Personality disorders probably represent the extreme manifestation of traits that are normally distributed in the population. They have genetic and environmental components and may be adaptive or maladaptive depending on the circumstances (Del Giudice, 2018; Hunt and Jaeggi, 2022). The DSM-5 description of Narcissistic Personality Disorder (NPD) emphasizes a chronic pattern of dysfunction in relationships and trait-like personality features of arrogance, grandiosity, and lack of empathy (Gore and Widiger, 2016). The DSM-5 contains both a categorical (NPD) and dimensional (narcissistic traits) representation of narcissism. Non-pathological narcissistic traits are defined as “a relatively stable individual difference consisting of grandiosity, self-love and inflated self-views” (Campbell et al., 2011). In both NPD and non-pathological narcissism, self-identity and self-image are thought to depend on constant external validation from other individuals (e.g., their praise and adulation) or on continuous positive feedback through accomplishment and success (Campbell et al., 2006; Jones and Paulhus, 2011; Pincus, 2013; Roche et al., 2013). In this view, the outward confidence shown by subjects with strong narcissistic traits may be a self-regulation strategy to cope with low self-esteem (Bosson et al., 2003). The role of self-esteem in narcissism is a contentious issue in the literature, as narcissistic traits such as grandiosity and assertiveness may be displayed to conceal deep-seated feelings of inferiority, as claimed by the mask model (Miller et al., 2017); reports on the association between self-esteem and narcissistic traits range from negative to positive, complicating a possible association between self-esteem and narcissism subtypes (see Miller et al., 2017 for a review). There is evidence suggesting that narcissism may also be associated with schizotypy, psychopathy (March and Springer, 2019), and autistic-like traits (Cheek and Norem, 2022), as well as non-pathological personality traits such as extraversion, neuroticism, and agreeableness (Campbell and Crist, 2020; Zajenkowski and Szymaniak, 2021).

Acknowledging the divergent and sometimes contradictory findings obtained with individuals with NPD or non-pathological narcissistic traits, the field recognizes at least two subtypes of narcissism: vulnerable and grandiose (Wink, 1991; Jauk et al., 2017; Coleman et al., 2019; Jauk and Kanske, 2021). Individuals who express traits associated with vulnerable narcissism are more likely to be introverted, usually exhibit an insecure/fearful attachment style, and tend to show traits associated with negative emotionality such as anxiety, depression, and hostility, whereas individuals who express traits associated with grandiose narcissism are more likely to be extraverted and be at least outwardly dismissive of attachment relationships (Dickinson and Pincus, 2003; Miller et al., 2010; Campbell and Miller, 2013; Pincus et al., 2015; Miller et al., 2017). Negative associations between grandiose narcissism and some indicators of negative emotionality such as depression and sadness have also been reported (cf. Czarna et al., 2018). In addition to cognitive differences related to self-appraisal and interpretation of other individuals and their behavior (Pincus et al., 2009; Zajenkowski et al., 2018), differences in emotion regulation have been hypothesized to separate individuals with vulnerable and grandiose narcissistic traits, especially when their ego is threatened. For example, subjects who express traits associated with vulnerable narcissism appear to be more prone to anxiety, stress, and fear, whereas subjects who express traits associated with grandiose narcissism are more prone to anger and are more resilient to stress (e.g., Besser and Priel, 2010; Roche et al., 2013; Ng et al., 2014; Fernie et al., 2016; Sękowski et al., 2021; Underwood et al., 2021). As a notable example, Besser and Priel (2010) compared emotional reactions to two threat conditions (i.e., achievement failure and interpersonal rejection) between individuals high in grandiose narcissism and individuals high in vulnerable narcissism. The authors found that under a high-level interpersonal threat, but not a high-level achievement-threat, high scores of vulnerable narcissism were associated with greater change in negative outcomes, whereas under a high-level achievement-threat, but not a high-level interpersonal threat, high scores of grandiose narcissism predicted greater change in negative outcomes, supporting the claim of differential emotional responses under different contexts between subjects high in vulnerable and grandiose narcissism scores.

