1.1 Grandiose narcissism

When hearing the term “narcissism”, most people think of exaggerated self-worth, feelings of superiority, admiration seeking, entitlement, and arrogance (Buss & Chiodo, Reference Buss and Chiodo1991). These common associations actually very closely match the definition of grandiose narcissism as a personality trait, which encompasses self-importance and entitlement – the antagonistic core of narcissism – alongside extraverted and socially dominant behavior (Krizan & Herlache, Reference Krizan and Herlache2018; Miller et al., Reference Miller, Lynam, McCain, Few, Crego, Widiger and Campbell2016).

The concept of narcissism as a trait was originally devised from the diagnostic criteria for Narcissistic Personality Disorder (NPD) in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1980), and adapted for the assessment of narcissism in the general, nonclinical population in the Narcissistic Personality Inventory (NPI; Raskin & Hall, Reference Raskin and Hall1979, Reference Raskin and Hall1981). The original criteria included (1) a grandiose sense of self-importance, (2) fantasies of unlimited success, power, etc., (3) exhibitionism, (4) responding to criticism with either indifference or rage, (5) entitlement, (6) exploitativeness, (7) relationships vacillating between idealization and devaluation, and (8) lack of empathy. Importantly, these criteria focus strongly on narcissistic grandiosity, and only implicitly tap into narcissistic vulnerability (see section 1.2). Items of the NPI encompass statements such as “I think I am a special person”, “I am more capable than other people”, or “I have a natural talent for influencing people” (Raskin & Terry, Reference Raskin and Terry1988). The NPI has become the most widely used measure of grandiose narcissism, the total score displays good reliability and validity with respect to expert ratings (Miller et al., Reference Miller, McCain, Lynam, Few, Gentile, MacKillop and Campbell2014), and the vast majority of the neuroscience studies on grandiose narcissism reviewed here used the NPI as an indicator.

1.2 Vulnerable narcissism

Besides grandiose narcissism, increasing attention has been paid to vulnerable narcissism in the past decades (e.g., Fossati et al., Reference Fossati, Borroni, Grazioli, Dornetti, Marcassoli, Maffei and Cheek2009). While vulnerable aspects of narcissism have long been hypothesized to be overt or covert parts of the broader phenomenon of narcissism (Pincus & Lukowitsky, Reference Pincus and Lukowitsky2010), the social/personality study of narcissism delineated vulnerable narcissism as an independent trait – the “second face” of narcissism (Wink, Reference Wink1991). Vulnerable narcissism is uncorrelated to grandiose narcissism at a trait level in the general population, but might blend into grandiosity at high levels of grandiose narcissism (Jauk & Kaufman, Reference Jauk and Kaufman2018; Jauk, Weigle, et al., Reference Jauk, Weigle, Lehmann, Benedek and Neubauer2017). To date, only two neuroscience studies explicitly addressed vulnerable narcissism in that they used independent measures to assess this construct. Nevertheless, the construct may also be relevant to the interpretation of studies on grandiose and pathological narcissism as well as NPD.

Individuals high in vulnerable narcissism typically present very differently from those high in grandiose narcissism: vulnerable narcissistic individuals are anxious, defensive, and avoidant (Hart et al., Reference Hart, Adams, Burton and Tortoriello2017; Miller et al., Reference Miller, Price, Gentile, Lynam and Campbell2012). Yet, despite the outward presentation of self-consciousness, those high in vulnerable narcissism also share the common antagonistic core of entitlement and self-importance (Krizan & Herlache, Reference Krizan and Herlache2018; Miller et al., Reference Miller, Lynam, McCain, Few, Crego, Widiger and Campbell2016). A commonly used measure of vulnerable narcissism, the Hypersensitive Narcissism Scale (HSNS; Hendin & Cheek, Reference Hendin and Cheek1997), encompasses items such as “I easily become wrapped up in my own interests and forget the existence of others” or “I am secretely [sic!] ‘put out’ when other people come to me with their troubles, asking me for my time and sympathy“ (p. 592).

