Attuned to the Flux of Life: Relations Between Ability Emotional Intelligence and Emotional Reactivity

Abstract

The field of ability-related emotional intelligence (ability EI) could benefit from new perspectives concerning dynamic operations. According to a recent perspective, variations in ability EI are likely to be linked to variations in skills related to evaluation. This perspective contends, perhaps counterintuitively, that higher levels of ability EI are likely to be linked to higher levels of emotional reactivity, defined in terms of stronger event-emotion relationships. Two studies (total N = 245) pursue such ideas in the context of multilevel models involving event valence and emotional experience. Variations in ability EI modulated event-emotion relationships in the context of laboratory inductions involving hypothetical events (Study 1), affective images varying in valence (Study 1), and with respect to naturally occurring variations in positive and negative daily events (Study 2), such that higher levels of ability EI were linked to stronger event-emotion relationships, regardless of whether events and emotions were positive or negative in valence. These results provide new evidence for recent theorizing concerning ability EI while speaking to functional versus dysfunctional perspectives on emotional reactivity.

The premise of the ability-related conception of emotional intelligence (ability EI) is a bold one: Individuals are likely to differ in skills related to perceiving, understanding, and managing emotions and these skills are likely to matter with respect to functioning in both personal and professional domains (Mayer & Salovey, 1997). Such thinking gave rise to a number of ability-based EI tests that are reasonably well-established, such as the MSCEIT (Mayer et al., 2003), the STEU and the STEM (MacCann & Roberts, 2008), and the NEAT (Krishnakumar et al., 2016). However, the idea that higher levels of EI ability should be linked to higher levels of well-being, better educational and work performance, and better social relationships (Brackett et al., 2011) has resulted in modest findings, according to several reviews of this literature (e.g., Matthews et al., 2012; Ybarra et al., 2014).

In light of the state of the literature, commentators have advocated new directions in ability EI research, particularly with respect theory development (Fiori & Vesely-Maillefer, 2018; Miners et al., 2018; Ybarra et al., 2014). If we could develop a theory of the sorts of skills that result in high ability EI scores, for example, we could then link this theory to processes and outcomes that make sense from a mechanism-related perspective (Miners et al., 2018). Robinson, Asad, and Irvin (2023) developed such a theory, which introduced an “evaluative expertise” perspective on ability EI. Although evaluation is a central component of connotative (Osgood, 1962) and emotional (Barrett, 2006) meaning, individuals almost certainly differ in the extent to which they can assign evaluations in an accurate manner (Fazio, 2000; Jarvis & Petty, 1996). When we present test-takers with EI test materials, we are, at the very least, assessing their abilities to evaluate both stimuli (e.g., in assigning degrees of goodness or badness to described situations) and response options (e.g., in distinguishing benefit- versus harm-producing ways of responding). That evaluation skills contribute to ability EI scores was shown by Krishnakumar et al. (2016), who found that individuals whose evaluations of words (e.g., passion, illness, gossip) were more accurate obtained higher NEAT (ability EI) scores, whether related to perception (r = .40), understanding (r = .42), or management (r = .45). Note that evaluation skills seem to implicate global EI (what is shared among branches: MacCann et al., 2014) rather than any particular branch and we theorize in such terms.

Robinson and colleagues have amassed considerable additional evidence for the idea that high EI individuals, relative to low EI individuals, appear more capable of assigning evaluations to events and stimuli, though such evidence will only be briefly reviewed here (see Robinson, Asad, & Irvin, 2023). People possessing higher levels of ability EI are less prone to emotional (Robinson et al., 2020) and attitudinal (Robinson, Irvin, & Krishnakumar, 2023) ambivalence, both of which would seem to suggest some degree of evaluative confusion (Berrios et al., 2015; Conner & Armitage, 2008; van Harreveld et al., 2015). People possessing higher levels of ability EI are also more certain concerning the evaluations that they make, which in fact depart from neutrality to a greater extent, and they exhibit greater consistency across their cognitive, affective, and behavioral reactions to objects (Irvin et al., 2024). Links of this type extend to important domains such as job satisfaction in that variations in ability EI are linked to job satisfaction ratings that exhibit greater certainty and consistency (Irvin et al., 2024; Robinson, Irvin, & Krishnakumar, 2023). All of this research, in one way or another, converges to suggest that high EI individuals, relative to low EI individuals, are more capable of assigning evaluative meaning to the events and stimuli that they encounter.

If high EI individuals are more capable of evaluating events and stimuli, they should, according to the present analysis, be prone to higher levels of emotional reactivity. This should be the case because emotional reactions happen when events are appraised – that is, interpreted with respect to implications for personal wellbeing (Kuppens & Tong, 2010; Moors et al., 2013). Evaluative processes are central to this form of meaning analysis in that most appraisal dimensions are essentially forms of evaluation (Shuman et al., 2013). Roseman and Smith (2001) state this point in a particularly straightforward manner: “emotions are elicited by evaluations (appraisals) of events and situations” (p. 3). From this perspective, people (e.g., those high in EI) who are more capable of assigning evaluative meaning to events and stimuli should tend to exhibit higher levels of emotional reactivity to those events and stimuli. By contrast, people (e.g., those low in EI) who are less capable of assigning evaluations to the events and stimuli that they are exposed to should exhibit levels of emotional reactivity that are more muted in nature (Beshai et al., 2018; Klein et al., 2023).

