The Relationship Between Narcissistic Exploitativeness, Dispositional Empathy, and Emotion Recognition Abilities
- First Online:
- Cite this article as:
- Konrath, S., Corneille, O., Bushman, B.J. et al. J Nonverbal Behav (2014) 38: 129. doi:10.1007/s10919-013-0164-y
- 913 Views
The present research explores the link between the personality trait exploitativeness, a component of narcissism, and emotion recognition abilities. Prior research on this topic has produced inconsistent findings. We attempt to resolve these inconsistencies by testing the hypothesis that narcissistic exploitativeness, in particular, should be associated with emotion-reading abilities because it specifically taps into the motivation to manipulate others. Across two studies we find that narcissistic exploitativeness is indeed associated with increased emotion recognition, but in some cases the confounding effects of mood need to be considered (Study 1). Importantly, effect sizes of narcissistic exploitativeness were similar in magnitude to two different measures of dispositional empathy, which is an established correlate of emotion recognition. These studies suggest that emotional recognition abilities are associated with desirable and undesirable traits.
KeywordsNarcissismExploitativenessEmotional competenciesEmotional intelligenceEmotion recognitionEmpathyMind readingEmotion perception ability
“I can read people like a book.”
~Item from the Narcissistic Personality Inventory
Some might be surprised that the above item is included in a commonly used measure of narcissism, a personality trait characterized by excessive self-love combined with low empathy. Indeed, this item comes from the exploitativeness subscale of the Narcissistic Personality Inventory (Raskin and Terry 1988), and its inclusion is surprising because the ability to read other’s emotions is seen as central to empathy, which is the tendency to vicariously feel, imagine, and value others’ feelings and perspectives (Davis 1983). In turn, empathic competencies are important building blocks of deep and lasting social connections (Davis 1996). Yet people scoring high in narcissism are notoriously bad at connecting with others, often treating people as objects that exist only to facilitate their own desires. Narcissistic people have difficulties in maintaining relationships, and also tend to be hostile and aggressive, especially those high in narcissistic entitlement and exploitativeness (Bushman and Baumeister 1998; Campbell 1999; Konrath et al. 2006; Reidy et al. 2008). In short, narcissism is undesirable for social interaction partners.
Narcissism research has primarily focused on such undesirable interpersonally-relevant correlates. This is problematic because normal, non-clinical narcissism is distinct from Narcissistic Personality Disorder, and may be seen as an extreme form of individualism (Konrath et al. 2009), rather than a pathological state. Thus, narcissism can also be associated with a number of advantages, but mostly for narcissistic individuals themselves rather than for others (e.g., more happiness, lower anxiety and depression; Raskin and Novacek 1989; Rose 2002; Sedikides et al. 2004).
There has been limited attention to potential interpersonal advantages of narcissism. Overall, previous research finds that there may be some interpersonal advantages, but only in the short term. People scoring high in narcissism make positive first impressions. Based on short social interactions, brief video clips, or still photographs, they are perceived as attractive, agreeable, competent, well-adjusted, and popular by social interaction partners (Back et al. 2010; Holtzman and Strube 2010; Paulhus 1998). However, social interaction partners begin to rate them more negatively after repeated interactions, once they get to know what they are really like (Paulhus 1998). Up until now this research has focused on how others perceive people scoring high in narcissism, which is important to understand. Yet in the current study we extend research on interpersonal advantages associated with narcissism by examining how it (and more specifically, its exploitativeness dimension) relates to accurate judgments of others’ emotions and affective states.
Emotional Competencies: The Good and the Bad
Emotional competencies involve accurate emotion perception, using and regulating emotions appropriately, and understanding and communicating them effectively (Salovey and Mayer 1990). Some scholars define these competencies more broadly to also include social and relationship skills (“social intelligence”), and aspects of psychological well-being (Pérez et al. 2005). Researchers have studied emotional competencies as both personality traits (Mikolajczak et al. 2007; Petrides and Furnham 2001), and abilities, in which a variety of objective performance indicators are assessed (Salovey and Mayer 1990).
