Psychoneuroendocrinology (2014) 44, 20—29
Available online at www.sciencedirect.com
ScienceDirect j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Basal salivary oxytocin level predicts extra- but not intra-personal dimensions of emotional intelligence Nancy S. Koven *, Laura K. Max Program in Neuroscience, Department of Psychology, Bates College, 4 Andrews Road, Lewiston, ME 04240, USA Received 17 December 2013; received in revised form 18 February 2014; accepted 24 February 2014
KEYWORDS Oxytocin; Emotional intelligence; Mayer—Salovey—Caruso Emotional Intelligence Test; Emotional states; Peripheral biomarker
Summary A wealth of literature suggests that oxytocin is an important mediator of social cognition, but much of the research to date has relied on pharmaceutical administration methods that can raise oxytocin to artificially high levels. The present study builds upon previous work by examining whether basal oxytocin level predicts intra- and extra-personal (i.e., self- and otherfocused) elements of emotional intelligence (EI), independent of shared variance with current mood. The sample included 71 healthy young adults (46 women). Assessment measures included the Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0 (MSCEIT), the Trait Meta-Mood Scale, and the Profile of Mood States. Peripheral oxytocin levels were examined with enzyme-linked immunosorbent assay from saliva after solid phase extraction. Oxytocin level was unrelated to TMMS scores but was positively associated with performance in the Experiential EI domain of the MSCEIT. However, total mood disturbance was positively related to MSCEIT scores. Hierarchical regression analysis indicated that oxytocin level added unique variance to the prediction of MSCEIT performance beyond that of current mood. These results confirm an association between endogenous levels of oxytocin in healthy adults and a subset of EI abilities, including extra-personal emotion recognition and the channeling of emotions to enhance social proficiency. # 2014 Elsevier Ltd. All rights reserved.
1. Introduction Social cognition requires the interconnection of many component processes to enable adaptive affiliations with others (Adolphs, 2001), and growing evidence highlights oxytocin
* Corresponding author. Tel.: +1 207 786 6426; fax: +1 207 786 8338. E-mail address: [email protected] (N.S. Koven). http://dx.doi.org/10.1016/j.psyneuen.2014.02.018 0306-4530/# 2014 Elsevier Ltd. All rights reserved.
(OT) as an important mediator of several of these processes. OT is thought to directly influence emotion perception by enhancing detection accuracy of emotional stimuli (Schulze et al., 2011), boosting the capacity to attend to the eye region of the face (Gamer et al., 2010), and increasing mentalization through a decoding of eye expression (MacDonald et al., 2013). Neuroimaging data have shown that OT administration attenuates activity in the amygdala during processing of fearful stimuli (Gamer et al., 2010) and dampens the functional connectivity between the amygdala and
Oxytocin and emotional intelligence other brain regions involved in fear processing (Kirsch et al., 2005), which, in conjunction with behavioral measures, has been taken to suggest that OT primarily enhances the encoding of positive social information (Guastella et al., 2008). In the domain of emotional understanding, exogenous OT enhances trust and willingness to accept risks during social tasks (Kosfeld et al., 2005), increases expressions of empathy (Krueger et al., 2013), and improves empathic accuracy in those with elevated Autism Spectrum Quotient scores (Bartz et al., 2010). An evolving view of OT, however, suggests that its behavioral effects are moderated by the social nature of the experimental context and by stable characteristics of the individuals under study (Bartz et al., 2011). Regarding context, it appears that OT facilitates approach behavior when social stimuli are salient but that some minimum number of social cues is needed; while these cues do not necessarily have to be derived from live, face-to-face encounters, the influence on prosocial behavior is contingent upon some form or expectation of a social milieu (Declerck et al., 2010). Regarding individual differences, data suggest that exogenous OT induces adverse hypersensitivity to social cues in individuals who are, for various reasons, already socially attuned (for review, see Olff et al., 2013). There is speculation that differential outcomes reflect pre-existing differences in basal OT (Bartz et al., 2011), such that negative socioemotional consequences of OT administration might be localized to individuals who already have naturally occurring high levels of OT. It is possible that inconsistencies in social cognition research in which OT levels are pharmaceutically manipulated reflect unmeasured trait-like individual variations in endogenous OT; however, to determine this, we need research that assesses the covariance between socioemotional functioning and endogenous OT levels. As social cognition is a broad domain, there are still gaps in our knowledge about the extent to which OT is associated with enhanced emotion processing in healthy adults. In this study, we focused on emotional intelligence (EI) as a more specific indicator of social cognition. EI includes the ability to perceive and generate emotions, respond appropriately to emotions in self and others, and utilize emotions to assist cognition. Although some of the aforementioned studies have incorporated behavioral paradigms related to EI (e.g., detection of emotional content in facial expressions), there is scarce research that simultaneously examines multiple aspects of EI in relation to OT. Within the domain of EI, a further distinction can be made between the personality traits that define the construct, which are typically assessed via self-report, and aptitudes, which are usually measured with objective, performance-based tests. A second differentiation can be made of EI as a set of meta-cognitive processes that enable reflection upon one’s own emotions versus those of others, two related but not perfectly overlapping activities (Dimaggio et al., 2008). Mindful of these points of difference, we included in our behavioral protocol a self-report EI measure, the Trait Meta Mood Scale (Salovey et al., 1995), and a broadband, performance-based EI test, the Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0 (MSCEIT; Mayer et al., 2002). The TMMS parses intra-personal (i.e., self-focused) EI into three variables: attention to emotion, clarity among emotions, and mood regulation. To our knowledge, no one has examined
21 whether systematic differences in TMMS scores co-vary with endogenous OT levels. The MSCEIT assesses both intra- and extra-personal (i.e., other-focused) EI across four discrete domains: the ability to identify emotions in self and others, use emotions to improve thinking and facilitate problemsolving, understand complex emotional meanings and situations, and manage emotions in self and others to achieve positive outcomes. In a recent study using a subset of the MSCEIT scales, Cardoso et al. (2014) found that people who received exogenous OT performed worse in facial emotion recognition than those in the placebo group, which they interpreted as supporting the idea that intranasal OT can, at times, be insalubrious by promoting emotional hypersensitivity. Because pharmaceutical administration typically elevates OT to levels many times greater than what is seen outside of the laboratory (Weisman et al., 2012), it is vital to understand the extent to which basal OT levels predict individual differences in EI. The present study extends the work of Cardoso et al. (2014) by assessing endogenous peripheral OT concentration in conjunction with the TMMS and the full MSCEIT, while controlling for negative affect as a potential confounder. Higher OT has been associated with greater acute anxiety (MacDonald et al., 2013) and greater depression (Holt-Lunstad et al., 2011). Furthermore, emotional well-being is known to account for significant variance in performance-based EI (Webb et al., 2013). In order to determine how OTrelates to EI independent of negative affect, we included the Profile of Mood States (POMS; McNair et al., 1992) in our protocol to provide an index of mood disturbance. For this study, we chose a nonclinical sample of young adults in order to capture a wide range of EI individual differences yet avoid the psychiatric comorbidities inherent in clinical samples. Given that OT enhances emotion processing in social contexts, we expected a relationship between basal OT concentration and the extra- but not intra-personal elements of EI. In other words, we expected OT level to co-vary with MSCEIT performance specifically in domains that emphasize the application of EI skills to the recognition, understanding, and management of emotions in other people; in contrast, we did not expect OT to co-vary with TMMS scores or with performance in MSCEIT domains that assess the recognition, understanding, and management of emotions within the self. Barring outlying subjects with very high OT levels, we anticipated the hypothesized relationship to be positive in direction, such that higher OT would be associated with higher extra-personal EI.