We should further unpack this analysis. There has been considerable interest in the idea that some events or stimuli are evaluatively complicated, giving rise to mixed evaluations and mixed emotions (Larsen & McGraw, 2014). However, such events and stimuli seem to constitute the exception rather than the rule in typical cases of emotion elicitation (Russell, 2017). For example, Ito et al. (1998) asked participants to make separable positive and negative evaluations of images from the International Affective Picture system, which are commonly used in emotion elicitation studies (Lang & Bradley, 2013). The correlation between positive and negative image evaluations was -.78, suggesting that there are preciously few stimuli that can be characterized as ambivalent or mixed. Briesemeister et al. (2012) found a similar correlation of -0.88 between positive and negative evaluations of words and Yik (2007) found a similar -.89 correlation with respect to the emotions elicited by recalled events. These results converge on the idea that the events that tend to elicit emotions are only rarely ambivalent. In other words, pleasant events or stimuli tend not to be unpleasant and unpleasant events or stimuli tend not to be pleasant.

In this context, we have suggested that people higher in EI seem more capable of assigning evaluations to affective events and stimuli. When encountering a positive event (or stimulus), high EI individuals should evaluate it more positively, giving rise to higher levels of positive affect. Conversely, when encountering a negative event, high EI individuals should evaluate it more negatively, giving rise to higher levels of negative affect. Altogether, then, the emotional reactions of high EI individuals, relative to low EI individuals, should more closely track the valence (positivity versus negativity: Barrett, 2006) of the events that they are exposed to. This analysis does not suggest that high EI individuals should be prone to positive or negative emotional experiences under all (e.g., neutral) circumstances. Instead, what the analysis suggests is that event-emotion relationships should tend to be systematically stronger at higher, relative to lower, levels of the ability EI continuum.

The present predictions also follow from the “hypersensitivity hypothesis”, which contends that high EI individuals seem to be more sensitive to emotions or emotional information than low EI individuals are (Fiori & Ortony, 2021; Fiori et al., 2023). Thus far, this hypothesis has primarily been supported in two paradigms. In one, Fiori and Ortony (2021) showed that embedding positive or negative pieces of information in a person perception text influenced social impressions to a greater extent at higher relative to lower ability EI levels. In another paradigm, Nicolet-dit-Félix et al. (2023) showed that only individuals very high in emotional understanding displayed an attentional bias for emotional faces in a dot-probe task. These results involve social impressions (Fiori & Ortony, 2021) or attention-related processes (Nicolet-dit-Félix et al., 2023), but the basic hypothesis – namely, that high EI individuals may be more sensitive to valence-related cues – would reasonably extend to the emotional reactivity processes that are examined in the present studies. Regardless, the present phenomena are novel to the literature.

The focus of the present studies is also an important one. Traditionally, tendencies toward emotional reactivity have been viewed as problematic (Carver & Johnson, 2018). Emerging perspectives and sources of data have countered this idea. Blunted (i.e., lesser or insufficient) reactions to inductions have been observed in a number of clinical conditions as well as among individuals with compromised physical or mental health (Bylsma et al., 2008; Carroll et al., 2017). Some of the relevant mechanisms that have been identified – such as disengagement from the environment (Bylsma et al., 2008) or a lack of emotional flexibility (Coifman & Bonanno, 2009) – would reasonably vary by levels of ability EI (Vanuk et al., 2019). Other sources of data, not involving clinical populations, have suggested that stimulus-appropriate emotional responding may be an indicator of psychological health (Hardy & Segerstrom, 2017) or well-being (Klein et al., 2023), in part because stronger emotional reactions give rise to behaviors that are marked by greater emotional wisdom (Robinson et al., 2021). Findings linking ability EI to emotional reactivity would provide hitherto unavailable support for such “emotional flexibility” (Beshai et al., 2018; Waugh et al., 2011) perspectives.

The present research consists of two studies, one of which examined emotional reactivity tendencies in the laboratory and the other of which examined them within a daily life protocol. In both cases, we used a multilevel modeling approach to data collection as well as its analysis, positing that within-person relationships (or within-person slopes: Nezlek, 2008) linking event valence to emotional reactions would be stronger at higher levels of ability EI. In both studies, too, we contrasted moderating effects pertaining to ability EI with potential moderating effects involving extraversion and neuroticism, which are two temperament-related personality traits that have sometimes been linked to emotional reactivity tendencies in previous studies (e.g., Gross et al., 1998; Zelenski & Larsen, 1999). Moderating effects pertaining to ability EI should be distinct, both because ability EI tends not to correlate with these personality traits to a high degree (Joseph & Newman, 2010) and because ability EI would be expected to moderate event-emotion slopes with respect to events and reactions of both valences. Such predictions contrast with those that that have been made concerning extraversion and neuroticism, which have been hypothesized to be specific to positive events and reactions (extraversion) or negative events and reactions (neuroticism).

Study 1

In Study 1, we designed two emotional reactivity tasks, one of which focused on simulated reactions to realistic daily events (Irvin et al., 2023) and the other of which focused on emotional responses to affective images (Robinson & Clore, 2001). Owing to the range of stimuli used, which varied from strongly negative to strongly positive, event reactivity should be evident across levels of the ability EI continuum. However, we hypothesized that these within-person slopes linking event valence to emotional reactions would be stronger as levels of EI increased, giving rise to systematic cross-level interactions (Nezlek, 2008).