Both approaches commonly find that emotional competencies are associated with desirable traits and abilities. For example, self-reported (trait) emotional competencies are associated with qualities that help to initiate and maintain relationships (e.g., empathy, social skills, cooperativeness; Schutte et al. 2001), and they are also associated with a higher quantity and quality of social relationships (Lopes et al. 2003; Schutte et al. 2001). Parallel results have been found for ability measures of emotional competencies (Davis and Kraus 1997; Hall et al. 2009; Mayer et al. 2008).
The vast majority of research to date has examined more desirable correlates of emotion recognition capacities, such as higher dispositional empathy (Davis and Kraus 1997; Hall et al. 2009). A recent meta-analysis examined the relationship between interpersonal sensitivity, which includes emotion recognition abilities, and various correlates (Hall et al. 2009). Over five times as many results focused on positive personality traits and social outcomes (k = 316; Table 3) compared to negative personality traits (k = 60). Of the positive traits, the majority of results showed that interpersonal sensitivity and positive traits such as dispositional empathy were significantly correlated.
Of the less desirable traits, there were some negative correlations between interpersonal sensitivity and intrapersonally relevant traits (e.g., neuroticism, shyness). Yet only a few studies focused on traits with important interpersonal implications (i.e., aggression: k = 9, Mean r = −.03, ns; Machiavellianism: k = 4; Mean r = −.07, p < .05; Hall et al. 2009). Overall, there has not been sufficient research attention to the potential darker side of such capacities, especially with respect to interpersonally relevant traits.
In the current research, we examine the relationship between aspects of the personality trait narcissism and one form of emotional competency: emotion recognition. Just because a construct often has positive correlates does not rule out a potentially darker side. Some scholars are aware of this, admitting that emotional competencies can be “channeled antisocially [and] may create manipulative scenes or lead others sociopathically to nefarious ends” (Salovey and Mayer 1990, p. 198). Other work has linked high trait emotional intelligence to excessive pride (Petrides 2010). Unfortunately though, the majority of research has largely neglected this potentially darker side of emotional competencies in general, and emotion recognition abilities, in particular (Hall et al. 2009).
Some intriguing new research, however, is illuminating. For example, forensic patients scoring high in psychopathy, a trait that is highly correlated with narcissism (Paulhus and Williams 2002), score higher on trait measures of emotional intelligence, and report that they find it easy to identify and recognize emotions (Pham et al. 2010). In other words, psychopathic individuals perceive themselves as excelling in emotional competencies. It is unlikely that psychopaths use their emotion recognition skills for empathic purposes, but they may instead use these skills to manipulate others. Yet, contradictory findings exist for the relationship between antisocial personality disorder and emotion recognition ability, with a meta-analysis finding specific deficits in recognizing fearful facial expressions (Marsh and Blair 2008).
Among non-clinical populations, there are intriguing findings, but the literature remains sparse. For example, childhood bullies score higher on “theory of mind” tests assessing their ability to imagine others’ perspectives and feelings compared to non-bullies (Sutton et al. 1999). And although some work has found fewer emotional competencies among Machiavellians, or emotionally manipulative people (Austin et al. 2007), one recent paper finds that certain emotionally competent people (i.e., those scoring high in emotion regulation) channel this ability antisocially if they also score high on Machiavellianism, but channel it more prosocially if they score high on a more other-oriented personality trait (Côté et al. 2011).
Taken together, there is reason to believe that the relationships between emotional competencies and darker psychological traits may exist at times, but may be complex.
Narcissism and Emotional Competencies
To date there has been limited research on the extent to which people scoring high in narcissism understand and use emotional information. High narcissism scorers self-report being emotionally competent (Ames and Kammrath 2004; Petrides et al. 2011). However, narcissistic people self-enhance, so it is possible that they say they are emotionally gifted, but that in fact they are not. Supporting this idea, one prior study found that although narcissists report better-than-average emotion recognition skills within social interactions, there was actually no relationship between narcissism and actual emotion recognition (Ames and Kammrath 2004). This suggests that narcissists may inflate their emotion perception abilities
However, narcissism is a multifactorial construct, with a number of more (e.g., entitlement, exhibitionism, exploitativeness) or less (e.g., self-sufficiency, superiority, vanity, leadership/authority) dysfunctional aspects (Raskin and Novacek 1989; Raskin and Terry 1988). Theoretically, some factors should be associated with advantages in emotional competencies, while others should not. For example, people who score high in the exploitativeness factor of narcissism find it easy to control others and take advantage of them (Raskin and Novacek 1989; Raskin and Terry 1988). They have non-reciprocal social interactions and see other people as a means to achieve their own goals, rather than as ends in themselves. As such, exploitativeness might be associated with good emotion-reading abilities, because reading others’ emotions could be a useful skill in order to successfully get one’s own way.