2. Method 2.1. Participants The sample included 71 right-handed, young adults (46 women; mean age = 18.9 1.2 yrs) who participated in the study for college course credit. Self-reported ethnicity was 90.8% White, 7.9% Asian/Asian-American, and 1.3% Latino/ Hispanic. Individuals were excluded if they reported current or past psychiatric/neurologic illness, current significant medical illness, current use of psychotropic medication, and nonfluency in English. The protocol was consistent with Ethical Guidelines of the Declaration of Helsinki and was approved by the local Institutional Review Board.
22
N.S. Koven, L.K. Max
2.2. Procedure Participants were individually administered a test battery in a quiet, distraction-minimized room. The protocol included provision of written informed consent, completion of a health and demographics questionnaire, the POMS, TMMS, and MSCEIT, followed by saliva collection.
2.3. Oxytocin assay Using a passive drool collection procedure (Grewen et al., 2010), participants were given pre-chilled 1.8 mL CyroTubes (Salimetrics, State College, PA, USA) and asked to provide at least 1.5 mL of saliva. Samples were promptly stored in a 80 8C freezer until assay. Peptide extraction was carried out prior to assay. This extraction step reduces matrix interference and concentrates the sample 3.2 times to ensure a level that is above the sensitivity threshold of the assay and detectible on the linear portion of the standard curve. The extraction procedure followed the recommendations of Grewen et al. (2010), utilizing Strata-X 33 mm polymeric reversed phase SPE sorbent tubes (Phenomenex, Torrance, CA, USA). The eluent was freeze-dried using a lyophilizer (Labconco, Kansas City, MO, USA) and then reconstituted in assay buffer. The ELISA protocol followed manufacturer’s suggestions (Enzo Life Sciences, Farmingdale, NY, USA). Optical density was measured at 405 nm using a BioTek Synergy HT microplate reader (Winooski, VT, USA), with concentration calculated according to a standard curve. Samples were run in duplicate, averaging the two values to generate one concentration value per participant. After correcting for the concentration resulting from the extraction step, the lower sensitivity limit of the assay was 2.04 pg/mL. Intra- and interCV were 7.4% and 9.8%, respectively.
2.4. Psychological assessment measures The TMMS is a 30-item EI scale consisting of three subscales: Attention to Feelings (e.g., ‘‘I pay a lot of attention to how I feel’’), Clarity of Feelings (e.g., ‘‘I am usually very clear about my feelings’’), and Mood Repair (e.g., ‘‘When I become upset, I remind myself of all the pleasures in my life’’). Higher scores reflect higher EI. Full-scale Cronbach’s alpha in this sample was .81.
Figure 1
The MSCEIT is a computer-administered EI battery, with data scored by Multi-Health Systems (North Tonawanda, NY, USA) using consensus-based norms, in which the score awarded to each response reflects the proportion of participants in the standardization sample who selected the same answer, and correcting for sex and age. The 141 items are divided across eight subtests. In the Faces task, participants are presented with images that display faces of strangers and asked to rate the degree to which a specified emotion is present. The Pictures task is similar to the Faces task, but the stimuli are designs and landscapes. In the Sensations task, participants are asked to match a labeled emotion to the corresponding bodily sensations that one would expect to experience simultaneously. In the Facilitation task, participants are asked to judge which mood best accompanies certain cognitive behaviors. In the Blends task, participants are asked to identify how emotions can be combined to form other emotions. The Changes task requires participants to select an emotion that would result from the intensification of a given feeling. In the Emotion Management task, participants are asked to read a story and identify the action that would help the character in the story achieve a specified emotional outcome. The Emotional Relationships task requires participants to identify which action would be most effective for a person to use in order to manage the feelings of other individuals. Performance across the eight subtests is calculated at the level of four branch scores, two area scores, and a total score (Fig. 1). Variables are reported as standard scores (M = 100; SD = 15) with higher values indicating higher EI. The full-scale reliability is .93, with area score reliabilities >.88 and branch score reliabilities >.79. The MSCEIT has shown high test—retest reliability at the branch, area, and total scale levels (Brackett and Mayer, 2003). Given concerns that MSCEIT subtest scores are less reliable than the superordinate scores, we followed the advice of Mayer et al. (2002) to focus specifically on area and branch scores. As a measure of current mood, the POMS contains 65 adjectives that comprise six subscales: Tension/Anxiety, Depression/Dejection, Anger/Hostility, Vigor/Activity, Fatigue/Inertia, and Confusion/Bewilderment. Total mood disturbance, calculated by subtracting the Vigor/Activity score from the sum of the other five scores, acts as a barometer of overall negative affect at the time of testing. A z-score was calculated (M = 0, SD = 1) based on available sex-related norms (McNair et al., 1992); higher z-scores reflect greater mood disturbance. Full-scale Cronbach’s alpha in this sample was .84.