Method

Participants and General Procedures

Key hypotheses were examined using a multilevel modeling (MLM) platform and we followed rules of thumb for this platform in planning sample size. In particular, we sought to exceed 900 rows of data, which, according to simulation data, would provide over .90 power to detect main effects (Scherbaum & Ferreter, 2009) and approximately .80 power to detect cross-level interactions (Mathieu et al., 2012). Participants were students from a Midwestern University in the United States who received course credit for their psychology classes. They signed up for a study concerning attitudes and emotions, provided informed consent, and then accessed a link to a Qualtrics-programmed website. After providing demographic information, participants completed the ability-based emotional intelligence instrument described below, following which they completed the other measures. A total of 184 individuals started the study, but 23 did not finish it and 13 failed attention checks, resulting in a final sample size of 148 (66.89% female; 84.46% White; M age = 19.35). Data, code, and materials for this project are available using the following link: https://osf.io/c6z4v/?view_only=5d47d878950c4d2eac93546723aa9a08 (Authors, 2023).

Ability-Related Variations in Emotional Intelligence

Ability-related variations in emotional intelligence were assessed using the North Dakota Emotional Abilities Test (NEAT: Krishnakumar et al., 2016). In its typical form (Robinson et al., 2019), this situational judgment test (Libbrecht & Lievens, 2012) presents test-takers with 20 situational descriptions (e.g., “There have been widespread layoffs in Margie’s organization recently”). With respect to 10 perception-related scenarios, participants rate the extent (1–5 scale) to which the protagonist would experience each of 4 emotions (in the Margie case, anxiety, anger, fear, confusion). With respect to another 10 management-related scenarios (e.g., “Stephan saw his co-worker struggling to a considerable extent”), test-takers rate the effectiveness (also a 1–5 scale) of 4 potential ways of responding (in the Stephan case, take over the co-worker’s more challenging tasks, hope the co-worker eventually “gets” it, encourage the co-worker to try harder, focus his attention on other things).

All 80 ratings are then rescored using a proportion consensus metric (Mayer et al., 2003), based on a key involving the ratings of 82 organizational experts (e.g., CEOs, presidents of companies) with an average of 18.53 years of work experience and an average of 27.15 subordinates (e.g., if the participant made a 3 rating for a particular scenario/item combination and 32 experts also made a 3 rating, the participant would receive a .3902 score for that response). These expert proportion consensus scores are then averaged per scenario and then across scenarios (in Study 1, M = .3306; SD = .0399; α = .84). For exploratory purposes, we also computed separate scores for the perception (M = .3520; SD = .0519; α = .81) and management (M = .3092; SD = .0452; α = .75) branches of the NEAT, while noting that our focus was on global EI – that is, skills that are shared across branches (MacCann et al., 2014). With respect to global EI, women (M = .3382; SD = .0312) received higher average scores than men (M = .3154; SD = .0504), F(1, 146) = 11.44, p < .001, η_p² = .07, and participant sex will be controlled for in follow-up analyses.

The NEAT has been extensively validated. It correlates with personality traits and cognitive ability in expected manners (e.g., in the form of a .35 correlation with ACT scores). Further, total NEAT scores correlate highly with alternative assessments of ability EI. In particular, Krishnakumar et al. (2016) found that total NEAT scores correlated at .69 with the Situational Test of Emotional Understanding and at .59 with the Situational Test of Emotion Management (MacCann & Roberts, 2008). In organizational settings, higher NEAT scores have been linked to higher levels of job performance (e.g., Krishnakumar, Perera, Hopkins, & Robinson 2019) and organizational citizenship (e.g., Robinson, Irvin, & Krishnakumar, 2023) as well as lower levels of workplace deviance (e.g., Robinson et al., 2019). The NEAT also performs well in non-employee samples, in which it predicts affiliative problem-solving tendencies (Krishnakumar, Perera, Persich, & Robinson, 2019), lesser tendencies toward aggression (Robinson et al., 2013), and a wide variety of findings suggestive of greater affective attunement (for a review, see Robinson, Asad, & Irvin, 2023).

Extraversion and Neuroticism

For the sake of discriminant validity, we assessed the personality traits of extraversion and neuroticism, which are the two personality traits of the Big Five model that have most frequently been linked to emotional reactivity (Bolger & Schilling, 1991; Gross et al., 1998; Zelenski & Larsen, 1999; though see Hisler et al., 2020; Lucas & Baird, 2004). In Study 1, extraversion and neuroticism were assessed using Mini-IPIP scales, which correlate highly with other Big Five measures and exhibit test–retest correlations similar to longer scales (Donnellan et al., 2006). Participants rated their agreement with statements indicative of extraversion (e.g., “I am the life of the party”) and neuroticism (e.g., “I get upset easily”) and scores were computed by averaging across items (extraversion: M = 3.18; SD = 0.95; α = .78; neuroticism: M = 2.93; SD = 0.79; α = .63). In Study 1, ability EI levels correlated with extraversion, r = .20, p = .013, but not neuroticism, r = .02, p = .802.

A Simulation-Based Test of Emotional Reactivity

We sought to assess emotional reactivity to a variety of realistic everyday events and we used a simulation-based method (Irvin et al., 2023; Tangney et al., 1996) in capturing such tendencies, given data indicating that simulated and actual forms of emotional reactivity correlate highly with each other (Robinson & Clore, 2001). Participants were asked to imagine themselves in 10 situations, 5 of which involved pleasant events or conditions (e.g., “You just watched an entertaining movie”) and 5 of which involved unpleasant events (e.g., “You just misplaced your house keys”). In response to each situation or event, participants were asked to rate the extent to which (1 = none; 7 = an extreme amount) they would experience 3 positive emotions (pride, joy, and interest) as well as 3 negative emotions (sadness, fear, and disgust), based on Fredrickson (2013).