Although the narcissism dimensions tend to be correlated with one another, they each represent slightly different aspects of egotism that may have less direct relevance to emotion reading capabilities. For example, people scoring high in narcissistic entitlement believe that they deserve the best, which makes them focus on discrepancies between what they want and what they actually have. Exploitativeness may be the motivational engine that drives entitled people to get what they “deserve”, but entitlement itself shouldn’t necessarily be associated with good emotion reading. Similarly, the other narcissism subscales do not necessarily power the motivational engine to read others well.
Unfortunately, prior research has not specifically examined the relationship between narcissistic dimensions, such as exploitativeness, and emotion reading (Ames and Kammrath 2004). Thus, it remains possible that some aspects of narcissism are indeed associated with advantages in reading others’ emotions.
In addition, it is possible that narcissism is associated with better performance on certain emotions, but not others. One recent study found that narcissism was associated with an increased ability to read angry facial expressions, but no advantage when it came to identifying happy, sad, or fearful expressions (Wai and Tiliopoulos 2012). However, this study also did not specifically examine the role of specific dimensions such as exploitativeness. Given the very limited amount of past research and these mixed results, it is important to further pursue this area of inquiry.
In the current research, we examined the extent to which one aspect of narcissism (i.e., exploitativeness) is associated with emotional recognition abilities on validated emotion recognition tasks. We predicted that exploitativeness would be associated with better emotion recognition performance. We also examined whether exploitativeness would be associated with better recognition of positive versus negative emotions. We did not have a priori theoretical predictions about valence, but examining this question can help us to better understand the potential utility of accurate emotion recognition among exploitative people. For example, it is possible that those scoring high in exploitativeness might be especially sensitive to positive emotions because of the potential for positive emotions to signal others’ positive regard for them, but it is also possible that exploitative individuals are especially good at identifying negative emotions because such emotions often imply some sort of vulnerability and thus ripeness for potential exploitation. We return to these ideas in the General Discussion.
In both studies, we also aim to conceptually replicate prior research that people scoring high in dispositional empathy have better emotion recognition performance compared to low scorers. Demonstrating that exploitativeness and dispositional empathy both simultaneously are associated with emotion recognition ability would help to demonstrate both the desirable and undesirable correlates of this ability.
Study 1 examined the extent to which exploitativeness is associated with emotional competencies compared to other aspects of narcissism and compared to an established correlate of emotion recognition—dispositional empathy. We predicted that exploitativeness would be associated with a better recognition of affective states. In addition, we expected to replicate established research findings that empathic individuals are better able to recognize others’ affective states and emotions (Hall et al. 2009; Riggio et al. 1989).
Participants were 96 American college students who received $15 for their participation. One participant was dropped after admitting that he put random numbers on the questionnaire, leaving a final sample size of 95, although the results did not significantly differ when his data were included. The final sample was 77 % female with a mean age of 20.9 (SD = 2.3). The ethnic composition of the sample was 44 % Caucasian, 38 % Asian-American, 12 % African-American, 6 % Other.
Procedure and Measures
Participants completed measures of narcissism, empathy, and emotion recognition. They also reported their current mood because it can influence emotion recognition performance (Schmid and Schmid Mast 2010), and because positive mood is also consistently related to narcissism (Rose 2002; Zuckerman and O’Loughlin 2009).
Narcissism was assessed using the 16-item forced-choice Narcissistic Personality Inventory (Ames et al. 2006). Three items were used to calculate the exploitativeness subscale (Cronbach α = .50; This alpha is low, but comparable to the alpha of .52 for the exploitativeness subscale found in the original NPI development paper). A sample item is: “I find it easy to manipulate people” (scored 1) versus “I don’t like it when I find myself manipulating people” (scored 0).1 Dispositional empathy was assessed using the empathy subscale from the Trait Emotional Intelligence Questionnaire (Petrides and Furnham 2003), which consists of 9 items (e.g., “I’m normally able to ‘get into someone’s shoes’ and experience their emotions”) scored using a 7-point scale (1 = completely disagree, 7 = completely agree; Cronbach α = .77).