Hierarchical structure of the MSCEIT.
Oxytocin and emotional intelligence
2.5. Data analytic approach Analyses were computed using SPSS software version 21.0 (Armonk, NY, USA). Normality of data was checked using Kolmogorov—Smirnov tests and with inspection of skew and kurtosis. Analyses involving the MSCEIT followed its hierarchical structure, beginning at the level of area scores and then probing downward into the relevant branch if/when statistical significance was achieved at the superordinate level. One-way analyses of variance (ANOVA) were used to determine if target variables (i.e., OT level, MSCEIT area and branch scores, TMMS subscale scores) varied by sex. To determine the relevance of the POMS total score as a confounder, we calculated two-tailed correlations between this variable, OT concentration, and EI scores. We examined twotailed bivariate correlations between OT level and EI scores. Finally, we used hierarchical regression to ascertain whether OT level contributed unique variance to EI beyond that of the POMS total score. An alpha value of 0.05 was used as the significance threshold for analyses except when application of a Bonferroni correction for multiple comparisons was warranted. For analyses concerning the MSCEIT, the family-wise alpha value was determined based on the number of comparisons being conducted per hierarchical level. Adjusted significance thresholds are indicated parenthetically in Section 3.
3. Results 3.1. Descriptive statistics Descriptive statistics are shown in Table 1. Two saliva samples were excluded because they were either too viscous for reliable extraction and/or contained less than the 0.8 mL of fluid necessary for extraction. Four samples had concentrations below the minimum detection level of 2.04 pg/mL and were therefore excluded from analysis. The remaining values were normally distributed and ranged from 2.05 pg/mL to 22.0 pg/ mL, which is comparable to other studies that have combined extraction with ELISA to measure peripheral OTconcentration (e.g., Holt-Lunstad et al., 2008; Hoffman et al., 2012). Superordinate MSCEIT variables were normally distributed, with mean performances in the Experiencing EI and Strategic EI areas falling within the average and high average ranges, respectively. Summary statistics for the TMMS subscale scores were consistent with those from other young adult samples (Coffey et al., 2003). The POMS total score was also normally distributed; the average z-score was less than half a standard deviation unit from the population mean, indicating minimal mood disturbance at the time of testing.
23 the branch level (a = .05/2) by a sex difference in Managing Emotions, F(1,69) = 9.59, p = .003, in which men (M = 111.2) outperformed women (M = 100.4). There were no sex differences in the Facilitating Thought branch of Strategic EI, F(1,69) = .89, p = ns. Similarly, there were no sex differences in the Experiential EI area, F(1,69) = .11, p = ns, or in its two constituent branches. Among TMMS subscales, men scored higher than women on the Clarity of Emotion, F(1,69) = 8.69, p = .004, and the Mood Repair, F(1,69) = 4.26, p = .04, subscales, although the latter effect does not survive Bonferroni correction (a = .05/3). Zero-order correlations (Table 1) in the full sample revealed that the POMS total score was unrelated to OT concentration. However, it was positively and uniformly correlated with both MSCEIT area scores (a = .05/2) and each subordinate branch score (a = .05/4) such that EI performance increased as mood disturbance increased. The POMS total score was also positively associated with the TMMS Clarity of Emotion subscale (a = .05/3).
3.3. Relationship between OT and EI At the level of MSCEIT area score (a = .05/2), OT level was positively associated with the MSCEIT Experiential EI area score (Table 1). As shown in Fig. 2, similar relationships emerged in the Perceiving Emotions and Facilitating Thought branches (a = .05/2). OT concentration had no relationship with MSCEIT scores in the Strategic EI branch or any of the TMMS subscales. We used hierarchical regression to determine if OT level contributed unique variance to MSCEIT performance beyond that of the POMS total score. Given the statistically significant zero-order correlations noted above, we targeted the Experiential EI area, the Perceiving Emotions branch, and the Facilitating Thought branch scores as separate dependent variables. In each regression analysis, the POMS total score was entered in Step 1 and OT concentration was entered in Step 2. Sex was excluded from these analyses, as no sex differences were found among scores in this arm of the MSCEITor in OTconcentration. Results of the three regression analyses are shown in Table 2. OT level added to the prediction of each dependent variable over and above the variance associated with the POMS total score, yet both OT level and POMS total score remained statistically significant contributors to MSCEIT performance in the full models. Examination of the square of the semi-partial correlations showed that OT contributed 10.8%, 11.9%, and 7.1% unique variance to Experiential EI, Perceiving Emotions, and Facilitating Thought, respectively. Mood disturbance contributed 16.3%, 9.8%, and 10.6% unique variance, respectively.
4. Discussion
3.2. Testing for sex differences and confounds ANOVA revealed no sex differences in OT concentration, F(1,63) = .14, p = ns. Even though the MSCEIT raw scores had been adjusted for sex during norming, men outperformed women at the level of area score (a = .05/2) in Strategic EI, F(1,69) = 11.36, p = .001, by a 15-point margin. Average Strategic EI standard scores for men and women were 125.6 and 110.6, respectively. This finding was driven at
The main findings of this study are twofold. First, we found that peripheral OT level, as measured with ELISA from extracted saliva samples, positively predicted Experiential EI on the MSCEIT in both men and women, independent of negative affect. Secondly, negative affect did contribute meaningful variance to Experiential EI though in a negative direction, such that EI scores improved with worsening mood. Each one is discussed in turn.