To examine event-emotion relationships at the participant level, we needed event valence predictors and outcome variables. A positive event (or positivity) predictor was computed by characterizing each event in terms of the average levels of positive affect (in the sample as a whole) that it gave rise to. A negative event (or negativity) predictor was also computed in a parallel manner. For analysis purposes, both of these continuous event valence predictors were z-scored (Nezlek, 2008). Outcomes for the simulation task consisted of person- and event-specific reports of positive affect (M = 3.77; SD = 0.56; α = .87) and negative affect (M = 2.49; SD = 0.59; α = .77). This multilevel dataset consisted of 1480 rows, with event and emotion values nested within participant (Nezlek, 2008).

Emotional Reactions to Affective Images

To investigate emotional reactivity in a non-simulated manner, we used an image induction task. Participants were presented with 10 images from the International Affective Picture System (IAPS: Lang et al., 2005), given that this induction method has been shown to induce subjective, physiological, and expressive indicators of emotional responding in many previous studies (Lang & Bradley, 2010, 2013). The 10 images, chosen from Robinson and Clore (2001), varied widely in content (e.g., a filthy toilet, a hostile dog, individuals on a roller coaster, a serene desert scene) as well as valence norms and were presented in a randomized order. An event valence predictor was created by z-scoring the image valence norms collected by Lang et al. (2005).

Participants were instructed to view each image, following which they rated how sad, angry, excited, fearful, happy, and affectionate (Robinson & Clore, 2001) they felt in response to it (1 = not at all; 7 = extremely). On the basis of these ratings, we computed person- and image-specific scores for positive (M = 2.68; SD = 0.66; α = .81) and negative (M = 2.98; SD = 0.66; α = .83) emotional experience ratings by averaging across emotion markers of a given valence. The resulting multilevel dataset consisted of 1480 rows (148 participants times 10 images).

Results

Analysis Plan

Analyses focused on the multilevel datasets. Within-subject predictors consisted of z-scored versions of positivity or negativity (simulation task) and/or image valence (image task). Between-subject predictors, which were also z-scored, consisted of ability EI scores, sex (-1 = male; + 1 = female), and/or a personality variable. Outcomes retained their original units in primary analyses, but a second set of analyses employed z-scored outcome variables, resulting in “standardized bs” that can be considered measures of effect size (Lorah, 2018).

A first set of analyses will focus on “level 1” (or normative) relationships between event valence and experiences of positive or negative emotion. A second set of analyses will focus on potential “level 2” main effects for ability EI. The final critical set of analyses will examine whether level 1 associations linking event valence to a given class of emotions vary by ability EI. Such cross-level interactions were hypothesized. Subsequent to critical analyses, we performed other cross-level MLMs to characterize effects for EI branches, sex, and/or personality. All analyses were performed using the PROC MIXED procedure of SAS (Singer, 1998).

Were Normative Reactivity Tendencies Present?

In the simulation task, situation positivity led to reactions that were more positive, b = 2.001 [1.926, 2.076], t = 52.07, p < .001, standardized b = .931, and less negative, b = -1.245 [-1.321, -1.168], t = -31.90, p < .001, standardized b = -.774. Similarly, situation negativity led to reactions that were more negative, b = 1.329 [1.249, 1.409], t = 32.76, p < .001, standardized b = .826, and less positive, b = -1.874 [-1.947, -1.801], t = -50.50, p < .001, standardized b = -.872. In the image viewing task, valence was a robust predictor of both positive, b = 1.220 [1.127, 1.313], t = 25.74, p < .001, standardized b = .679, and negative, b = -1.1636 [-1.731, -1.546], t = -34.68, p < .001, standardized b = -.827, emotional reactions. These results indicate that event-emotion relationships were strong.

Did Ability EI Predict Average Levels of Emotional Experience?

Investigators have been interested in whether ability EI predicts affective well-being, defined in terms of higher levels of positive affect and lower levels of negative affect (Fernández-Berrocal & Extremera, 2016). We could speak to this literature with respect to the level 2 MLMs that were performed. In the simulated emotion task, ability EI predicted lower levels of negative affect, b = -.157 [-.249, -.064], t = -3.35, p = .001, standardized b = -.097, but not higher levels of positive affect, b = .075 [-.036, .186], t = 1.34, p = .182, standardized b = .035. In the image reactivity task, ability EI did not predict average levels of either positive, b = -.021 [-.131, .089], t = -.27, p = .785, standardized b = -.012, or negative, b = .015 [-.092, .122], t = .27, p = .785, standardized b = .007, affect.

Did Levels of Emotional Reactivity Vary by Ability EI?

The primary hypothesis was that higher levels of ability EI would be linked to higher levels of emotional reactivity, defined in terms of stronger event/emotion relationships or within-person slopes (Nezlek, 2008). Given that such cross-level interactions among the predictors were central to hypotheses, we display results from these MLMs in Table 1. As can be seen in the table, all cross-level interactions were significant and all were of the same form – namely, event-emotion slopes were stronger at higher levels of ability EI. So, for example, the simple slope linking situation positivity to positive affect had a standardized b of .858 at a low (-1 SD) level of ability EI and a standardized b of 1.005 at a high (+ 1 SD) level. Differences among slopes were .316, -.154, -.282, .186, .323, and -.367 for positive event/positive affect, positive event/negative affect, negative event/positive affect, negative event/negative affect, image valence/positive affect, and image valence/negative affect relationships, respectively. In summary, although emotion systems were attuned to event valence for everyone, this was particularly the case at higher EI levels.

Table 1.

