Emotional Intelligence Partially Mediates the Association between Anxiety Sensitivity and Anxiety Symptoms 1,2

Introduction

Anxiety is a complex phenomenon that generally includes a number of physiological responses and associated sensations (e.g., rapid heartbeat, muscle tension, “fluttering” of the stomach, gastrointestinal activity, skin flushing, dizziness, or racing thoughts) in conjunction with the cognitive perception of impending threat, harm, or danger, either in the present or future. For most people, these sensations serve as a warning signal that prepares the body and mind to respond to potential threat. For a minority of people, however, the physical sensations and cognitive activities generated during an anxious reaction may be so aversive that they are interpreted as potentially harmful or even life threatening, leading to concerted attempts to alleviate or avoid such sensations. This propensity to fear the somatic, mental, and social consequences of anxiety is known as anxiety sensitivity (AS; Reiss, 1991). Over time, heightened AS may serve as a risk factor for developing various forms of psychiatric problems, including panic attacks, anxiety disorders, depression, and somatic illnesses. For instance, in correlational studies, concerns related to AS have been associated with higher symptoms of post-traumatic stress disorder (PTSD; Asmundson & Stapleton, 2008), time spent in panic disorder episodes (Benitez, Shea, Raffa, Rende, Dyck, Ramsawh et al., 2009), increases in panic symptoms after a year follow-up (Grant, Beck, & Davila, 2007), and hypochondriacal concerns in depressed patients (Otto, Demopulos, McLean, Pollack, & Fava, 1998). Although AS is established as a stable trait with a strong genetic basis (Stein, Jang, & Livesley, 1999; Zavos, Gregory, & Eley, 2012), the psychological mechanisms underlying its association with the manifestation of pathological symptoms of anxiety remain poorly understood.

Some evidence suggests that people with high AS may have hyperawareness of interoceptive cues and atypical sensitivity to other somatic sensations (Domschke, Stevens, Pfleiderer, & Gerlach, 2010). Within the brain, the insular cortex plays an important role in processing these interoceptive cues (Craig, 2002; Gray, Harrison, Wiens, & Critchley, 2007) and determining how a stimulus might affect the internal somatic state (Paulus & Stein, 2006). The anterior insula, particularly within the right hemisphere, may be especially important for establishing prediction signals about potential future body states in response to a given stimulus or circumstance (Paulus & Stein, 2006). These expectations regarding the potential effects of a stimulus on the future somatic condition can be critical for many cognitive processes, such as decision-making. Such body state prediction signals can help people avoid potentially unpleasant outcomes while facilitating potentially pleasurable outcomes (Craig, 2009). On the other hand, when the strength of these predictive signals is poorly calibrated or exaggerated, particularly for future unpleasant or painful body states, people may be predisposed to developing an anxiety disorder (Paulus & Stein, 2006). Recent neuroimaging data suggest that higher AS is positively correlated with functional activation within the insular cortex during emotional processing (Stein, Simmons, Feinstein, & Paulus, 2007; Killgore, Britton, Price, Gold, Deckersbach, & Rauch, 2011) and is associated with larger volume of the right anterior insular region among people with simple phobias (Rosso, Makris, Britton, Price, Gold, Zai et al., 2010). People who are more sensitive to the sensations of anxiety through heightened awareness of these internal somatic signals might therefore be more easily conditioned by aversive stimuli, which would serve to predispose those with high AS to the development of anxiety disorders (Clark, 1986; McNally, 1990; Barlow, 2002).

While AS appears to be a risk factor for developing an anxiety disorder, not everyone with high AS will manifest panic attacks or demonstrate symptoms of pathologic anxiety (Deer & Calamari, 1998). In other words, the expression of anxiety cannot be fully explained by AS. Other factors must also contribute to this process. Because AS reflects not only a person's “sensitivity” to physiological sensations (e.g., “I feel my heart racing”), but also the cognitive interpretation of those sensations as they pertain to survival (e.g., “a racing heart means I might be having a heart attack”), it is important to consider the role that emotional understanding (e.g., “my heart is racing because I just got excited about winning a game” vs. “my heart is racing because there is something wrong with me”) and management capacities (i.e., the ability to regulate emotions via cognitive, affective, behavioral, or social strategies) play in the expression of anxious symptoms. In other words, the translation of somatic and cognitive experiences into an anxious reaction may depend on the ability to understand and regulate emotional information. The ability to accurately perceive, understand, and manage emotions in order to engage adaptively with the environment has often been termed Emotional Intelligence (EI; Mayer, Salovey, & Caruso, 2002). Thus, it could be hypothesized that EI plays an important mediating role between the experience of AS and the expression of anxious symptoms.