24
Table 1
Zero-order correlations among and descriptive statistics for variables of interest.
Variable 1. OT level 2. MSCEIT total 3. Area 1 4. Branch 1 5. Faces 6. Pictures 7. Branch 2 8. Sensations 9. Facilitation 10. Area 2 11. Branch 3 12. Blends 13. Changes 14. Branch 4 15. Management 16. Relations 17. TMMS Att 18. TMMS Clar 19. TMMS Rep 20. POMS total M SD
1
2
—
3 .32
*
4 **
.35 .85 **
—
5 **
.36 .69 ** .83 **
—
6 *
.30 .58 ** .70 ** .86 **
—
7 *
.25 .50 ** .61 ** .69 ** .33 **
—
8 *
.28 .74 ** .82 ** .38 ** .31 ** .31 **
—
9 .18 .67 ** .74 ** .43 ** .36 ** .37 ** .82 **
—
10 *
.29 .57 ** .62 ** .25 * .17 .22 .78 ** .34 **
—
11 .18 .79 ** .39 ** .28 ** .21 .23 .41 ** .37 ** .33 **
—
12 .11 .61 ** .27 * .17 .20 .02 .29 ** .15 .35 ** .76 **
—
13 .15 .60 ** .30 ** .19 .21 .05 .31 ** .17 .33 ** .68 ** .86 **
—
14 .06 .52 ** .22 .11 .11 .01 .27 * .15 .30 ** .68 ** .76 ** .59 **
—
15 .03 .55 ** .29 ** .20 * .11 .27 * .31 ** .36 ** .17 .71 ** .20 * .13 .30 **
—
16 .05 .37 ** .16 .05 .03 .16 .23 .23 .14 .53 ** .15 .07 .21 .78 **
—
17 .08 .58 ** .33 ** .26 * .17 .28 * .30 ** .38 ** .14 .72 ** .26 * .21 .37 ** .87 ** .43 **
—
18 .00 .26 * .29 * .22 .24 * .14 .24 * .29 * .06 .12 .02 .06 .03 .19 .29 * .06
—
19 .14 .31 ** .31 ** .26 * .26 * .08 .28 * .27 * .18 .23 .23 .22 .10 .12 .08 .09 .13
—
20 .11 .05 .00 .06 .05 .03 .06 .06 .16 .15 .09 .02 .12 .15 .16 .11 .21 .28 *
—
.05 .47 ** .42 ** .38 ** .31 ** .28 * .29 ** .30 ** .23 .39 ** .31 ** .31 ** .27 * .34 ** .15 .36 ** .08 .31 ** .22
— — 9.4 5.5
109.4 17.1
107.0 15.2
104.9 13.6
113.7 24.1
111.0 14.1
109.5 18.6
102.2 15.2
110.7 16.2
115.9 19.2
125.5 25.3
110.1 15.7
130.7 16.2
104.2 14.8
103.2 12.5
106.1 14.6
48.7 6.4
37.8 7.3
22.6 3.9
.2 .8
Note: OT, oxytocin (pg/mL); MSCEIT, Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0; Area 1, MSCEIT Experiential emotional intelligence; Branch 1, MSCEIT Perceiving Emotions; Branch 2, MSCEIT Facilitating Thought; Area 2, MSCEIT Strategic emotional intelligence; Branch 3, MSCEIT Understanding Emotions; Branch 4, MSCEIT Managing Emotions; TMMS, Trait Meta-Mood Scale; Att, Attention to Feelings; Clar, Clarity of Feelings; Rep, Mood Repair; POMS, Profile of Mood States. MSCEIT variables are reported as standard scores, with suggested interpretation as follows: 131 = very superior. * p < .05 ** p < .01
N.S. Koven, L.K. Max
Oxytocin and emotional intelligence
25
Figure 2 Scatterplots of oxytocin level versus the MSCEIT Perceiving Emotion (left panel) and Facilitating Thought (right panel) branch scores across subjects.
Experiential EI indexes the ability to perceive emotions accurately, appraise circumstances based on affective cues, and recognize that emotions can facilitate certain behaviors. There was a positive association between OT level and performance in both the Perceiving Emotions and Facilitating Thought branches within the Experiential EI area. With the caveat that MSCEIT subtest scores are less reliable than superordinate scores (Mayer et al., 2002), post hoc analysis at the subtest level suggests that the effect within the Perceiving Emotions branch was carried evenly by the Faces and Pictures subtests and that the effect within the Facilitating Thought branch was carried exclusively by the Facilitation subtest (Table 1). As tests of emotion recognition that use photographed strangers and pictures of landscapes as stimuli, respectively, the Faces and Pictures subtests clearly fall into the category of extra-personal EI. The Facilitation task reflects the degree to which individuals can harness
Table 2
emotions to facilitate reasoning, problem solving, and communication (Salovey and Grewal, 2005), content that is harder to designate as purely intra- or purely extra-personal. In support of its relevance to intra-personal abilities, Schneider et al. (2005) reported a positive correlation between the Facilitation subtest score and accuracy of autonomic, visceral self-perception (e.g., heartbeat detection). However, the Facilitation subtest also indexes the use of emotion to direct attention toward salient information about the environment and other people, to aid social judgment, and to facilitate perspective-taking (Brackett et al., 2011), all of which are arguably extra-personal in scope. Indeed, a high score in Facilitating Thought has implications for interpersonal behavior, including having a secure relationship attachment style in adulthood (Kafetsios, 2004) and being able to resist peer pressure in order to reason about risky behaviors in adolescence (Duncan et al., 2013). Interestingly, in a study
Summary of hierarchical regression results for predictions of MSCEIT performance.