Ability EI as a Moderator of Event/Emotion Relationships, Study 1

Analysis and Predictors	b [95% CI]	t	p	b(S)
Situation Positivity/PA
Positivity	2.001 [1.930, 2.072]	55.14	< .001	.931
Ability EI	.075 [-.016, .166]	1.63	.106	.035
Interaction	.158 [.087, .230]	4.35	< .001	.074
Low EI (-1 SD)	1.843 [1.742, 1.944]	35.85	< .001	.858
High EI (+ 1 SD)	2.159 [2.058, 2.260]	42.00	< .001	1.005
Situation Positivity/NA
Positivity	-1.245 [-1.320, -1.169]	-32.22	< .001	-.774
Ability EI	-.157 [-.249, -.064]	-3.35	.001	-.097
Interaction	-.077 [-.153, -.001]	-1.99	.047	-.048
Low EI (-1 SD)	-1.168 [-1.275, -1.060]	-21.34	< .001	-.726
High EI (+ 1 SD)	-1.322 [-1.429, -1.214]	-24.16	< .001	-.822
Situation Negativity/PA
Negativity	-1.874 [-.1943, -1.805]	-52.96	< .001	-.872
Ability EI	.075 [-.016, .166]	1.63	.106	.035
Interaction	-.141 [-.210, -.071]	-3.96	< .001	-.066
Low EI (-1 SD)	-1.733 [-1.832, -1.635]	-34.58	< .001	-.807
High EI (+ 1 SD)	-2.015 [-2.113, -1.916]	-40.19	< .001	-.938
Situation Negativity/NA
Negativity	1.329 [1.251, 1.407]	33.25	< .001	.826
Ability EI	-.157 [-.249, -.064]	-3.35	.001	-.097
Interaction	.093 [.015, .172]	2.33	.020	.058
Low EI (-1 SD)	1.236 [1.125, 1.347]	21.82	< .001	.768
High EI (+ 1 SD)	1.422 [1.311, 1.533]	25.12	< .001	.884
Image Valence/PA
Valence	1.220 [1.131, 1.310]	26.72	< .001	.679
Ability EI	-.021 [-.131, .089]	-.38	.705	-.012
Interaction	.162 [.072, .251]	3.52	< .001	.090
Low EI (-1 SD)	1.059 [.932, 1.186]	16.37	< .001	.589
High EI (+ 1 SD)	1.382 [1.255, 1.508]	21.36	< .001	.769
Image Valence/NA
Valence	-1.639 [-1.727, -1.550]	-36.47	< .001	-.827
Ability EI	.015 [-.092, .122]	.27	.785	.007
Interaction	-.184 [-.272, -.095]	-4.07	< .001	-.093
Low EI (-1 SD)	-1.455 [-1.580, -1.330]	-22.86	< .001	-.735
High EI (+ 1 SD)	-1.822 [-1.947, -1.697]	-28.63	< .001	-.920

b(S) = standardized b; PA = Positive Affect; NA = Negative Affect

Estimated means (± 1 SD: Aiken & West, 1991) for the image reactivity task are displayed in Fig. 1. As shown there, differences were subtle, but event valence had a stronger effect on both positive and negative affect at higher, relative to lower, levels of ability EI. For example, relative to individuals with lower ability EI levels, those with higher levels experienced less negative affect when exposed to positive images and more negative affect when exposed to negative images.

Fig. 1.

Positive Affect (Top Panel) and Negative Affect (Bottom Panel) Estimated Means from Image Reactivity Task, Study 1

Additional Analyses

A series of additional MLMs were performed. The cross-level interactions displayed in Table 1 remained significant when controlling for participant sex (-1 = male; + 1 = female), |ts|> 1.50, ps < .05, standardized |bs|> .040. They also remained significant when controlling for extraversion and neuroticism, |ts|> 1.50, ps < .05, standardized |bs|> .040. Extraversion interacted with image valence to predict positive affect, b = .122 [.031, .214], t = 2.62, p = .009, standardized b = .068, but extraversion did not interact with event-related predictors with respect to the other emotion-related outcomes, |ts|< 1.50, ps > .10, standardized |bs|< .030. Neuroticism did not significantly moderate any of the event reactivity slopes, |ts|< 2.00, ps > .10, standardized |bs|< 0.050. Hence, the reactivity effects pertaining to ability EI were unique relative to those related to personality.

In another set of MLMs, we replaced NEAT total scores with scores particular to perception (one set of MLMs) or management (a second set of MLMs). In the simulated emotion task, perception-related cross-level interactions were consistently significant, |ts|> 2.50, ps < .01, standardized |bs|> .060, whereas management-related cross-level interactions were not, |ts|< 1.50, ps > .20, standardized |bs|< .030. However, in the task involving non-simulated induction stimuli (affective images), both perception scores, |ts|> 2.50, ps < .01, standardized |bs|> .050, and management, |ts|> 2.50, ps < .01, standardized |bs|> .050, moderated event-emotion slopes. We will revisit the question of particular branches after Study 2.

Discussion and Study 2

We have suggested that individuals possessing higher levels of ability EI are more skilled in evaluating the events they are exposed to (Robinson, Asad, & Irvin, 2023). This should render them more emotionally reactive to both positive and negative events, owing to the fact that evaluation is central to the appraisals that generate emotion (Roseman & Smith, 2001; Shuman et al., 2013). In Study 1, we provided evidence linking ability EI to increased emotional reactivity within two tasks, one of which involved simulated emotional reactions and the other of which involved responses to affective images. Each of these induction paradigms has advantages, particularly in examining within-person processes (Irvin et al., 2023; Waugh et al., 2011). Nonetheless, the evidence base would be strengthened if similar results occur in a daily diary protocol, which is suited to examine event-emotion slopes in the context of high levels of ecological validity (Lischetzke, 2014).

