is not known, as this has not yet been empirically tested.

In addition to IU measured at a dispositional level, recent studies have begun to investigate the impact of uncertainty in situations, and findings suggest that intolerance to uncertainty at a situational level accounts for significant variance in anxiety symptom measures. Mahoney and McEvoy (Reference Mahoney and McEvoy2012a) developed and tested a new specific situational measure of IU (IUS-SS) in a sample of individuals diagnosed with clinical levels of anxiety and found that situational uncertainty – as opposed to the dispositional measure of IU – was reported more in the anxiety group than in the benchmarking control group. Over and above the IU measured at a dispositional level, situational uncertainty accounted for variance in anxiety symptom measures. This indicates that difficulty tolerating uncertainty in situations is closely related to negative affect experienced, and this may be more predictive of situation-specific anxiety than IU measured at a dispositional level. This illustrates that situational IU may be contributing significantly to presentations of anxiety. However, as with previous studies, this measure was based on ambiguous rather than uncertain scenarios, and hence the conceptual clarity is unclear.

An unpublished study has resolved some of the conceptual criticism in its exploration of the cognitive, behavioural and emotional appraisals of uncertain situations in a group of non-clinical adolescent participants. In the design stage of this study, M. Ford (unpublished doctoral thesis, Reference Ford2011) distinguished uncertainty and ambiguity in measure development, to ensure only the inclusion of uncertain situations. It was found that participants’ levels of trait IU were significantly impacted on by appraisal of arousal in situational IU. The measure of appraisal of affect and arousal together explained 23.8% variance in IU – over and above the appraisal of the ‘unpleasantness of the situation’, which was based on threat appraisal items (e.g. likelihood of awful outcome, severity of this outcome). This showed that increases in trait IU in uncertain situations appear to be accounted for by the evoked internal arousal which was interpreted negatively; this contribution was over and above the situation appraisal which was inclusive of estimations of threat. IU may therefore be linked with emotional experiences of uncertain stimuli that are aversive, and this may be independent of threat appraisal. This finding adds to the understanding of why uncertainty may be so intolerable through investigating various appraisals of uncertainty; however, the findings are limited in generalizability as an adolescent population only was used.

A more recent study by Pepperdine et al. (Reference Pepperdine, Lomax and Freeston2018) clearly distinguished between IU and intolerance of ambiguity using a vignette-based measure to look at the role of situational IU in predicting dispositional IU. Findings showed situational IU predisposed participants to see more threat in uncertain situations with a potentially negative outcome, a neutral outcome and even, in uncertain situations with a potential positive outcome. Situational IU was found to be bothersome across all situations even when threat appraisal was controlled for, and this intolerance to uncertainty in the situation contributed unique variance to IU at a dispositional level. Pepperdine et al. (Reference Pepperdine, Lomax and Freeston2018) for the first time clearly separated between threat and IU in personally salient but non-threatening situations. This provides evidence of the relative independence of IU and TA, as uncertainty was still bothersome even in positive situations (where there is not direct possible negative outcome). However, it is unclear how in salient anxiety-specific situations whether these two concepts would relate or inter-relate with one another, or whether they would operate as relatively independent constructs in anxiety.

Clinically, these data suggest that the ‘not-knowing’ component of a situation is difficult for people high in IU to tolerate and that it is important to reduce this bothersome and negative experience. Crucially, this can occur independently of threat (i.e. a probable negative outcome). However, we do not know if the bothersomeness of IU in possible negative situations may inflate subjective threat estimates. An example may be seen in social anxiety and interactions with others – such as, if an individual is uncertain how they may look asking a question and they are bothered by this uncertainty, this negative experience of IU could lead them to make a greater over-estimation of threat (e.g. I am not sure how my question will sound to others, I don’t like not knowing, this doesn’t feel ok > it is extremely likely that I am going to look stupid which will be humiliating and shameful). Although there is no evidence at present of this relationship between IU and TA in anxiety, a scenario such as this does seem plausible.

However, IU could be operating separately to threat appraisals, therefore it seems important to consider which construct (IU or TA) leads to more anxiety in specific situations, including whether there are situations where individuals are unable to identify negative or catastrophic outcomes, but still are distressed or impaired by their feelings. This may be evident in OCD where a significant sub-group of clients report seeking a ‘just-right’ feeling and are very distressed without it. Does this group experience high IU but are lower on evaluation of threat appraisal, which would imply an additional component to the TA equation is present? Additionally, in GAD it may be that high IU is present but that again individuals are unable to identify negative outcomes as a result of threat; rather it is the sensation of IU that is intolerable and aversive to them. This is important to ascertain in order to consider where to target interventions in anxiety, as it would suggest that exposure and response prevention (ERP) may not be effective for all individuals, and that a different target around IU might be more helpful.

Treatment studies incorporating intolerance of uncertainty

There are a small number of treatment studies which have looked at IU as a focus of intervention, most of which have concentrated on GAD. A study examining CBT for worry utilizing the IU model of GAD demonstrated changes in IU preceded or co-occurred with changes in worry, which evidences further the central role of IU in anxiety (Dugas and Ladouceur, Reference Dugas and Ladouceur2000). Similar findings have been demonstrated in CBT for illness anxiety that focus on the IU component (Langlois and Ladouceur, Reference Langlois and Ladouceur2004). Furthermore, a small number of CBT outcome studies which have used transdiagnostic treatments of anxiety have found that changes in IU predict outcome in levels of anxiety, although IU was not specifically targeted in these therapy protocols (Boswell et al., Reference Boswell, Thompson-Hollands, Farchione and Barlow2013; McEvoy and Ercer-Hurn, Reference McEvoy and Erceg-Hurn2016). This supports the suggestion that IU is a transdiagnostic cognitive behavioural process, contributing to a range of anxiety treatments. Although the above evidence suggests that IU may be key in the treatment of anxiety there are few treatment studies specifically aimed at treating IU, therefore further evaluation of the treatment of anxiety targeting IU is required.

In summary, there is now considerable research to suggest that IU is an independent construct that contributes to anxiety. Research has evidenced that IU plays a core role in the cognitive model of GAD (Dugas et al., Reference Dugas, Gagnon, Ladouceur and Freeston1998), a proposition that has been supported by research on worry (e.g. Dugas et al., Reference Dugas, Freeston and Ladouceur1997; Ladouceur et al., Reference Ladouceur, Gosselin and Dugas2000). Studies also evidenced that the manifestation of higher levels of uncertainty guides processing and contributes to a wide range of anxiety difficulties (e.g. Koerner and Dugas, Reference Koerner and Dugas2008). Nonetheless, the mechanisms through which IU may have an impact on anxiety symptoms remain unclear, although there is a suggestion that IU underpins concerning appraisals of uncertain situations, which may explain why people high in IU have been shown to react more negatively to uncertain anxiety-provoking situations than those low in IU. The impact of situational uncertainty on anxiety symptoms appears to be superior in explaining levels of anxiety symptoms in comparison with dispositional levels of IU (Mahoney and McEvoy, Reference Mahoney and McEvoy2012a). However, as most studies have not clearly distinguished uncertain situations from ambiguous ones, the extent these findings can be attributed to IU is limited. IU may be considered an important mechanism for intervention in anxiety, and treatment outcome studies reflect this as levels of IU have been shown to correspond to levels of symptom changes. However, in order to demonstrate causality, further experimental studies using a broader range of tasks across relevant anxiety situations are required to establish this link with anxiety beyond correlational evidence and novel experimental tasks.