This hypothesis has some potential basis in neurophysiology. According to one conceptualization, EI relies heavily upon interoceptive signals from the insular cortex (Bar-On, Tranel, Denburg, & Bechara, 2003), which become integrated with other emotional salience signals (e.g., from the amygdala) within higher order brain regions such as the ventromedial prefrontal cortex to help guide judgment, decision-making, and other emotional problem-solving processes. Because EI involves insight into the causal and modulating factors underlying a particular emotional experience, as well as the ability to voluntarily regulate the intensity of emotional experience and expression, EI may be an important mechanism through which AS may ultimately manifest as pathological anxiety. If interoceptive prediction signals are excessively skewed or poorly calibrated, as seen in AS, this would lead to altered attunement of the emotional processing systems. Without an accurately attuned and reliable emotional perception and prediction system, it would be difficult to make sense of emotional experiences and their regulating factors and to use that knowledge to effectively guide decisions and behavior. In other words, AS would be associated with reduced EI.

Low EI would also be expected to contribute to a wide range of symptoms of psychopathology, including anxiety and depression. By definition, low EI should be associated with deficits in the ability to accurately perceive the emotions and intentions of others, poor ability to understand feelings and their true antecedents, and limited ability to modulate mood and emotional reactions through voluntary means (Mayer et al., 2002). In this conceptualization, even cumulative minor daily hassles and stresses could lead to exaggerated reactions among those with lower EI, culminating in anxious or depressive experiences. There is considerable support for this notion, as relatively lower EI is often found in people with anxious conditions such as panic disorder (Perna et al., 2010), social anxiety disorder (Jacobs et al., 2008; Nolidin, Downey, Hansen, Schweitzer, & Stough, 2013), generalized anxiety disorder (Lizeretti & Extremera, 2011), depression (Downey, Johnston, Hansen, Schembri, Stough, Tuckwell et al., 2008), and other internalizing problems (Resurreccion, Salguero, & Ruiz-Aranda, 2014). On the other hand, enhancement of emotional skills is associated with improvement in depression (Jahangard, Haghighi, Bajoghli, Ahmadpanah, Ghaleiha, Zarrabian et al., 2012; Fehlinger, Stumpenhorst, Stenzel, & Rief, 2013), as well as other symptoms such as anxiety and social functioning (Pourmohamadreza-Tajrishi, Ashori, & Jalilabkenar, 2013). These findings suggest that EI may be an important mediating variable between predisposing traits such as AS and the eventual manifestation of anxiety symptoms, but to our knowledge, this role of EI in the development of pathological anxiety has not been explored.

One complicating factor that has hindered the scientific study of EI has been the ongoing debate over the nature and definition of the EI construct (Mayer, Salovey, & Caruso, 2008). While several approaches and theoretical models of EI have emerged over the past two decades (Mayer, Roberts, & Barsade, 2008), two have received the most research attention: (1) the Trait model, which is typically assessed via self-report measures and posits that EI consist of a set of noncognitive personal competencies and traits, similar to personality, which help a person cope and succeed under stress and environmental challenges (Bar-On, 2002) and (2) the Ability model, which assumes that EI comprises a variety of demonstrable skills and abilities related to the capacity to reason about emotions and use emotions to solve problems (Mayer, Salovey, Caruso, & Sitarenios, 2001, 2003; Caruso, Mayer, & Salovey, 2002). Each of these models appears to have merit in certain contexts, although they measure different aspects of emotional functioning (Webb, Schwab, Weber, DelDonno, Kipman, Weiner et al., 2013).