Method

Participants and General Procedures

Event-emotion relationships were again assessed using a multilevel design and we sought the same target as in Study 1 – namely, over 900 rows of data (Scherbaum & Ferreter, 2009). Undergraduate students from a Midwestern University in the United States signed up for a “Daily Experiences Study” using SONA software. After providing informed consent, they began a daily diary protocol that ran for 14 consecutive days. During each of these days, we emailed participants with subject number information and a link to a daily survey, which remained active from 7 p.m. until 9 a.m. the next morning. If a report was not completed within this time frame, it was deemed missing. Given the interest in within-person event-emotion slopes (West et al., 2011), we deleted data from the handful of participants who did not complete at least 9 of 14 daily reports. On the basis of adequate compliance, 137 participants were asked to complete a final online survey, which included assessments of personality and ability EI. Ninety-seven participants (69.07% female; 92.78 White; M age = 18.99) completed these assessments, resulting in a full, analyzable dataset consisting of 1244 rows.

Ability-Related Variations in Emotional Intelligence

Ability-related variations in emotional intelligence were again assessed with the NEAT (Krishnakumar et al., 2016). In Study 2, we presented the NEAT in its entirety, which consists of 30 scenarios, 10 each for perception, understanding, and management (120 total ratings). The perception and understanding branches are similar, in that both involve attributing emotions to protagonists encountering particular situations, but the understanding branch requires test-takers to rate the extent to which characters would experience blends of emotion (e.g., pride and gratitude) or transitions (e.g., interest, then pride), following Mayer et al. (2003). All 1–5 ratings were rescored using the same expert proportion consensus scoring system detailed in Study 1. We were most interested in a total score (M = .3190; SD = .0408; α = .89), but we also computed branch-specific scores for perception (M = .3668; SD = .0694; α = .88), understanding (M = .2822; SD = .0349; α = .63), and management (M = .3082; SD = .0438; α = .76), which inter-correlated with each other at about .50. Sex differences in ability EI (total scores) were not apparent, F(1, 96) = 1.76, p = .188, η_p² = .02, and follow-up analyses will omit this variable.

Extraversion and Neuroticism

For the sake of discriminant validity, we again assessed the personality traits of extraversion and neuroticism, which have been linked to emotional reactivity in some studies (e.g., Gross et al., 1998), but not others (e.g., Howell & Rodzon, 2011). In Study 2, these traits were assessed with 10-item Goldberg (1999) scales, which correlate highly with their NEO and BFI counterparts (John & Srivastava, 1999). Participants indicated how accurately particular statements described them (e.g., “I start conversations” for extraversion and “I worry about things” for neuroticism). After reverse-scoring statements reflective of low levels of the traits, we averaged across items to quantify individual differences in extraversion (M = 2.79; SD = 0.82; α = .82) and neuroticism (M = 2.21; SD = 0.76; α = .79). Ability-related variations in EI did not correlate with extraversion, r = -.02, p = .854, or neuroticism, r = -.10, p = .344.

Daily Diary Variables

The daily diary protocol included measures of emotional experience, behavior (not relevant to current report), and daily event occurrences, in that order, thus guarding against emotion reports that are biased by a prior consideration of behaviors or events. Given that the event measures were treated as predictors in MLMs, however, we will describe the event measures first. Participants were asked to indicate whether, and to what extent (1 = not at all true today; 4 = very much true today), positive and negative events had occurred to them on particular days. The event measures were deliberately generic and therefore encompassing of a wide variety of event types, whether social, academic, work-related, etc. (Robinson et al., 2022; Suls et al., 1998). Positive events were assessed with two items (“Today, something good happened to me” and “Today, I experienced a lot of pleasant events”), which were averaged (at the level of days, M = 2.57; SD = 0.87; α = .89). Negative events were also assessed with two items (“Today, something bad happened to me” and “Today, I experienced a lot of unpleasant events”), which were also averaged (M = 1.66; SD = 0.77; α = .86). For analyses purposes, event measures were person z-scored, which is a form of person-mean centering, given that person-mean centering is recommended for designs of this type (Enders & Tofighi, 2007; Wang & Maxwell, 2015).

Positive and negative emotional experiences were assessed with PANAS markers (Watson et al., 1988) that have performed well in previous daily diary studies (e.g., Moeller et al., 2014). With respect to positive experiences, participants were asked to rate the extent to which (1 = not at all; 4 = very much) they felt “excited” and “enthusiastic” on particular days. Positive affect was quantified by averaging across items (at the level of days, M = 2.62; SD = 0.9 = 0; α = .92). Participants were also asked to rate the extent to which they felt “distressed” and “nervous” on particular days and these two ratings were similarly averaged to quantify daily negative affect (M = 1.87; SD = 0.78; α = .63).

Results

Analysis Plan

Analyses were conducted on a multilevel dataset, with daily reports nested in participants. Within-subject predictors, which were person z-scored, consisted of the daily event measures. Between-subject predictors, which were z-scored, consisted of ability EI scores (primary analyses) and/or personality variables. The outcome was always a particular type of emotional experience, either positive or negative.

As in Study 1, we first conducted “level 1” models that focused on whether daily events predicted daily emotions within the sample as a whole. We then performed “level 2” models, which focused on the question of whether ability EI mattered for average levels of positive and negative affect. We then performed the critical cross-level analyses, which spoke to the primary question of whether event-emotion relationships varied as a function of ability EI levels. Follow-up analyses considered particular EI branches and/or the personality variables.

Were Normative Reactivity Tendencies Present?

The event measures were robust predictors of emotional experience. The occurrence of positive events predicted positive affect, b = .405 [.359, .452], t = 17.21, p < .001, standardized b = .451, as well as negative affect, b = -.138 [-.185, -.091], t = 5.79, p < .001, standardized b = -.177. Similarly, the occurrence of negative events predicted daily levels of positive affect, b = -.228 [-.285, -.172], t = -7.99, p < .001, standardized b = -.254, as well as daily levels of negative affect, b = .241 [.197, .286], t = 10.67, p < .001, standardized b = .310.