Dependent variable
Predictor
Full model 2
1. Experiential EI area score
R = .29 F(2,62) = 12.36 **
Change statistics
R2 = .23 F(2,62) = 9.10 **
R2 = .19 F(2,62) = 7.07 ** POMS total OT level
t value
p value
.16 .11
.41 .33
3.77 3.07
Available online at www.sciencedirect.com
ScienceDirect j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Basal salivary oxytocin level predicts extra- but not intra-personal dimensions of emotional intelligence Nancy S. Koven *, Laura K. Max Program in Neuroscience, Department of Psychology, Bates College, 4 Andrews Road, Lewiston, ME 04240, USA Received 17 December 2013; received in revised form 18 February 2014; accepted 24 February 2014
KEYWORDS Oxytocin; Emotional intelligence; Mayer—Salovey—Caruso Emotional Intelligence Test; Emotional states; Peripheral biomarker
Summary A wealth of literature suggests that oxytocin is an important mediator of social cognition, but much of the research to date has relied on pharmaceutical administration methods that can raise oxytocin to artificially high levels. The present study builds upon previous work by examining whether basal oxytocin level predicts intra- and extra-personal (i.e., self- and otherfocused) elements of emotional intelligence (EI), independent of shared variance with current mood. The sample included 71 healthy young adults (46 women). Assessment measures included the Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0 (MSCEIT), the Trait Meta-Mood Scale, and the Profile of Mood States. Peripheral oxytocin levels were examined with enzyme-linked immunosorbent assay from saliva after solid phase extraction. Oxytocin level was unrelated to TMMS scores but was positively associated with performance in the Experiential EI domain of the MSCEIT. However, total mood disturbance was positively related to MSCEIT scores. Hierarchical regression analysis indicated that oxytocin level added unique variance to the prediction of MSCEIT performance beyond that of current mood. These results confirm an association between endogenous levels of oxytocin in healthy adults and a subset of EI abilities, including extra-personal emotion recognition and the channeling of emotions to enhance social proficiency. # 2014 Elsevier Ltd. All rights reserved.
1. Introduction Social cognition requires the interconnection of many component processes to enable adaptive affiliations with others (Adolphs, 2001), and growing evidence highlights oxytocin
* Corresponding author. Tel.: +1 207 786 6426; fax: +1 207 786 8338. E-mail address: [email protected] (N.S. Koven). http://dx.doi.org/10.1016/j.psyneuen.2014.02.018 0306-4530/# 2014 Elsevier Ltd. All rights reserved.
(OT) as an important mediator of several of these processes. OT is thought to directly influence emotion perception by enhancing detection accuracy of emotional stimuli (Schulze et al., 2011), boosting the capacity to attend to the eye region of the face (Gamer et al., 2010), and increasing mentalization through a decoding of eye expression (MacDonald et al., 2013). Neuroimaging data have shown that OT administration attenuates activity in the amygdala during processing of fearful stimuli (Gamer et al., 2010) and dampens the functional connectivity between the amygdala and
Oxytocin and emotional intelligence other brain regions involved in fear processing (Kirsch et al., 2005), which, in conjunction with behavioral measures, has been taken to suggest that OT primarily enhances the encoding of positive social information (Guastella et al., 2008). In the domain of emotional understanding, exogenous OT enhances trust and willingness to accept risks during social tasks (Kosfeld et al., 2005), increases expressions of empathy (Krueger et al., 2013), and improves empathic accuracy in those with elevated Autism Spectrum Quotient scores (Bartz et al., 2010). An evolving view of OT, however, suggests that its behavioral effects are moderated by the social nature of the experimental context and by stable characteristics of the individuals under study (Bartz et al., 2011). Regarding context, it appears that OT facilitates approach behavior when social stimuli are salient but that some minimum number of social cues is needed; while these cues do not necessarily have to be derived from live, face-to-face encounters, the influence on prosocial behavior is contingent upon some form or expectation of a social milieu (Declerck et al., 2010). Regarding individual differences, data suggest that exogenous OT induces adverse hypersensitivity to social cues in individuals who are, for various reasons, already socially attuned (for review, see Olff et al., 2013). There is speculation that differential outcomes reflect pre-existing differences in basal OT (Bartz et al., 2011), such that negative socioemotional consequences of OT administration might be localized to individuals who already have naturally occurring high levels of OT. It is possible that inconsistencies in social cognition research in which OT levels are pharmaceutically manipulated reflect unmeasured trait-like individual variations in endogenous OT; however, to determine this, we need research that assesses the covariance between socioemotional functioning and endogenous OT levels. As social cognition is a broad domain, there are still gaps in our knowledge about the extent to which OT is associated with enhanced emotion processing in healthy adults. In this study, we focused on emotional intelligence (EI) as a more specific indicator of social cognition. EI includes the ability to perceive and generate emotions, respond appropriately to emotions in self and others, and utilize emotions to assist cognition. Although some of the aforementioned studies have incorporated behavioral paradigms related to EI (e.g., detection of emotional content in facial expressions), there is scarce research that simultaneously examines multiple aspects of EI in relation to OT. Within the domain of EI, a further distinction can be made between the personality traits that define the construct, which are typically assessed via self-report, and aptitudes, which are usually measured with objective, performance-based tests. A second differentiation can be made of EI as a set of meta-cognitive processes that enable reflection upon one’s own emotions versus those of others, two related but not perfectly overlapping activities (Dimaggio et al., 2008). Mindful of these points of difference, we included in our behavioral protocol a self-report EI measure, the Trait Meta Mood Scale (Salovey et al., 1995), and a broadband, performance-based EI test, the Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0 (MSCEIT; Mayer et al., 2002). The TMMS parses intra-personal (i.e., self-focused) EI into three variables: attention to emotion, clarity among emotions, and mood regulation. To our knowledge, no one has examined
21 whether systematic differences in TMMS scores co-vary with endogenous OT levels. The MSCEIT assesses both intra- and extra-personal (i.e., other-focused) EI across four discrete domains: the ability to identify emotions in self and others, use emotions to improve thinking and facilitate problemsolving, understand complex emotional meanings and situations, and manage emotions in self and others to achieve positive outcomes. In a recent study using a subset of the MSCEIT scales, Cardoso et al. (2014) found that people who received exogenous OT performed worse in facial emotion recognition than those in the placebo group, which they interpreted as supporting the idea that intranasal OT can, at times, be insalubrious by promoting emotional hypersensitivity. Because pharmaceutical administration typically elevates OT to levels many times greater than what is seen outside of the laboratory (Weisman et al., 2012), it is vital to understand the extent to which basal OT levels predict individual differences in EI. The present study extends the work of Cardoso et al. (2014) by assessing endogenous peripheral OT concentration in conjunction with the TMMS and the full MSCEIT, while controlling for negative affect as a potential confounder. Higher OT has been associated with greater acute anxiety (MacDonald et al., 2013) and greater depression (Holt-Lunstad et al., 2011). Furthermore, emotional well-being is known to account for significant variance in performance-based EI (Webb et al., 2013). In order to determine how OTrelates to EI independent of negative affect, we included the Profile of Mood States (POMS; McNair et al., 1992) in our protocol to provide an index of mood disturbance. For this study, we chose a nonclinical sample of young adults in order to capture a wide range of EI individual differences yet avoid the psychiatric comorbidities inherent in clinical samples. Given that OT enhances emotion processing in social contexts, we expected a relationship between basal OT concentration and the extra- but not intra-personal elements of EI. In other words, we expected OT level to co-vary with MSCEIT performance specifically in domains that emphasize the application of EI skills to the recognition, understanding, and management of emotions in other people; in contrast, we did not expect OT to co-vary with TMMS scores or with performance in MSCEIT domains that assess the recognition, understanding, and management of emotions within the self. Barring outlying subjects with very high OT levels, we anticipated the hypothesized relationship to be positive in direction, such that higher OT would be associated with higher extra-personal EI.