Given the possibility that outward behavior may not be a reliable index of underlying psychobiological processes, an important question regarding the validity of the two subtypes of narcissism is whether they are separable biologically. There is some evidence indicating possible psychophysiological differences between individuals with vulnerable and grandiose narcissistic traits. For example, Kelsey et al. (2002) found that women with characteristics of grandiose narcissism have a heightened cardiac pre-rejection period (i.e., the interval between myocardial contractile force and aortic opening reflecting sympathetic control on the hearth), whereas women with characteristics of vulnerable narcissism have a diminished skin conductance response after two stress tasks involving mental arithmetic and the Thematic Apperception Test. Given the hypothesized self-regulatory role of narcissistic fluctuations of self-esteem in the face of psychosocial stress, the functioning of the hypothalamic–pituitary–adrenal (HPA) axis may also reflect such differences. Some data suggest that narcissism is a façade, masking interpersonal hypersensitivity and stress reactivity, similar to that observed in borderline personality disorder (Gunderson and Lyons-Ruth, 2008). Consistent with this hypothesis, in a study of 129 men, baseline levels of salivary cortisol were found to be associated with grandiose narcissism (as a component of the Dark Triad of personality, cf. Jones and Paulhus, 2011), whereas no association was found with psychopathy or Machiavellianism (Pfattheicher, 2016). In another study, Reinhard et al. (2012) reported that ‘unhealthy narcissism’ (high grandiose narcissism scores), predicted higher basal salivary cortisol in men while a tendency was found for women. However, one of the largest studies so far (n = 366) failed to find any relationship between narcissism and baseline cortisol levels (Wardecker et al., 2018).

It is worth emphasizing that the cortisol response to a stressful situation was not assessed in any of these previous studies. However, in three studies that examined the relationship between grandiose narcissism (but not vulnerable narcissism) and HPA axis reactivity to psychosocial stress, the results showed that (a) cortisol response following a public speaking task was elevated in individuals with grandiose narcissism (Edelstein et al., 2010), (b) that grandiose narcissism, as measured by the Narcissistic Personality Inventory (NPI), was positively correlated with increased cortisol levels in association with perceptions of negative emotions across 3 days (Cheng et al., 2013), and (c) that, contrary to these two previous studies, grandiose narcissism was associated with decreased cortisol release after the social stress of telling a lie (Dane et al., 2018). Given that there may be further unpublished non-significant results concerning the association between narcissism and HPA axis function and that studies on vulnerable narcissism and HPA axis function are lacking, this research is difficult to interpret. Methodological differences among the studies could also account for inconsistencies in previous findings. Thus, the hypothesis of divergent mechanisms of psychobiological stress-coping in grandiose and vulnerable narcissism remains open and is just beginning to be empirically explored.

To the best of our knowledge, no previous study has investigated possible differences in responses to the TSST between subjects with vulnerable and grandiose narcissism within the same study. The aims of this exploratory study were therefore (1) to compare emotional and cortisol responses to the TSST between individuals with vulnerable and grandiose narcissistic traits and (2) to examine possible associations between narcissism, other personality traits, and stress responses. For the first aim, we predicted that individuals with vulnerable narcissistic traits would show stronger emotional and physiological responses to the TSST than subjects with grandiose narcissistic traits. For the second aim, we predicted differences in personality correlations between individuals high in grandiose narcissism and individuals high in vulnerable narcissism.

Consistent with our hypothesis of divergent mechanisms of psychobiological stress-coping in grandiose and vulnerable narcissism, the present exploratory study found a strong increase in cortisol levels after the TSST in subjects high in vulnerable narcissism, whereas a blunted response was observed in subjects high in grandiose narcissism. This significant difference in cortisol responses between subjects with vulnerable and grandiose narcissistic traits was consistent with our first prediction and emerged from the data with medium-to-large effect sizes despite the small sample size. One of the strengths of the present study, is that our results show that the observed difference in cortisol reactivity was not a byproduct of differences in extraversion/introversion, because the difference remained statistically significant even after controlling for this variable in the analysis. In support of our second prediction, we found that subjects with vulnerable narcissistic traits showed higher scores on the “social anxiety” and “interpersonal” dimensions of schizotypy, and reported increases in negative emotions such as anxiety, anger, shame behavior, and devaluation after the TSST, whereas subjects with grandiose narcissistic traits showed higher scores on the “fearlessness” and “social influence” dimensions of psychopathy and reported an increase in devaluation scores after the TSST, but not in anxiety. These findings provide additional validation for the two subtypes of narcissism. Additionally, our results also point to the occurrence of a third subtype of narcissism, the mixed-type, which is characterized by joint traits of both vulnerable and grandiose narcissism (see also Campbell and Crist, 2020). This conclusion is supported by our findings showing that the mixed-type narcissism can be physiologically and psychologically differentiated from the other two subtypes in some traits. For example, subjects in the mixed-type group scored higher in the schizotypal traits of “social anxiety” and “eccentric behavior” and lower in the psychopathic trait “stress immunity” than subjects high in grandiose narcissism, and that the mixed-type tended to have a smaller cortisol response than subjects high in vulnerable narcissism. Additionally, the medium-to-large effect sizes found in non-significant results may be indicative of further psychological and personality differences between the narcissism subtypes.