1.3 Pathological narcissism

Pathological narcissism, as an operationally defined and quantifiable construct, was introduced by Pincus and colleagues in recognition of the problem that previously existing self-report measures of (grandiose) narcissism mostly focused on adaptive qualities, and maladaptive as well as vulnerable aspects might have been underrepresented (Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009). Consequently, pathological narcissism is defined as comprising both grandiose and vulnerable aspects (Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009; Pincus & Lukowitsky, Reference Pincus and Lukowitsky2010). While these seemingly opposing experiential and behavioral tendencies may appear hard to reconcile at the first glance, Pincus and Lukowitsky (Reference Pincus and Lukowitsky2010) assert that grandiose and vulnerable self-states can fluctuate or co-occur in highly narcissistic individuals (Wright & Edershile, Reference Wright and Edershile2018), which is illustrated vividly in clinical case descriptions (Pincus et al., Reference Pincus, Cain and Wright2014). The current standard measure of pathological narcissism is the Pathological Narcissism Inventory (PNI; Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009). It encompasses facets related more to narcissistic grandiosity, namely entitlement rage (anger when entitled expectations are not met), exploitativeness (manipulative behavior in pursuit of own goals), grandiose fantasy (imagining grandiose self), and self-sacrificing self-enhancement (self-enhancement by means of seemingly altruistic behavior), as well as facets related more to narcissistic vulnerability, which are contingent self-esteem (dependence upon others’ regard), hiding the self (not showing own needs and weaknesses), and devaluing (devaluing others who do not meet their own needs). The grandiose and vulnerable factors are intrinsically related in the PNI (Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009). This makes it conceptually and empirically different from scales designed exclusively for the assessment of either grandiose or vulnerable narcissism, which are unrelated in the general population (Jauk, Weigle, et al., Reference Jauk, Weigle, Lehmann, Benedek and Neubauer2017).

1.4 Narcissistic personality disorder

The diagnostic category of NPD emerged in the DSM-III mainly from psychodynamic theory and case studies (Ronningstam, Reference Ronningstam2009). Since its first inclusion, the diagnostic criteria (mentioned above) have been revised. In the latest edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013), the categorical diagnosis of NPD is defined by at least five out of the following nine criteria: (1) grandiose sense of self-importance, (2) fantasies of unlimited success, power, etc., (3) thinks he/she is special and unique, (4) requires excessive admiration, (5) entitlement, (6) exploitativeness, (7) lack of empathy, (8) envious/thinks others are envious, and (9) arrogance. Contrary to the developments in the research field, the definition of NPD in the DSM-5 is strongly focused on grandiosity (Pincus & Lukowitsky, Reference Pincus and Lukowitsky2010; Ronningstam, Reference Ronningstam2009) and might – in its present form – best be understood as an extreme manifestation of grandiose narcissism, which is supported by concurrent validity evidence (Miller et al., Reference Miller, McCain, Lynam, Few, Gentile, MacKillop and Campbell2014). This does not mean that vulnerable aspect cannot accompany grandiosity (which is also explicitly acknowledged in the DSM), but they do not constitute a necessary condition for an NPD diagnosis. For the purpose of the present review, it is most important to note that the terms pathological narcissism and NPD, despite their similar connotation, cannot be used interchangeably. While pathological narcissism, sensu Pincus and Lukowitsky (Reference Pincus and Lukowitsky2010), refers to concurrent grandiosity and vulnerability, NPD is a clinical diagnosis for individuals with extreme levels of grandiosity, which may or may not be accompanied by vulnerability.

The prevalence of NPD is generally considered low, as compared to other mental disorders. Prevalence estimates in the general population range between 0 and 1% (Roepke & Vater, Reference Roepke and Vater2014) or 0–5% (Ronningstam, Reference Ronningstam2009), the DSM-5 provides an estimated range between 0 and 6.2% in community samples (American Psychiatric Association, 2013). Systematic empirical research on NPD is generally very sparse, putatively because patients are hard to find, and hard to recruit. We summarize some of the available evidence in the following.