Emotional competency was assessed using the Reading the Mind in the Eyes Test (Baron-Cohen et al. 2001), a widely used assessment of affect/emotion recognition in clinical (e.g., autism) and non-clinical populations (Baron-Cohen et al. 2001; Chapman et al. 2006; Luminet et al. 2011; Tso et al. 2010). Participants received a binder with 17 images of eyes depicting various specific affective expressions, and were asked to choose the affective term that best described each expression from an array of four terms. The RMET includes a variety of affective states that are more specific than the basic emotions (e.g., regretful, rather than sad). Participants could take as much time as they wished for this task and it was not timed. Scores ranged from 0 = none correct to 17 = all correct (Cronbach α = .23). Unfortunately the original RMET paper does not list the reliability so no comparisons could be made. We address the low Cronbach alphas in the General Discussion.
Affective states were objectively categorized into valence categories based on Whissell’s Dictionary of Affect in Language (DAL) program (Whissell 2002). The DAL is a validated computerized corpus of English words rated by subject volunteers on valence (1 = Unpleasant, 2 = In between, 3 = Pleasant), activation (1 = Passive, 2 = In between, 3 = Active), and imagery (1 = Hard to imagine, 2 = In between, 3 = Easy to Imagine; Sweeney and Whissell 1984; Whissell 2008, 2009; Whissell and Charuk 1985; Whissell and Dewson 1986; Whissell et al. 1986). Six RMET words were categorized as positive (playful, anticipating, desire, insisting, thoughtful, fantasizing) and 9 were categorized as negative (accusing, regretful, preoccupied, worried, cautious, doubtful, skeptical, upset, uneasy). Two words were not available in the dictionary, so could not be included (despondent, contemplative) in the valence-based analyses. Positive affective states were significantly higher in pleasantness (M = 2.23, SD = .28, Range = 1.9–2.4) than negative affective states (M = 1.23, SD = .11, Range = 1.0–1.3), F(1, 13) = 96.35, p < .001, but positive and negative affective states did not differ in activation or imagery (ps > .15).
We also included measures of positive (Cronbach α = .89) and negative mood (Cronbach α = .85) as covariates in our analyses (Watson et al. 1988).
Correlation table with all variables from Study 1 (above diagonal) and Study 2 (below diagonal)
1. Total emotion recognition
2. Positive emotion recognition
3. Negative emotion recognition
6. Positive mood
7. Negative mood
Predictors of emotion recognition on the Reading the Mind in the Eyes Task (Study 1)
Exploitativeness factor (NPI-16)
Empathy subscale of trait emotional intelligence (TEI)
Regression Model 1
Total emotion recognition (RMET)
Controlling for mood
Regression Model 2
Positive emotion recognition (RMET)
Controlling for mood
Regression Model 3
Negative emotion recognition (RMET)
Controlling for mood
Study 1 found that narcissistic exploitativeness and dispositional empathy were both associated with better recognition of affective states when mood was included as a covariate. Importantly, the relationship between exploitativeness and affective state recognition was similar in magnitude as a more established correlate (i.e., dispositional empathy). Mood may serve as an important covariate because participants’ traits (empathy, exploitativeness) might make certain moods more likely, and in turn, these moods might affect the recognition of affective and emotional states. In order to fully understand the role of mood, studies specifically manipulating narcissism, empathy, and mood would be needed. However, before such studies are recommended, it would be important to find a consistent relationship with respect to mood as a covariate in a separate study. This is something that we can address in Study 2.
Another potentially interesting finding that would be important to replicate in Study 2 was whether exploitativeness and dispositional empathy are associated with advantages in identifying positive versus negative affective states or emotions. In Study 1, exploitative individuals were especially good at recognizing negative affective states, whereas empathic individuals were better at recognizing positive affective states. We had no a priori predictions for these results, so Study 1 cannot yet make strong conclusions about the role of valence. Thus, in Study 2 we examine whether similar patterns emerge using different measures of narcissism, dispositional empathy, and emotional competency.