Based on the aforementioned discussion, it was hypothesized that EI would serve as an intervening process through which AS may or may not be expressed in terms of anxious symptoms. Thus, the aim of the present study was to determine whether EI serves as a mediating variable between AS and mild anxiety symptoms in healthy individuals using standard, well-validated measures of each construct. Furthermore, given that there currently exist several different approaches for measuring and conceptualizing EI, the role of both Trait and Ability models in the mediation process was investigated:

Hypothesis. Both Trait and Ability EI would mediate the association between AS and anxiety symptoms in a sample of healthy individuals.

Method

Results

Initial evaluation of the data distributions revealed no extreme outliers on any of the variables. Most variables were normally distributed with no evidence of excess skewness or kurtosis. The only exception was the ASI variable, which showed a slight positive skew (skew = 1.24), due to three participants having slightly elevated ASI scores. No gender differences were apparent for any of the variables. Table 1 presents the zero-order correlations among all of the variables.

OPEN IN VIEWER

Table 1.

Zero-order Correlations between Anxiety Sensitivity, Emotional Intelligence, and Anxiety variables.

	2	3	4	5
1. ASI	−.23	−.47†	−.08	.64†
2. MSCEIT	.11	.33*	−.22
3. EQ–i	.47†	−.64†
4. SREIS	−.30*
5. PAI ANX

The results of the mediation analysis are summarized in Figure 1. In all models, the ASI had a significant total effect on ANX (β = 0.57, SE = 0.09, t = 6.39, p < .0001). This is consistent with the hypothesized model, which suggests that people higher on AS are prone to the expression of generalized symptoms of anxiety. OPEN IN VIEWER

Discussion

It was hypothesized that EI would mediate the association between AS and the expression of anxious symptoms as measured by the ANX subscale of the PAI. This hypothesis was supported, although in a limited manner and only for one of the three theoretical perspectives of EI measurement assessed in the present study. Specifically, significant partial mediation between AS and ANX was observed for the Trait measure of EI (i.e., Bar-On EQ–i), but not for the two Ability indices of EI (i.e., MSCEIT or SREIS). Thus, the effect of AS on anxiety symptoms can be partly explained through its effects on the intervening variable of Trait EI.

These findings have important implications for understanding of anxiety disorders and the factors that may contribute to their development, as well as the identification of potential methods of intervention and/or prevention. Whereas prior research suggested that high AS is a risk factor for developing an anxiety disorder simply by virtue of heightened fear of the somatic and interoceptive sensations associated with anxiety (Domschke et al., 2010), the findings build upon this work and suggest some additional complexity to the hypothesized model, by proposing there may be an intervening psychological process. The data are consistent with the hypothesis that, in addition to the direct effects of AS on anxiety, higher AS may also affect symptom expression partly through its effects on Trait EI. As a partial mediating variable, it appears that higher AS might impair some facets of Trait EI, and this lower EI capacity may in turn affect the extent to which symptoms of anxiety are expressed. Of course, the directional nature of this mediation model is speculative because the data were collected at one time point and it is impossible to determine directionality of causation. Future work will need to test this model with a longitudinal design.

Based on the theoretical model proposed by Paulus and Stein (Paulus & Stein, 2006), as well as prior neuroimaging findings (Rosso et al., 2010; Killgore et al., 2011), it can be suggested that heightened AS is associated with hyper-responsiveness of the insula, which then leads to exaggerated sensitivity to interoceptive cues and altered prediction signals regarding future somatic states. Exaggerated interoceptive sensitivity and altered prediction signals would likely make it difficult to differentiate among various emotional experiences (Paulus & Stein, 2006). Such people might find it challenging to understand and make sense of their own emotional responses. According to this model, emotional acuity would suffer due to an insufficient capacity to accurately perceive and gauge the subtleties in the intensity of experienced emotions or the qualitative differences among such experiences. In other words, lower AS would affect the ability to accurately perceive and understand emotional processes—a key component of EI.