1.5 Summary: grandiosity and vulnerability in narcissism

According to current models, narcissism entails both grandiose and vulnerable aspects, which manifest in experiential and behavioral patterns of self-assuredness and dominance on the one hand, and insecurity and reactivity on the other (see Figure 1; Krizan & Herlache, Reference Krizan and Herlache2018; Weiss et al., Reference Weiss, Campbell, Lynam and Miller2019). In the general population, self-reports of trait grandiosity and vulnerability are unrelated or only weakly positively related, which points to relatively independent personality configurations – sharing a common core of self-importance and entitlement/antagonism (Krizan & Herlache, Reference Krizan and Herlache2018; Weiss et al., Reference Weiss, Campbell, Lynam and Miller2019) – at a global level (Jauk & Kaufman, Reference Jauk and Kaufman2018; Jauk, Weigle, et al., Reference Jauk, Weigle, Lehmann, Benedek and Neubauer2017). However, it has been posited that those high in grandiosity can fluctuate between grandiose and vulnerable states (e.g., Wright & Edershile, Reference Wright and Edershile2018), which is evident in informant reports (Gore & Widiger, Reference Gore and Widiger2016), increases in correlations among trait measures of grandiosity and vulnerability with increasing grandiosity (Jauk & Kaufman, Reference Jauk and Kaufman2018; Jauk, Weigle, et al., Reference Jauk, Weigle, Lehmann, Benedek and Neubauer2017), and ecological momentary assessment of state grandiosity and vulnerability (Edershile & Wright, Reference Edershile and Wright2020). Grandiosity and vulnerability are further intrinsically related in measures assessing primarily maladaptive, pathological aspects of narcissism (such as the PNI; Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009), and vulnerable self-states can be observed in NPD (American Psychiatric Association, 2013; Ronningstam, Reference Ronningstam2009), a personality disorder characterized by extreme grandiosity (Miller et al., Reference Miller, McCain, Lynam, Few, Gentile, MacKillop and Campbell2014). Taken together, while grandiose and vulnerable narcissism can be described in terms of dissociable personality configurations at a trait level, there is also increasing evidence that these two “faces” of narcissism go hand in hand – particularly at high levels of grandiosity. An interesting question for the neuroscience of intrapersonal functioning in narcissism might thus be to identify conditions under which individuals high in grandiose narcissism display reactions which are indicative of vulnerability, even if not directly evident in self-reports. Another question concerns interpersonal functioning, which is altered in all expressions in narcissism discussed above. We elaborate on these questions in the following section 2.1.

2.1 Current research questions for the neuroscience of narcissism

The overarching goal of this review is to highlight the possible contributions of neuroscience to the understanding of narcissism in its different expressions. Though self-report and behavioral research on narcissism have acquired a great deal of knowledge in the past decades, some questions still remain open. Broadly, these concern two fundamental aspects of personality functioning: (1) intrapersonal, self-related characteristics of narcissism and (2) interpersonal, other related characteristics of narcissism.

(1) It has long been hypothesized that narcissism, particularly grandiose narcissism and NPD as its clinical expression, entail vulnerable aspects, which need not necessarily be overtly expressed (e.g., Pincus & Lukowitsky, Reference Pincus and Lukowitsky2010). The mask model of narcissism posits that grandiosity is a façade to mask an underlying fragile self (Akhtar, Reference Akhtar1989); that is, that grandiosity compensates for underlying vulnerability. Whether or not grandiosity is indeed causally related to underlying vulnerability, as this model implies, clinicians working with narcissistic patients are inclined to agree with their coexistence (in correlational terms). In their seminal review, Pincus and Lukowitksy (Reference Pincus and Lukowitsky2010) assert that “many contemporary clinical experts on narcissistic personality disorder now recognize that grandiose self-states oscillate or co-occur with vulnerable self-states and affective dysregulation” (p. 428). Similarly, in her review on NPD, Ronningstam (Reference Ronningstam2009) states that: “the narcissistic individual may fluctuate between assertive grandiosity and vulnerability” (p. 113). Following this perspective, one might assume that individuals high in grandiose narcissism should hold implicit negative self-views along their explicit positive self-views (cf. Kuchynka & Bosson, Reference Kuchynka, Bosson, Hermann, Brunell and Foster2018). However, this view is not commonly supported by systematic research. Self-report studies on grandiose narcissism find positive correlations with explicit (Campbell et al., Reference Campbell, Rudich and Sedikides2002) and also implicit self-esteem (Campbell et al., Reference Campbell, Bosson, Goheen, Lakey and Kernis2007), and research on NPD suggests no difference in implicit self-esteem between NPD patients and controls (Marissen et al., Reference Marissen, Brouwer, Hiemstra, Deen and Franken2016; Vater et al., Reference Vater, Ritter, Schröder-Abé, Schütz, Lammers, Bosson and Roepke2013).