2. Method 2.1. Participants The sample included 71 right-handed, young adults (46 women; mean age = 18.9 1.2 yrs) who participated in the study for college course credit. Self-reported ethnicity was 90.8% White, 7.9% Asian/Asian-American, and 1.3% Latino/ Hispanic. Individuals were excluded if they reported current or past psychiatric/neurologic illness, current significant medical illness, current use of psychotropic medication, and nonfluency in English. The protocol was consistent with Ethical Guidelines of the Declaration of Helsinki and was approved by the local Institutional Review Board.
22
N.S. Koven, L.K. Max
2.2. Procedure Participants were individually administered a test battery in a quiet, distraction-minimized room. The protocol included provision of written informed consent, completion of a health and demographics questionnaire, the POMS, TMMS, and MSCEIT, followed by saliva collection.
2.3. Oxytocin assay Using a passive drool collection procedure (Grewen et al., 2010), participants were given pre-chilled 1.8 mL CyroTubes (Salimetrics, State College, PA, USA) and asked to provide at least 1.5 mL of saliva. Samples were promptly stored in a 80 8C freezer until assay. Peptide extraction was carried out prior to assay. This extraction step reduces matrix interference and concentrates the sample 3.2 times to ensure a level that is above the sensitivity threshold of the assay and detectible on the linear portion of the standard curve. The extraction procedure followed the recommendations of Grewen et al. (2010), utilizing Strata-X 33 mm polymeric reversed phase SPE sorbent tubes (Phenomenex, Torrance, CA, USA). The eluent was freeze-dried using a lyophilizer (Labconco, Kansas City, MO, USA) and then reconstituted in assay buffer. The ELISA protocol followed manufacturer’s suggestions (Enzo Life Sciences, Farmingdale, NY, USA). Optical density was measured at 405 nm using a BioTek Synergy HT microplate reader (Winooski, VT, USA), with concentration calculated according to a standard curve. Samples were run in duplicate, averaging the two values to generate one concentration value per participant. After correcting for the concentration resulting from the extraction step, the lower sensitivity limit of the assay was 2.04 pg/mL. Intra- and interCV were 7.4% and 9.8%, respectively.
2.4. Psychological assessment measures The TMMS is a 30-item EI scale consisting of three subscales: Attention to Feelings (e.g., ‘‘I pay a lot of attention to how I feel’’), Clarity of Feelings (e.g., ‘‘I am usually very clear about my feelings’’), and Mood Repair (e.g., ‘‘When I become upset, I remind myself of all the pleasures in my life’’). Higher scores reflect higher EI. Full-scale Cronbach’s alpha in this sample was .81.
Figure 1
The MSCEIT is a computer-administered EI battery, with data scored by Multi-Health Systems (North Tonawanda, NY, USA) using consensus-based norms, in which the score awarded to each response reflects the proportion of participants in the standardization sample who selected the same answer, and correcting for sex and age. The 141 items are divided across eight subtests. In the Faces task, participants are presented with images that display faces of strangers and asked to rate the degree to which a specified emotion is present. The Pictures task is similar to the Faces task, but the stimuli are designs and landscapes. In the Sensations task, participants are asked to match a labeled emotion to the corresponding bodily sensations that one would expect to experience simultaneously. In the Facilitation task, participants are asked to judge which mood best accompanies certain cognitive behaviors. In the Blends task, participants are asked to identify how emotions can be combined to form other emotions. The Changes task requires participants to select an emotion that would result from the intensification of a given feeling. In the Emotion Management task, participants are asked to read a story and identify the action that would help the character in the story achieve a specified emotional outcome. The Emotional Relationships task requires participants to identify which action would be most effective for a person to use in order to manage the feelings of other individuals. Performance across the eight subtests is calculated at the level of four branch scores, two area scores, and a total score (Fig. 1). Variables are reported as standard scores (M = 100; SD = 15) with higher values indicating higher EI. The full-scale reliability is .93, with area score reliabilities >.88 and branch score reliabilities >.79. The MSCEIT has shown high test—retest reliability at the branch, area, and total scale levels (Brackett and Mayer, 2003). Given concerns that MSCEIT subtest scores are less reliable than the superordinate scores, we followed the advice of Mayer et al. (2002) to focus specifically on area and branch scores. As a measure of current mood, the POMS contains 65 adjectives that comprise six subscales: Tension/Anxiety, Depression/Dejection, Anger/Hostility, Vigor/Activity, Fatigue/Inertia, and Confusion/Bewilderment. Total mood disturbance, calculated by subtracting the Vigor/Activity score from the sum of the other five scores, acts as a barometer of overall negative affect at the time of testing. A z-score was calculated (M = 0, SD = 1) based on available sex-related norms (McNair et al., 1992); higher z-scores reflect greater mood disturbance. Full-scale Cronbach’s alpha in this sample was .84.