In addition to Total EI scores of the Bar-On EQ–i being a partial mediator between AS and anxiety, four of the five subscales (i.e., facets) also contributed individually as significant partial mediators of this relationship. These facets included the IntRApersonal, Stress management, Adaptability, and General Mood subscales, but not the IntERpersonal facet. Building upon the speculation above, it is further proposed that the types of deficits in emotional functioning that could presumably be expected to result from altered insular prediction signals might correspond closely to the domains measured by these factor scales of the Bar-On EQ–i. For example, people with exaggerated insular responses would theoretically be prone to misinterpret their own emotional states and therefore have limited understanding regarding the factors contributing to their emotional reactions, which would correspond to the associations observed between ASI scores and the IntRApersonal facet. Similarly, exaggerated insular responses might also be associated with a lower threshold of tolerance for aversive and stressful activation of these systems, which would be consistent with the associations seen between greater ASI and lower Stress management facet scores. Because altered insular prediction signals provide inaccurate information about the environment, those with exaggerated responses might also have greater difficulty adapting to novel situations, as evidenced by the negative association between ASI and the Adaptability facet. Finally, it is conceivable that these same processes combine to affect mood, potentially contributing to a negative and apprehensive outlook on life, as evident by the negative association between AS and the General Mood facet. The general model proposed here posits that AS leads to lower competencies in the various facets of Trait EI, which in turn increase the likelihood that a person would respond to physiological sensations with increased anxiety. The aforementioned facets showed statistical associations consistent with partial mediation between AS and anxiety symptoms, but it should also be kept in mind that while the mediation model is consistent with our speculative explanations, the causal direction of these associations cannot be verified in the present design. It is quite plausible that the directionality of some or all of the associations could be reversed. This will be a question for further research.

Interestingly, the IntERpersonal facet was not found to be a significant mediator, suggesting that AS was not statistically associated with how an individual perceives and understands the emotions of others, and such skills were unrelated to the expression of pathological anxiety. This is also consistent with the lack of mediation effects found for the indices associated with the Ability model of EI (i.e., MSCEIT and SREIS). Together, these findings suggest that Trait aspects of EI, particularly those involving one's self-attunement to the somatic aspects of emotions and the capacity to maintain a flexible coping style under stress, may partially mediate the extent to which AS will ultimately manifest as symptoms of anxiety.

These findings also point to the potential importance of developing anxiety interventions that target improvement of both AS and EI. Keeping in mind that causal direction cannot be inferred from the current model, reduction in AS might be expected to have benefits in terms of reducing the probability of developing future anxiety disorders and may also be important for enhancing EI skill more broadly. Although very few intervention studies have focused directly on reducing AS, a recent meta analysis suggests that cognitive-behavioral therapy (CBT) approaches appear to be particularly effective at reducing AS (Smits, Berry, Tart, & Powers, 2008). Even brief CBT approaches involving Internet or telephone interventions are effective at reducing AS (Deacon & Abramowitz, 2006; Pier, Austin, Klein, Mitchell, Schattner, Ciechomski et al., 2008; Brenes, Ingram, & Danhauer, 2012; Brenes, Miller, Williamson, McCall, Knudson, & Stanley, 2012; Olthuis, Watt, Mackinnon, & Stewart, 2014). On the other hand, EI training programs are still in their infancy, and there is relatively little rigorous research regarding their effectiveness. While there is much discussion in the literature about the presumed relevance of EI training in adult contexts (Taylor, Farver, & Stoller, 2011), most studies have had limited if any experimental control, focused on broad and loosely defined constructs such as “life skills training” (Lolaty, Ghahari, Tirgari, & Fard, 2012) or have included primary outcomes based on ratings of participant satisfaction with the program or self-reported perceptions of improvement (Codier, Kamikawa, & Kooker, 2011; Flowers, Thomas-Squance, Brainin-Rodriguez, & Yancey, 2014). The limited evidence, however, has been positive and suggests that EI skill enhancement is possible and may play an important role in treatment of psychopathology (Fletcher, Leadbetter, Curran, & O'Sullivan, 2009; Jahangard et al., 2012; Dugan, Weatherly, Girod, Barber, & Tsue, 2014). Moreover, growing evidence suggests that higher EI is associated with better coping and lower anxiety and other psychopathological conditions (Jacobs et al., 2008; Lizeretti & Extremera, 2011; Lizeretti, Extremera, & Rodriguez, 2012; Asberg, 2013; Nolidin et al., 2013; Resurreccion et al., 2014). Thus interventions aimed at improving EI may help develop the skills necessary to remain resilient against stress and maintain greater mental health and well being.