The question remains, thus, if and how negative self-views are represented in individuals high in grandiose narcissism, and neuroscience might help to shed light on the respective mechanisms. In the long run, this might also contribute to understanding state changes between grandiosity and vulnerability discussed above (e.g., Edershile & Wright, Reference Edershile and Wright2020). Neuroscience studies targeting intrapersonal functions can investigate the neurophysiological reactions to situations in which the individual is exposed to ego threat, for instance, in terms of achievement failure, social exclusion, or confrontation with self-referential stimuli. These might induce stronger involuntary stress responses in individuals with higher levels of narcissism (cf. Coleman et al., Reference Coleman, Pincus and Smyth2019). Indicators that are known to be sensitive to these responses include, among others, autonomic measures such as skin conductance (Jacobs et al., Reference Jacobs, Friedman, Parker, Tofler, Jimenez, Muller and Stone1994) or blood pressure (Ulrich-Lai & Herman, Reference Ulrich-Lai and Herman2009) as well as neuroendocrine markers such as cortisol level (Kirschbaum et al., Reference Kirschbaum, Pirke and Hellhammer1993). Using neuroimaging methods, activity within regions of the salience network (comprising the anterior insula [AI] and dorsal anterior cingulate cortex [dACC]; Bressler & Menon, Reference Bressler and Menon2010) has been associated with the processing of aversive stimuli that induce conflict, pain, or negative affect (de la Vega et al., Reference de la Vega, Chang, Banich, Wager and Yarkoni2016; Shackman et al., Reference Shackman, Salomons, Slagter, Fox, Winter and Davidson2011; Somerville et al., Reference Somerville, Heatherton and Kelley2006).Footnote ² These indicators might thus unveil involuntary vulnerable reactions and are less prone to cognitive bias than pure self-report measures. Moreover, habitual experiential and behavioral tendencies related to narcissism might manifest in chronic functional alterations in the mentioned systems, which can be revealed by studies of baseline function.

(2) Another controversial aspect of narcissism concerns the exact nature of altered interpersonal functioning observed in different expressions of narcissism. Interpersonal functioning includes – alongside more complex constructs such as attachment (see Table 1) – fundamental social–affective and social–cognitive functions frequently termed empathy and perspective-taking. Empathy refers to sharing others’ affective and emotional states, whereas perspective-taking (or Theory of Mind) circumscribes the capacity to cognitively represent others’ mental states (de Vignemont & Singer, Reference de Vignemont and Singer2006; Frith & Frith, Reference Frith and Frith2005). While both are sometimes subsumed under empathy as an umbrella term (being also referred to as affective and cognitive empathy; e.g., Baron-Cohen & Wheelwright, Reference Baron-Cohen and Wheelwright2004), they can be selectively impaired and draw on distinct neural networks (Stietz et al., Reference Stietz, Jauk, Krach and Kanske2019; see also below). Importantly, empathy and perspective-taking display considerable state variation and can depend on contextual and also motivational factors (ibid.), raising the question of whether altered interpersonal functioning in narcissism is more a question of reduced capacity (lower ability) or propensity (lower engagement; cf. Baskin-Sommers et al., Reference Baskin-Sommers, Krusemark and Ronningstam2014).