Study 2 addresses other potential problems in Study 1. For example, it is possible that the exploitativeness results were relatively weak because the narcissism scale that we used, the NPI-16, was not intended for subscale analysis. Although we selected three exploitativeness items that completely overlapped with NPI-40 ones, the authors of the NPI-16 conceptualized their scale as being used to calculate an overall score only (Ames et al. 2006). Thus, in Study 2 we use the full five items from the NPI-40 that were intended to address exploitative tendencies. It is also possible that we would find stronger effects when using a more prototypical measure of empathy and a different test of emotional competency that examines the more basic emotions.
Finally, in Study 2 we test our hypotheses in a wider group of American adults, in order to determine whether our results generalize beyond a college student sample.
In Study 2 we attempt to conceptually replicate findings from Study 1 using different measures and a different participant group. We again expected both exploitativeness and dispositional empathy to be associated with better emotion recognition ability.
Participants were 88 Americans recruited from Amazon’s MTurk to complete a survey on “personality and emotion.” MTurk has been used extensively to conduct psychological studies (Buhrmester et al. 2011; Mason and Suri 2012). Participants received a small payment of $0.50, which is the current typical rate of payment for short psychology studies. Participants were 59 % female with a mean age of 37.2 (SD = 12.6). The ethnic composition of the sample was 81 % Caucasian, 8 % Asian-American, 7 % African-American, and 4 % Other.
Procedure and Measures
Participants completed narcissism, dispositional empathy, and mood measures. Narcissism was measured using the 40-item Narcissistic Personality Inventory, which had an overall Cronbach α of .86 (Raskin and Terry 1988). In the NPI-40, five items comprise the exploitativeness subscale (Cronbach α = .55). Three of these items are identical to the exploitativeness subscale of the NPI-16, and there are two additional items. Dispositional empathy was assessed using the 7-item empathic concern subscale of the Interpersonal Reactivity Index (e.g., “I often have tender, concerned feelings for people less fortunate than me”; 1 = not very true of me, 5 = very true of me; Cronbach α = .89; Davis 1983).
Participants next completed an emotion recognition task that assessed more basic emotional expressions rather than the specific affective states in the RMET. They saw 10 facial expressions (anger, contempt, disgust, embarrassment, fear, happiness, pride, sadness, shame, and surprise) posed by two individuals (White man, White woman)—20 photos in all (presented randomly), from the University of California, Davis, Set of Emotion Expressions (Tracy et al. 2009). For each presentation, a fixation cross first appeared on a computer screen (1 s), followed by a face (1 s). Next, participants were asked: “Which emotion, if any, do you think is being expressed by the person in the photo? Choose the emotion that best matches the emotion expressed by the person in the photo.” Response options included: anger, contempt, disgust, embarrassment, excitement, fear, happiness, pride, sadness, shame, surprise, and none of these emotions. The online survey program recorded response times. A total score was computed (0 = none correct to 20 = all correct; Cronbach α = .51). Unfortunately the original USDSEE paper does not list the Cronbach alphas so no comparisons could be made. We address this in the “General Discussion” section.
Finally, since Study 2 was online and brevity was important, we created a shortened 5-item mood measure. Participants rated their positive (happy, interested) and negative mood (irritable, bored, sad) on a 5-point scale (1 = not at all, 5 = extremely; Cronbach α’s = .55 and .61, respectively).
On average, participants correctly recognized 14.2 out of 20 emotions (71.0 %). Of 6 potential positive emotions (happiness, pride, and surprise, times 2 targets) the mean correct response was 5.2 (86.7 %), and of 14 negative emotions (anger, contempt, disgust, embarrassment, fear, sadness, and shame, times 2 targets) the mean correct response was 9.0 (64.3 %). Unlike in Study 1, participants were better at recognizing positive emotions compared to negative emotions, t(88) = 11.07, p < .001.