Hierarchical structure of the MSCEIT.
Oxytocin and emotional intelligence
2.5. Data analytic approach Analyses were computed using SPSS software version 21.0 (Armonk, NY, USA). Normality of data was checked using Kolmogorov—Smirnov tests and with inspection of skew and kurtosis. Analyses involving the MSCEIT followed its hierarchical structure, beginning at the level of area scores and then probing downward into the relevant branch if/when statistical significance was achieved at the superordinate level. One-way analyses of variance (ANOVA) were used to determine if target variables (i.e., OT level, MSCEIT area and branch scores, TMMS subscale scores) varied by sex. To determine the relevance of the POMS total score as a confounder, we calculated two-tailed correlations between this variable, OT concentration, and EI scores. We examined twotailed bivariate correlations between OT level and EI scores. Finally, we used hierarchical regression to ascertain whether OT level contributed unique variance to EI beyond that of the POMS total score. An alpha value of 0.05 was used as the significance threshold for analyses except when application of a Bonferroni correction for multiple comparisons was warranted. For analyses concerning the MSCEIT, the family-wise alpha value was determined based on the number of comparisons being conducted per hierarchical level. Adjusted significance thresholds are indicated parenthetically in Section 3.
3. Results 3.1. Descriptive statistics Descriptive statistics are shown in Table 1. Two saliva samples were excluded because they were either too viscous for reliable extraction and/or contained less than the 0.8 mL of fluid necessary for extraction. Four samples had concentrations below the minimum detection level of 2.04 pg/mL and were therefore excluded from analysis. The remaining values were normally distributed and ranged from 2.05 pg/mL to 22.0 pg/ mL, which is comparable to other studies that have combined extraction with ELISA to measure peripheral OTconcentration (e.g., Holt-Lunstad et al., 2008; Hoffman et al., 2012). Superordinate MSCEIT variables were normally distributed, with mean performances in the Experiencing EI and Strategic EI areas falling within the average and high average ranges, respectively. Summary statistics for the TMMS subscale scores were consistent with those from other young adult samples (Coffey et al., 2003). The POMS total score was also normally distributed; the average z-score was less than half a standard deviation unit from the population mean, indicating minimal mood disturbance at the time of testing.
23 the branch level (a = .05/2) by a sex difference in Managing Emotions, F(1,69) = 9.59, p = .003, in which men (M = 111.2) outperformed women (M = 100.4). There were no sex differences in the Facilitating Thought branch of Strategic EI, F(1,69) = .89, p = ns. Similarly, there were no sex differences in the Experiential EI area, F(1,69) = .11, p = ns, or in its two constituent branches. Among TMMS subscales, men scored higher than women on the Clarity of Emotion, F(1,69) = 8.69, p = .004, and the Mood Repair, F(1,69) = 4.26, p = .04, subscales, although the latter effect does not survive Bonferroni correction (a = .05/3). Zero-order correlations (Table 1) in the full sample revealed that the POMS total score was unrelated to OT concentration. However, it was positively and uniformly correlated with both MSCEIT area scores (a = .05/2) and each subordinate branch score (a = .05/4) such that EI performance increased as mood disturbance increased. The POMS total score was also positively associated with the TMMS Clarity of Emotion subscale (a = .05/3).
3.3. Relationship between OT and EI At the level of MSCEIT area score (a = .05/2), OT level was positively associated with the MSCEIT Experiential EI area score (Table 1). As shown in Fig. 2, similar relationships emerged in the Perceiving Emotions and Facilitating Thought branches (a = .05/2). OT concentration had no relationship with MSCEIT scores in the Strategic EI branch or any of the TMMS subscales. We used hierarchical regression to determine if OT level contributed unique variance to MSCEIT performance beyond that of the POMS total score. Given the statistically significant zero-order correlations noted above, we targeted the Experiential EI area, the Perceiving Emotions branch, and the Facilitating Thought branch scores as separate dependent variables. In each regression analysis, the POMS total score was entered in Step 1 and OT concentration was entered in Step 2. Sex was excluded from these analyses, as no sex differences were found among scores in this arm of the MSCEITor in OTconcentration. Results of the three regression analyses are shown in Table 2. OT level added to the prediction of each dependent variable over and above the variance associated with the POMS total score, yet both OT level and POMS total score remained statistically significant contributors to MSCEIT performance in the full models. Examination of the square of the semi-partial correlations showed that OT contributed 10.8%, 11.9%, and 7.1% unique variance to Experiential EI, Perceiving Emotions, and Facilitating Thought, respectively. Mood disturbance contributed 16.3%, 9.8%, and 10.6% unique variance, respectively.
4. Discussion
3.2. Testing for sex differences and confounds ANOVA revealed no sex differences in OT concentration, F(1,63) = .14, p = ns. Even though the MSCEIT raw scores had been adjusted for sex during norming, men outperformed women at the level of area score (a = .05/2) in Strategic EI, F(1,69) = 11.36, p = .001, by a 15-point margin. Average Strategic EI standard scores for men and women were 125.6 and 110.6, respectively. This finding was driven at
The main findings of this study are twofold. First, we found that peripheral OT level, as measured with ELISA from extracted saliva samples, positively predicted Experiential EI on the MSCEIT in both men and women, independent of negative affect. Secondly, negative affect did contribute meaningful variance to Experiential EI though in a negative direction, such that EI scores improved with worsening mood. Each one is discussed in turn.
24
Table 1
Zero-order correlations among and descriptive statistics for variables of interest.