Narcissistic individuals are known to display lowered empathy (indeed, limited empathy is among the defining criteria of grandiose narcissism; see section 1.1), and there is also evidence for impaired social cognition in NPD (Marissen et al., Reference Marissen, Deen and Franken2012). However, behavioral evidence suggests that limited empathy in narcissism might be more due to lower propensity than capacity (Hepper et al., Reference Hepper, Hart and Sedikides2014), and similar mechanisms are being discussed for putative perspective-taking deficits in NPD (Baskin-Sommers et al., Reference Baskin-Sommers, Krusemark and Ronningstam2014). It might thus be the case that altered interpersonal functioning in narcissistic individuals is more related to the consequences of a self-focused and antagonistic interpersonal style (Krizan & Herlache, Reference Krizan and Herlache2018; Miller et al., Reference Miller, Lynam, McCain, Few, Crego, Widiger and Campbell2016) than ability deficits per se. Again, neuroscience might help to gain a closer understanding of the mechanisms involved in interpersonal functioning in narcissism. A neuroscience approach seems particularly fruitful given that global self-reports of empathy and perspective-taking (as measured by the frequently used Interpersonal Reactivity Index, for instance; Davis, Reference Davis1983) display only low-to-moderate correlation with task-based ratings and corresponding neural activation (Hildebrandt et al., Reference Hildebrandt, Jauk, Lehmann, Maliske and Kanske2021), which points to differences between global self-perceptions of typical behavior and interpersonal capacities assessed under laboratory conditions. This gap might be even stronger for individuals high in narcissism, whose self-ratings of emotional competencies, in general, deviate substantially from task-based assessments (Jauk, Freudenthaler, et al., Reference Jauk, Freudenthaler and Neubauer2016; Mota et al., Reference Mota, Leckelt, Geukes, Nestler, Humberg, Schröder-Abé, Schmukle and Back2019).

Neuroscience research on interpersonal functioning has delineated empathy from perspective-taking; the two main capacities that enable access to others’ inner, mental states, thus, providing crucial ground for successful social interaction. Empathic reactions to others’ emotional states are commonly observed in the same regions that are activated during the firsthand experiences of these states (Kanske, Reference Kanske2018). Thus, for instance, empathy for another’s aversive experience, be it pain, disgust or unfair treatment, activates the AI and the (middle) anterior cingulate cortex (parts of the salience network; see above), which are also activated during firsthand aversive stimulation (Corradi-Dell’Acqua et al., Reference Corradi-Dell’Acqua, Tusche, Vuilleumier and Singer2016; Oosterwijk et al., Reference Oosterwijk, Snoek, Rotteveel, Barrett and Scholte2017). This supports the view that empathy for others’ emotional states might be regarded as an internalization of these states (Kanske, Reference Kanske2018), or in other words, we literally feel what others are feeling when empathizing. With respect to narcissism, this might make alexithymia, in terms of the general incapacity to experience emotions, a prime candidate as a mediating factor (Valdespino et al., Reference Valdespino, Antezana, Ghane and Richey2017).

Perspective-taking, or Theory of Mind, draws on a different neuronal network than empathy (Kanske, Böckler, Trautwein, & Singer, Reference Kanske, Böckler, Trautwein and Singer2015). This network comprises medial prefrontal regions, the temporoparietal junction, the precuneus/posterior cingulate cortex, as well as the temporal poles and superior temporal sulcus (Bzdok et al., Reference Bzdok, Schilbach, Vogeley, Schneider, Laird, Langner and Eickhoff2012; Molenberghs et al., Reference Molenberghs, Johnson, Henry and Mattingley2016; Schurz et al., Reference Schurz, Radua, Tholen, Maliske, Margulies, Mars and Kanske2021); structures that partially overlap with the default mode network (Mars et al., Reference Mars, Neubert, Noonan, Sallet, Toni and Rushworth2012). Activation within the network implicated in perspective-taking directly corresponds to task performance (Kanske et al., Reference Kanske, Böckler, Trautwein and Singer2015). Most importantly for the present review, neurophysiological research showed that empathy and perspective-taking are clearly independent and distinguishable processes, both on a behavioral and neurophysiological basis (Kanske, Böckler, Trautwein, Parianen Lesemann, & Singer, Reference Kanske, Böckler, Trautwein, Parianen Lesemann and Singer2016). Intra- and interindividual variations in the two capacities are independent as well (Stietz et al., Reference Stietz, Jauk, Krach and Kanske2019). This means that empathic functioning and perspective-taking could be selectively altered in different expressions of narcissism. Neuroscience research might unveil alterations in the involved networks during empathy and perspective-taking tasks, during rest, and in structural measures.