Predictors of emotion recognition on the University of California, Davis, Set of Emotion Expressions (Study 2)
Exploitativeness factor (NPI-40)
Davis empathic Concern Scale (EC)
Regression Model 1
Total emotion recognition (USDSEE)
Controlling for mood
Regression Model 2
Positive emotion recognition (USDSEE)
Controlling for mood
Regression Model 3
Negative emotion recognition (USDSEE)
Controlling for mood
Among a broader age range of participants, and using a different emotion recognition task, we found that exploitativeness and dispositional empathy were again associated with better emotion recognition abilities. In Study 2, however, controlling for participants’ mood was not influential in observing these results. Adding mood as a covariate slightly dampened the exploitativeness-emotion recognition result and slightly strengthened the empathy-emotion recognition one, yet both remained significant (see Table 3). Taken together, the role of mood as a covariate is inconsistent across the two studies, and future research should continue to include an assessment of participants’ mood when examining similar research questions. Better yet, we recommend that future researchers conduct experimental studies that manipulate narcissism, empathy, and mood in order to fully understand the causal role of each variable in predicting emotion recognition. For example, narcissistic versus empathic states could be experimentally varied via explicit instructions (e.g., asking participants to focus on their own feelings and goals versus others’ feelings and goals during a social interaction) or via implicit subliminally presented information (Riketta and Dauenheimer 2003). Mood could be experimentally varied via established laboratory mood induction procedures (Westermann et al. 1996).
In Study 2, the relationship between exploitativeness and emotion recognition was again similar in magnitude as an established correlate (i.e., dispositional empathy). As in Study 1, exploitative individuals were especially good at recognizing negative emotions, but unlike in Study 1, empathic individuals were also better at recognizing negative emotions. Thus, the pattern with respect to valence is more consistent for exploitativeness than for empathy (see Table 3). Still, in Study 2 we must consider that there was a ceiling effect for positive emotion identification, which might make it less likely that variations in personality traits are associated with variations in performance.
Is narcissism associated with better emotion recognition abilities? Prior research on this topic has been limited, with inconsistent findings. One study found that high narcissism scorers believe that they are better than average at detecting emotions, but this did not bear out in reality (Ames and Kammrath 2004). Another study found that narcissists are good at reading angry expressions, but not happy, sad, or fearful ones (Wai and Tiliopoulos 2012). We add to this literature by finding that only certain aspects of narcissism are associated with better emotion recognition abilities (i.e., exploitativeness).
We also note that the role of participants’ mood must be better understood before making strong conclusions about how mood affects emotion recognition ability in those scoring high in narcissism and empathy. For now, we hesitate to make any claims with regard to interpretations of the mood effects. First, the relationships between mood and the other variables are not entirely consistent across the two studies. In Study 1, results became stronger when mood was added as a covariate, but in Study 2, results were significant without including mood as a covariate. This could have been because different measures of emotion recognition (more affective states in Study 1 and more basic emotions in Study 2) might have responded differently to participants’ moods, but there were other differences across the two studies that make it impossible to make conclusions (e.g., participant population, laboratory versus online study, etc.). Second, these data are correlational, and experimental studies will confirm whether being more narcissistic or empathetic is associated with a general tendency to experience certain moods (e.g., more positive), and whether in turn, being in these moods affects one’s ability or motivation to recognize emotions. Based on the two studies we cannot make any strong conclusions on the role of mood as a covariate, but simply must note that it appears to be important and should be included in future research on this topic.
Taken together, these studies clarify under which circumstances there may be a relationship between narcissism and emotion recognition ability, and as such, they may help to explain inconsistent findings in prior work. Moreover, this work sheds light on the potential darker side of emotional competencies. Simply being skilled at reading others’ emotions could imply a sense of empathy or caring, but it could also translate into more effective manipulation for certain individuals, such as those high in exploitativeness.
When considering our results across both studies, exploitative people are better at recognizing negative (vs. positive) emotions in Study 1 and Study 2, but the role of valence in dispositional empathy is less clear because of the opposite results for Study 1 (positive emotions) and Study 2 (negative emotions). Since the two studies consistently found that exploitativeness was associated with better recognition of negative emotions, we can speculate as to why this might be the case. Exploitative people may be especially attuned to vulnerability in others in order to find people who are easy to take advantage of and prey on, and who may not be in a position to fight back. Regardless of whether the measures are assessing specific affective states (e.g., cautious, doubtful: Study 1) or negative emotions (e.g., fear, sadness: Study 2), negative emotions can often signal vulnerability. For example, an exploitative person may see cautiousness and doubtfulness in others as signals of uncertainty, hesitancy, tentativeness, and low confidence, making individuals who display such facial expressions potentially ripe for manipulation.