One of the main characteristics of subjects high in vulnerable narcissism is a high sensitivity to social criticism and concomitant fear to be exposed (Twenge and Campbell, 2003; Miller et al., 2011). Subjects high in vulnerable narcissism may show stronger cortisol responses to the TSST because their unstable self-image is generally associated with anxiety and fear of challenges. The observed high scores of social anxiety and constricted affect as well as the increases in negative emotions within this subtype of narcissism are consistent with this explanation. In contrast, inflated ego, exhibitionism, and vanity are common characteristics of subjects high in grandiose narcissism (Jones and Paulhus, 2011). Subjects with grandiose narcissistic traits may not react with anxiety and fear to social evaluation because, similar to psychopaths, they are relatively fearless in socially challenging situations. Thus, it is possible that the performance in front of an audience that is part of the TSST procedure is not interpreted by them as anxiety-provoking or stressful, but it may even reinforce their vanity and competitiveness (see Wallace and Baumeister, 2002). The observed higher scores of social influence, fearlessness, and stress immunity, as well as the lack of increases in the vast majority of negative emotions in the subjects high in grandiose narcissism, may have played a role in their observed blunted cortisol response to the TSST. Consistent with these explanations, we found a weak cortisol response to the TSST in the group of subjects with a mixed-type of narcissism, who shared some psychological traits (e.g., in terms of schizotypy, psychopathy, and negative emotions) with subjects within the vulnerable and the grandiose subtypes of narcissism (see Campbell and Crist, 2020, for a discussion of subjects who show both grandiose and vulnerable narcissistic traits).

Previous research has reported that subjects high in grandiose narcissism tend to show positive affect after stress or following provocation (Wolven, 2015; Hart et al., 2021), are more resilient and have higher adaptive capacity (Sękowski et al., 2021), and are psychologically healthier and report lower perceived stress and higher life satisfaction (Ng et al., 2014) than subjects high in vulnerable narcissism. However, previous research on the association between cortisol reactivity to stress and narcissism has not accounted for narcissism subtypes.

Our findings may help to explain why the results of some previous studies are inconsistent. For example, two studies found that individuals with grandiose narcissism exhibited elevated cortisol response after the TSST (Edelstein et al., 2010; Cheng et al., 2013) while a third found that cortisol levels decreased after a social evaluative test (lying while being videotaped; Dane et al., 2018). Our results suggest that future studies examining the role of stress-reactivity in NPD and non-pathological narcissism, should take the vulnerable subtype, and even the mixed-type into consideration. The results could be seen as supporting some previous reports of elevated interpersonal hypersensitivity in NPD, such as the finding of increased oxidative stress in NPD and borderline personality disorder, but not other personality disorders (Lee et al., 2020). On the other hand, the results may also point to clear biological boundaries between vulnerable and grandiose narcissism, and potentially less clear boundaries in the mixed-type, with grandiose narcissism lacking interpersonal hypersensitivity with blunted cortisol responses under socially stressful situations. Future studies of narcissism should continue to investigate possible differences in the (physiological, and perhaps also emotional-cognitive) mechanisms underlying grandiose, vulnerable, and the mixed-type narcissism as well as possible differences in their adaptive significance: for example, the extent to which they are accompanied by similar or different sociosexual strategies (e.g., Jonason et al., 2009) that modulate mate value and number of sexual partners (Borráz-León and Rantala, 2021; Burtăverde et al., 2021) in subjects with these narcissism subtypes.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Social Science Institutional Review Board at the University of Chicago. The patients/participants provided their written informed consent to participate in this study.

JIB-L: formal analysis, investigation, writing—original draft, and writing—review and editing. AS: formal analysis and writing—review and editing. CS and ML: conceptualization, investigation, methodology, and writing—review and editing. RL: writing—original draft and writing—review and editing. DM: conceptualization, investigation, resources, writing—original draft, and writing—review and editing. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.