2.2 Literature search and inclusion strategy

For this review, we considered quantitative empirical journal articles on grandiose and vulnerable narcissism, pathological narcissism, and NPD using neuroimaging, neurophysiological, or neuroendocrine measures. For the inclusion, we required a standardized assessment of narcissism that targets one of the constructs introduced above. We searched the academic databases Scopus and PubMed, complemented by Google Scholar searches, for the following terms in the article abstracts: narciss* + neuro*, bio*, physio*, EEG, *MRI, and *imaging. After the exclusion of 3 empirical articles using qualitative methods and 4 studies that did not use neuroscience methods, we obtained a set of 35 empirical journal articles. Of those journal articles, one (Yang, Sedikides, Gu, Luo, Wang, Yang, et al., Reference Yang, Sedikides, Gu, Luo, Wang, Yang and Cai2018) targeted the relatively novel construct of communal narcissism (an expression of agentic, narcissistic strivings through communal means, such as being the most helpful person; Gebauer, Sedikides, Verplanken, & Maio, Reference Gebauer, Sedikides, Verplanken and Maio2012). As this study is to date the only neuroscience investigation of the communal narcissism construct, and the construct itself is still relatively new,Footnote ³ we include this study in the discussion section, rather than the main part of the review. Another study (Kelsey et al., Reference Kelsey, Ornduff, Reiff and Arthur2002) used subscales of a multidimensional personality scale, the Bell Object Relations and Reality Testing Inventory (BORRTI; Bell, Reference Bell1995) for the assessment of overt and covert narcissism (which were sometimes equated with grandiose and vulnerable narcissism). As it was not possible to evaluate the convergent validity of the used subscales (egocentricity and alienation) with more widely used measures of grandiose and vulnerable narcissism, and also the authors themselves discuss the validity of the measure critically (Kelsey et al., Reference Kelsey, Ornduff, Reiff and Arthur2002), we excluded this study from our review. We excluded one further study as the indicator of narcissism was too indirect to allow for comparisons with the rest of the literature (peer ratings of grandiosity were transformed into measures of sociometric preference in the crucial analysis involving a neuroendocrine marker; Bukowski, Schwartzman, Santo, Bagwell, & Adams, Reference Bukowski, Schwartzman, Santo, Bagwell and Adams2009). Table 2 summarizes the remaining 32 studies. In addition to these published studies, we also found two unpublished doctoral dissertations meeting the abovementioned criteria (Krusemark, Reference Krusemark2009; Noser, Reference Noser2017), which we also included in the main part of this review and Table 2, leading to a total of 34 studies. Finally, our literature search identified one hypothesis proposal (Jankowiak-Siuda & Zajkowski, Reference Jankowiak-Siuda and Zajkowski2013), two review articles that each cover parts of the research reviewed here (Coleman et al., Reference Coleman, Pincus and Smyth2019; Di Sarno et al., Reference Di Sarno, Di Pierro and Madeddu2018), one conceptual paper (George & Short, Reference George and Short2018), and four book chapters (Konrath & Bonadonna, Reference Konrath, Bonadonna and Besser2014; Elizabeth A. Krusemark, Reference Krusemark, Campbell and Miller2012, Reference Krusemark, Hermann, Brunell and Foster2018; Schulze & Roepke, Reference Schulze, Roepke, Mulert and Shenton2014), which do not present empirical data but rather integrate previous studies. We include these works in section 3 of the present review.

Table 2.

Summary of neuroscience studies on narcissism

The studies are described in section 2 of the main text. F, female; M, male. Abbreviations of narcissism measures: CNS, Childhood Narcissism Scale; DTDD, Dark Triad Dirty dozen; HSNS, Hypersensitive Narcissism Scale; NARQ, Narcissistic Admiration and Rivalry Questionnaire; NI, Narcissism Inventory; NPD, narcissistic personality disorder; NPI, Narcissistic Personality Inventory; NPI-16, abbreviated 16-item – version; PNI, Pathological Narcissism Inventory; SCATI, Short Coolidge Axis II inventory; SCID II, structured clinical interview for DSM-IV, personality disorders; SDT, short dark triad. Abbreviations of neuroscience measures: ECG, Electrocardiogram; EEG, Electroencephalogram; EEG, Electromyogram; DTI, diffusion tensor imaging; fMRI, functional magnetic resonance imaging; rs-fMRI, resting-state fMRI; VBM, voxel-based morphometry. Abbreviations of brain structures: ACC, anterior cingulate cortex; dACC, dorsal ACC; vACC, ventral ACC; sgACC, subgenual ACC; AI, anterior insula; mPFC, medial prefrontal cortex.