Consistent with this analysis, narcissistic exploitativeness is associated with bullying among children (Ang et al. 2009), and bullies often seek out vulnerable individuals to harass (Sweeting and West 2001). In future research, it would be interesting to examine how exploitative individuals respond to other vulnerability cues besides negative emotional expressions (e.g., disheveled appearance, requests for help). Whereas these types of cues may elicit sympathetic responses and helping behavior from many people, exploitative people may see them as perfect opportunities to take advantage.
The pattern for empathy was inconsistent across the two studies, with empathy associated with better positive emotion recognition in Study 1 and better negative emotion recognition in Study 2. We cannot determine whether this was because of the different measures used across the two studies, the different participant populations, or some other reason. We do note that on average Study 1 participants had worse performance (74.3 %) on the positive emotions task compared to Study 2 (86.7 %), and Study 2 participants had worse performance (64.3 %) on the negative emotions task compared to Study 1 (74.0 %). Thus, it is possible that empathic individuals excel at recognizing emotions that others generally find more difficult to identify. However, because of the inconsistent pattern across studies, we hesitate to make conclusions about the relationship between empathy and valence, and instead note that on average, in both studies, empathic people scored higher on the total number of emotions recognized.
Caution is warranted because of the correlational nature of both studies. Good emotion recognition ability could lead people to develop bad motives, because they can see weak points in others more easily than many people. Conversely, it is also possible that having an exploitative personality can lead to increases in emotion recognition ability. This could occur through a direct deliberative process such that exploitative people might be aware of the value of emotion recognition and deliberately develop it. But this could also occur through an experiential process such that the repeated successes and failures in efforts at exploitation could build better emotion recognition abilities. It is also possible that both processes work hand-in-hand. Perhaps the most likely process is one of moderation—people with good emotion recognition skills may use it for exploitation or for more empathic purposes, depending on their goals. Future research needs to clarify the direction of causality by examining whether experimentally increasing exploitative goals can affect emotion recognition abilities, or whether training people to better recognize emotions affects individuals’ exploitative tendencies. Yet researchers should be cautious and also consider the ethics and desirability of possibly creating more exploitativeness in experimental studies.
We also note that the Cronbach alphas of all of our key measures (exploitativeness, emotion recognition) were low. However, for the exploitativeness factor, the alpha was comparable to that found in the original NPI development paper, and unfortunately the emotion recognition development papers do not list the Cronbach alphas so no comparisons could be made. Despite this, (a) these measures are all widely used and standardized, (b) our predicted effects were replicated across different procedures and populations, and (c) the emotion recognition measures were sensitive to measures of dispositional empathy, which offers an additional guarantee that these measures, although imperfect, do represent the construct of interest. Still, we recommend that future researchers develop more internally reliable instruments. We used the best instruments available.
The present research operationalized emotional competencies as emotion recognition, but there are many other types (e.g., emotion regulation), which future research should explore. Moreover, we did not specifically test for the motives or goals that exploitative people held while identifying emotions, which would be another potentially interesting future extension. We find that emotional competencies, like other abilities, are associated with both desirable and undesirable correlates. The extent to which exploitative people use emotional competencies for self-gain is an important question for future research.
The current studies find that exploitative people can indeed “read people like a book.” Taken together, they add to the emerging literature examining complex relationships between emotional competencies and darker psychological traits (Côté et al. 2011). We conclude by suggesting that people should not assume that someone who can easily read their feelings always has their best interests at heart.
For a priori theoretical reasons outlined in the Introduction we chose to specifically focus on the exploitativeness subscale of the Narcissistic Personality Inventory. However, readers may be curious whether the other aspects of narcissism also predict emotion recognition ability. We have no specific a priori hypotheses with respect to why other aspects of narcissism should be associated with better emotion recognition (i.e., exhibitionism, entitlement, authority, superiority, self-sufficiency, or vanity). When examining the specific subscales of the NPI in each of the two studies, there are no consistent patterns with any other subscales predicting emotion recognition across both studies. In addition, when combining these subscales into a “non-exploitative narcissism” scale by summing them together, there are no consistent relationships with emotion recognition across both studies. Thus, the non-exploitative aspects of narcissism were not consistently associated with emotion recognition, even when controlling for mood.
The first author was supported by a grant from Wake Forest University, The Psychology of Character (via the John Templeton Foundation), while writing this manuscript. We thank Judith Hall and two anonymous reviewers for helpful comments on this manuscript.
Conflict of interest
The authors declare that they have no conflict of interest.