Did Ability EI Predict Average Levels of Emotional Experience?

We were interested in whether ability EI predicted average levels of positive and negative affect, independent of daily events that were encountered. This was not the case with respect to experiences of positive affect, b = .057 [-.049, .163], t = 1.07, p = .289, standardized b = .063, or negative affect, b = .065 [-.035, .165], t = 1.29, p = .199, standardized b = .083. Thus, events may need to be considered in understanding relations between ability EI and emotional experience.

Did Levels of Emotional Reactivity Vary by Ability EI?

Results for the primary cross-level MLMs are displayed in Table 2. As indicated there, all event/emotion relations were stronger at higher levels of ability EI. So, for example, the standardized slope linking positive events to positive affect was .387 at the low (-1 SD) level of ability EI and .513 at the high (+ 1 SD) level. Differences among slopes were .113, -.105, -.214, and .104 for positive event/PA, positive event/NA, negative event/PA, and negative event/NA relationships, respectively.

Table 2.

Ability EI as a Moderator of Event/Emotion Relationships, Study 2

Analysis and Predictors	b [95% CI]	p	t	b(S)
Positive Events/PA
Events	.405 [.360, .450]	17.67	< .001	.450
Ability EI	.056 [-.049, .162]	1.07	.289	.063
Interaction	.056 [.011, .102]	2.44	.015	.063
Low EI (-1 SD)	.348 [.284, .413]	10.65	< .001	.387
High EI (+ 1 SD)	.461 [.398, .525]	14.25	< .001	.513
Positive Events/NA
Events	-.137 [-.183, -.092]	-5.88	< .001	-.176
Ability EI	.057 [-.042, .156]	1.14	.257	.073
Interaction	-.053 [-.099, -.007]	-2.24	.025	-.068
Low EI (-1 SD)	-.085 [-.150, -.019]	-2.53	.011	-.109
High EI (+ 1 SD)	-.190 [-.255, -.126]	-5.57	< .001	-.244
Negative Events/PA
Events	-.230 [-.282, -.178]	-8.68	< .001	-.256
Ability EI	.057 [-.050, .163]	1.06	.293	.063
Interaction	-.107 [-.159, -.056]	-4.06	< .001	-.119
Low EI (-1 SD)	-.123 [-.196, -.050]	-3.30	.001	-.136
High EI (+ 1 SD)	-.337 [-.411, -.263]	-8.96	< .001	-.375
Negative Events/NA
Events	.242 [.199, .285]	10.93	< .001	.310
Ability EI	.073 [-.023, .170]	1.51	.135	.094
Interaction	.052 [.009, .095]	2.36	.019	.067
Low EI (-1 SD)	.190 [.129, .251]	6.12	< .001	.244
High EI (+ 1 SD)	.294 [.232, .356]	9.34	< .001	.377

b(S) = standardized b; PA = Positive Affect; NA = Negative Affect

Estimated means (± 1 SD) for positive event/PA and negative event/NA relationships are displayed in Fig. 2. The interactions were parallel to Fig. 1 in the sense that reactivity effects were evident at both low and high levels of EI, but were more pronounced at high levels. Viewed in another way, individuals with low and high levels of EI had comparable levels of positive and negative affect when daily event levels were low. When event levels were high, however, reactivity was more evident at higher EI levels.

Fig. 2.

Positive Event/Positive Affect (Top Panel) and Negative Event/Negative Affect (Bottom Panel) Relationships as a Function of Ability-Related Emotional Intelligence, Study 2

Additional Analyses

Extraversion moderated the relationship between negative events and negative affect, b = .033 [.005, .062], t = 2.29, p = .022, standardized b = .043, but did not moderate other event/emotion slopes, |ts|< 1.50, ps > .15, standardized |bs|< .030. None of the interactions involving neuroticism were significant, |ts|< 1.50, ps > .15, standardized |bs|< .030. Ability EI continued to moderate the event-emotion slopes depicted in Fig. 2 when controlling for both extraversion and neuroticism, |ts|> 2.00, ps < .05, standardized |bs|> .050 (in these analyses, p-values were .062 and < .001 for positive event/NA and negative event/PA interactions involving ability EI). Hence, the reactivity interactions observed in Fig. 2 were unique to ability EI relative to extraversion and neuroticism.

In another series of MLMs, we replaced NEAT total scores with scores particular to perception, understanding, or management. All three branches appeared to moderate event-emotion slopes in similar ways. Two of four interactions were significant for perception, 3 of 4 were significant for understanding, and 3 of 4 were significant for management. Hence, reactivity effects were not particular to only one branch of EI, but were evident across branches. Or, stated another way, it makes sense to emphasize global EI (as we have done) rather than any particular branch or set of skills with respect to the present findings.

General Discussion

The present studies are the first that we know of to apply experience-sampling procedures in investigating potential relations between ability EI and event-reactivity processes. In their simulated emotional reactions to a wide variety of realistic positive (e.g., “you just watched an entertaining movie”) and negative (e.g., you just stubbed your toe”) events, slopes linking event valence to emotional reactions were stronger at higher, relative to lower, levels of ability EI. Stronger event-emotion relationships at higher levels of EI were also found when assessing emotional reactions to affective images (Study 1) and in a daily diary protocol (Study 2). These modulatory effects were graded in nature, in that variations in event valence mattered for all or nearly all individuals. Nonetheless, the results provide consistent evidence for the hypothesized link between ability EI and emotional reactivity processes. That is, the emotional experiences of high EI individuals appear to be more contingent on events that occur, whether pleasant or unpleasant.