Variable 1. OT level 2. MSCEIT total 3. Area 1 4. Branch 1 5. Faces 6. Pictures 7. Branch 2 8. Sensations 9. Facilitation 10. Area 2 11. Branch 3 12. Blends 13. Changes 14. Branch 4 15. Management 16. Relations 17. TMMS Att 18. TMMS Clar 19. TMMS Rep 20. POMS total M SD
1
2
—
3 .32
*
4 **
.35 .85 **
—
5 **
.36 .69 ** .83 **
—
6 *
.30 .58 ** .70 ** .86 **
—
7 *
.25 .50 ** .61 ** .69 ** .33 **
—
8 *
.28 .74 ** .82 ** .38 ** .31 ** .31 **
—
9 .18 .67 ** .74 ** .43 ** .36 ** .37 ** .82 **
—
10 *
.29 .57 ** .62 ** .25 * .17 .22 .78 ** .34 **
—
11 .18 .79 ** .39 ** .28 ** .21 .23 .41 ** .37 ** .33 **
—
12 .11 .61 ** .27 * .17 .20 .02 .29 ** .15 .35 ** .76 **
—
13 .15 .60 ** .30 ** .19 .21 .05 .31 ** .17 .33 ** .68 ** .86 **
—
14 .06 .52 ** .22 .11 .11 .01 .27 * .15 .30 ** .68 ** .76 ** .59 **
—
15 .03 .55 ** .29 ** .20 * .11 .27 * .31 ** .36 ** .17 .71 ** .20 * .13 .30 **
—
16 .05 .37 ** .16 .05 .03 .16 .23 .23 .14 .53 ** .15 .07 .21 .78 **
—
17 .08 .58 ** .33 ** .26 * .17 .28 * .30 ** .38 ** .14 .72 ** .26 * .21 .37 ** .87 ** .43 **
—
18 .00 .26 * .29 * .22 .24 * .14 .24 * .29 * .06 .12 .02 .06 .03 .19 .29 * .06
—
19 .14 .31 ** .31 ** .26 * .26 * .08 .28 * .27 * .18 .23 .23 .22 .10 .12 .08 .09 .13
—
20 .11 .05 .00 .06 .05 .03 .06 .06 .16 .15 .09 .02 .12 .15 .16 .11 .21 .28 *
—
.05 .47 ** .42 ** .38 ** .31 ** .28 * .29 ** .30 ** .23 .39 ** .31 ** .31 ** .27 * .34 ** .15 .36 ** .08 .31 ** .22
— — 9.4 5.5
109.4 17.1
107.0 15.2
104.9 13.6
113.7 24.1
111.0 14.1
109.5 18.6
102.2 15.2
110.7 16.2
115.9 19.2
125.5 25.3
110.1 15.7
130.7 16.2
104.2 14.8
103.2 12.5
106.1 14.6
48.7 6.4
37.8 7.3
22.6 3.9
.2 .8
Note: OT, oxytocin (pg/mL); MSCEIT, Mayer—Salovey—Caruso Emotional Intelligence Test Version 2.0; Area 1, MSCEIT Experiential emotional intelligence; Branch 1, MSCEIT Perceiving Emotions; Branch 2, MSCEIT Facilitating Thought; Area 2, MSCEIT Strategic emotional intelligence; Branch 3, MSCEIT Understanding Emotions; Branch 4, MSCEIT Managing Emotions; TMMS, Trait Meta-Mood Scale; Att, Attention to Feelings; Clar, Clarity of Feelings; Rep, Mood Repair; POMS, Profile of Mood States. MSCEIT variables are reported as standard scores, with suggested interpretation as follows: 131 = very superior. * p < .05 ** p < .01
N.S. Koven, L.K. Max
Oxytocin and emotional intelligence
25
Figure 2 Scatterplots of oxytocin level versus the MSCEIT Perceiving Emotion (left panel) and Facilitating Thought (right panel) branch scores across subjects.
Experiential EI indexes the ability to perceive emotions accurately, appraise circumstances based on affective cues, and recognize that emotions can facilitate certain behaviors. There was a positive association between OT level and performance in both the Perceiving Emotions and Facilitating Thought branches within the Experiential EI area. With the caveat that MSCEIT subtest scores are less reliable than superordinate scores (Mayer et al., 2002), post hoc analysis at the subtest level suggests that the effect within the Perceiving Emotions branch was carried evenly by the Faces and Pictures subtests and that the effect within the Facilitating Thought branch was carried exclusively by the Facilitation subtest (Table 1). As tests of emotion recognition that use photographed strangers and pictures of landscapes as stimuli, respectively, the Faces and Pictures subtests clearly fall into the category of extra-personal EI. The Facilitation task reflects the degree to which individuals can harness
Table 2
emotions to facilitate reasoning, problem solving, and communication (Salovey and Grewal, 2005), content that is harder to designate as purely intra- or purely extra-personal. In support of its relevance to intra-personal abilities, Schneider et al. (2005) reported a positive correlation between the Facilitation subtest score and accuracy of autonomic, visceral self-perception (e.g., heartbeat detection). However, the Facilitation subtest also indexes the use of emotion to direct attention toward salient information about the environment and other people, to aid social judgment, and to facilitate perspective-taking (Brackett et al., 2011), all of which are arguably extra-personal in scope. Indeed, a high score in Facilitating Thought has implications for interpersonal behavior, including having a secure relationship attachment style in adulthood (Kafetsios, 2004) and being able to resist peer pressure in order to reason about risky behaviors in adolescence (Duncan et al., 2013). Interestingly, in a study
Summary of hierarchical regression results for predictions of MSCEIT performance.
Dependent variable
Predictor
Full model 2
1. Experiential EI area score
R = .29 F(2,62) = 12.36 **
Change statistics
R2 = .23 F(2,62) = 9.10 **
R2 = .19 F(2,62) = 7.07 ** POMS total OT level
t value
p value
.16 .11
.41 .33
3.77 3.07