As Table 2 shows, among the 34 studies, 21 studies used experimental paradigms, in which participants were confronted with stimuli or tasks. Ten studies used baseline measures of brain activity or neuroendocrine markers, and five studies used structural brain imaging measures.Footnote ⁴ We classified the experimental paradigm studies as either primarily intrapersonal (self-related paradigms) or interpersonal (other related paradigms) in focus. This classification is in line with social/personality models of narcissism (e.g., Morf & Rhodewalt, Reference Morf and Rhodewalt2001) and with prevailing models of personality functioning, such as the DSM-5 AMPD (American Psychiatric Association, 2013). Though this distinction may not appear straightforward in each case (such as, for instance, the touch anticipation paradigm used by Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), we capitalized on whether the main focus of the paradigm is to manipulate intrapersonal, self-regulatory (e.g., emotion regulation), or interpersonal (e.g., empathizing, emotion recognition) psychological processes.

As Table 2 further shows, the vast majority of neuroscience research investigated the construct of grandiose narcissism (25 studies). Three studies investigated pathological narcissism.Footnote ⁵ Only two studies, to date, explicitly focused on vulnerable narcissism in which separate measures were used to assess vulnerability (though some discuss narcissistic vulnerability as a potential mechanism of the observed results). All studies on grandiose or pathological narcissism used nonclinical samples, mostly comprised of students. Six studies investigated NPD in patients.

In the following, we will summarize the findings of these studies grouped in functional, baseline, and structural studies of the different narcissism constructs. Table 2 provides an overview of all studies including sample characteristics. We will then integrate this research into a general discussion, in which we also consider prior reviews and book chapters on the topic.

2.3 Experimental studies

2.4 Baseline function studies

We identified 10 studies that investigated different biophysiological measures of baseline functioning in narcissism: six studies used measures of neuroendocrine markers including cortisol, other neuroendocrine markers targeting habitual stress levels, or testosterone, three investigated resting-state fMRI (one of them is the already discussed social touch study by Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), and one study used auditory evoked potentials in EEG.Footnote ¹² Except for two studies, which were carried out on NPD patients, all other studies investigated continuous differences in grandiose narcissism in relatively large samples.

Studies targeting neuroendocrine markers of habitual stress mostly speak to elevated stress in relation to narcissism, partially moderated by other variables: Reinhard and colleagues (Reference Reinhard, Konrath, Lopez and Cameron2012) carried out the first study on the association between a standardized measure of grandiose narcissism (NPI) and baseline cortisol levels. In a relatively large sample of young individuals, they observed an interaction among sex and narcissism in the way that narcissistic men have higher baseline cortisol across two time points separated by a 25-minute period. The effect sizes were substantial with correlations around r = .40. Critically, this effect could not be explained by potentially confounding psychological variables (positive/negative mood, stress, social support, or relationship status). The effects were stronger for the entitlement/exploitativeness factor than for other factors of the NPI, and were also evident in females when considering only this factor. This shows that particularly those aspects of grandiose narcissism that are most maladaptive (Ackerman et al., Reference Ackerman, Witt, Donnellan, Trzesniewski, Robins and Kashy2011) are associated with chronic stress. Subsequently, two studies attempted to indirectly or directly replicate the results observed by Reinhard and colleagues (Reference Reinhard, Konrath, Lopez and Cameron2012): the first study successfully replicated the associations between grandiose narcissism and baseline measures of cortisol, and extended it by measures of testosterone among young men (Pfattheicher, Reference Pfattheicher2016). It was found that higher levels of narcissism – but not of its near neighbor’s psychopathy or Machiavellianism – were related to higher levels of both baseline cortisol and baseline testosterone as assessed by two salivary probes separated by 6 min. The author interprets the findings in the way that men with higher levels of narcissism are at the same time responsive to social threat (cortisol) and socially dominant (testosterone). The second study, however, did not replicate the findings: Wardecker and colleagues (Reference Wardecker, Chopik, LaBelle and Edelstein2018) investigated the association of grandiose narcissism (NPI) and baseline cortisol in a large sample of young women and men. They did neither observe any overall associations between narcissism and cortisol, nor interactions with sex or effects for specific NPI factors, as reported in the original study. Though the replication attempt was well powered, methodological differences such as the saliva sampling procedure could still have contributed to the failure to replicate. Finally, an unpublished study replicated the associations between grandiose narcissism (assessed using the Dark Triad Dirty Dozen, which focuses on grandiosity, but also correlates with vulnerability; Jonason & Webster, Reference Jonason and Webster2010; Küfner et al., Reference Küfner, Dufner and Back2015) and baseline testosterone: Noser (Reference Noser2017) observed a correlation (r = 0.22) of narcissism and salivary testosterone in a large sample of adult men, but not of its neighboring personality dimensions psychopathy or Machiavellianism.