Implications and Analysis

When emotional intelligence is thought of in noun-like terms (e.g., as a body of knowledge about emotion), it can be difficult to know how that knowledge should interact with events to support particular experiences or behaviors (Mayer et al., 2016). Following several leads (e.g., Fiori, 2009; Miners et al., 2018), Robinson, Asad, & Irvin, (2023), instead, developed a verb-like conception of ability EI, contending that high ability individuals are more skilled at evaluating the events and stimuli that they are exposed to. In the present research, this theory gave rise to a new set of predictions concerning how ability EI should operate: If high EI individuals evaluate events and stimuli more accurately (Robinson, Asad, & Irvin, 2023), they should be more prone to positive feelings when exposed to positive stimuli, but more prone to negative feelings when exposed to negative stimuli (Barrett, 2006). Results supported these predictions and they did not support the alternative idea that higher levels of ability EI are necessarily linked to higher levels of affective well-being (i.e., average emotion-related states that tend to be more positive and less negative, irrespective of current circumstances).

Another way of thinking about the present designs is that they took seriously the possibility that a key signature of ability EI would involve dynamic processes (i.e., those varying in a within-subject manner: Fleeson, 2001). By varying event valence (or by allowing it to vary naturalistically: Study 2) and by repeatedly sampling positive and negative affective states, we were able to show that slopes linking events to emotional experiences were more pronounced at higher levels of ability EI. These results meet the call for dynamic investigations of ability EI (Ybarra et al., 2014), with the idea that EI is likely to operate in ways that produce dynamic signatures (MacCann et al., 2020). Because emotions change (Kuppens & Verduyn, 2017), and because such changes should be linked to distinct ecological and behavioral contexts (Moeller et al., 2014), key processes involved in ability EI may very well depend on dynamic links involving events, emotions, and behaviors (Moeller et al., 2014). Clearly, more research of this type is warranted.

Our results also speak to perspectives on emotional reactivity. Traditionally, investigators have operated under the assumption that emotional reactivity is problematic, in that it implicates an emotional system that is more easily destabilized (Bolger & Schilling, 1991; Rothbart et al., 2004). This perspective has given rise to the idea that emotional reactivity can compromise health (Smith, 2006) and result in behaviors that are impulsive in nature (Carver & Johnson, 2018). However, Gratz and Roemer (2004) have suggested that emotional reactivity should not be equated with emotion regulation difficulties and many traits and conditions that have traditionally been linked to emotional reactivity – such as neuroticism or borderline personality disorder – are inconsistent predictors of emotional reactivity phenomena (Bortolla et al., 2020; Howell & Rodzon, 2011; Kuo et al., 2016).

Meanwhile, a number of theories and sources of data have accrued to suggest that emotional reactivity is either functional or health-promoting. In psychological flexibility theory, in particular, psychological health is thought to be marked by flexible emotional and behavioral systems that are attuned to one’s current circumstances (Kashdan & Rottenberg, 2010). A core component of psychological flexibility is emotional flexibility, defined in terms of more pronounced emotional reactions to both positive and negative events (Beshai et al., 2018). In support of this theory, stronger emotional reactions to laboratory inductions have been found among individuals who are mindful (Beshai et al., 2018), resilient (Waugh et al., 2011), or happy (Klein et al., 2023). Conversely, a body of work has concluded that deficits in emotional reactivity appear to characterize conditions such as alexithymia (Aaron et al., 2018) and major depression (Bylsma et al., 2008).

Klein et al. (2023) extended the latter analysis in useful ways. According to these authors, emotional reactivity is functional when the emotional reactions are suited to the stimuli that produced them (also see Coifman & Bonanno, 2009). So, for example, strong reactions to weak (or barely valenced) stimuli would not be functional. In the present studies, valence was manipulated in strong ways (e.g., in relation to images that implicate physical harm or reproduction opportunities: Lang & Bradley, 2013). Klein et al. (2023) also suggest that it may be critical to distinguish the intensity and duration facets of emotional experience, as intense reactions that are short-lived are particularly consistent with the emotional flexibility construct (Koval et al., 2012). Finally, Klein et al. (2023) argue – and provide support for – the idea that stronger reactions are functional because they support adaptive behavioral decisions related to approach and avoidance (Elliot, 2006). Given the present results, further application of this model would be useful in the context of variations in ability EI.

Conclusions

Modest results have led multiple commentators to advocate for new directions in ability EI research (e.g., Miners et al., 2018; Ybarra et al., 2014). In the present research, we drew from a new theory of EI emphasizing its links to evaluation propensities and skills (Robinson, Asad, & Irvin, 2023). Such propensities, we reasoned, would tend to give rise to higher levels of emotional reactivity at higher levels of EI. Results, which were supportive of predictions, suggest new perspectives on ability EI as well as emotional reactivity phenomena.

Additional Information

Funding

There is no funding for this research.

Competing Interests

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Data Availability

Datasets for all studies can be found at https://osf.io/c6z4v/?view_only=5d47d878950c4d2eac93546723aa9a08. Neither study was preregistered.

Authors’ Contribution

All authors contributed to the designs and analyses of these studies. The first author wrote the paper, the second author collaborated heavily with respect to method and findings, and the second and third authors provided comments on earlier drafts.

Ethics Approval

The studies were approved by the first author’s institutional IRB (NDSU IRB0003825) and conducted in accordance with the 1964 Declaration of Helsinki.

Consent to Participate

Informed consent was obtained from all participants.

Consent for publication

Not applicable.