Two more studies investigated cortisol and further neuroendocrine markers in narcissism: Cheng and colleagues (Reference Cheng, Tracy and Miller2013) studied the relationships between cortisol, alpha-amylase, and grandiose narcissism in young women who provided saliva samples and ratings of positive and negative affect for three consecutive days. Both cortisol and alpha-amylase were higher in those who were high in grandiose narcissism and experienced more negative affect. The authors conclude that narcissism acts as a moderator in the relationship between perceived stress and negative health outcomes, as evident in cortisol (related to activity of the hypothalamic-pituitary-adrenal axis) and alpha-amylase (related to sympathetic nervous system activity). Finally, a recent study (Lee et al., Reference Lee, Gozal, Coccaro and Fanning2020) investigated baseline levels of 8-hydroxy-2'-deoxyguanosine (8-OH-DG), an indicator of oxidative stress at a cellular level (which has been associated with a variety of mental disorders; e.g., Salim, Reference Salim2014), in patients with different personality disorders, other mental disorders, and controls. Results showed higher levels of 8-OH-DG in patients with NPD and borderline personality disorder; no such effects were observed for other personality disorders. The effects were independent of relevant covariates including age, gender, alcohol, and cigarette use, major depression, and posttraumatic stress disorder. The authors interpret the findings in terms of increased biological stress responses to interpersonal hypersensitivity. Considered together with the studies by Reinhard and colleagues (Reference Reinhard, Konrath, Lopez and Cameron2012), Pfattcheicher (Reference Pfattheicher2016), Cheng and colleagues (Reference Cheng, Tracy and Miller2013), as well as the TSST study by Edelstein and colleagues (Reference Edelstein, Yim and Quas2010), these results provide relatively consistent evidence for exaggerated biological stress responses (cortisol, alpha-amylase, and 8-OH-DG) in relation to grandiose narcissism or NPD as its extreme expression, though one study did not find such associations (Wardecker et al., Reference Wardecker, Chopik, LaBelle and Edelstein2018). The exact nature of stress responses might depend on sex or gender (see section 4), the expression of entitlement/exploitativeness, which most closely reflects the antagonistic core of the construct (Krizan & Herlache, Reference Krizan and Herlache2018; Weiss et al., Reference Weiss, Campbell, Lynam and Miller2019) and is considered most maladaptive (Ackerman et al., Reference Ackerman, Witt, Donnellan, Trzesniewski, Robins and Kashy2011), and subjective stress levels as moderating factors.

Three studies investigated resting-state functional brain connectivity by means of fMRI. The first of these (Yang et al., Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015) studied functional connectivity differences in relation to grandiose narcissism (NPI) in a large sample of young women and men. The authors identified brain structural correlates of narcissism in the first step (see section 2.5), and then used regions in the right superior parietal lobe and the frontal eye fields – parts of the dorsal attention network – as seeds for subsequent functional connectivity analyses. In women, increased anticorrelations between the dorsal attention network and the default mode network were found as a function of narcissism. In men, the opposite pattern emerged, with decreased anticorrelation between the dorsal attention network and the default mode network as a function of narcissism. The authors interpret the effects in terms of a proneness toward outward-directed admiration seeking in more narcissistic women and loose boundaries between internal and external events in more narcissistic men. The second and most recent resting-state fMRI study of grandiose narcissism (NPI, Chinese Likert scale adaption) attempted to predict trait narcissism in a large sample of young individuals by means of machine learning applied to functional connectivity maps (Feng et al., Reference Feng, Yuan, Geng, Gu, Zhou, Wu and Luo2018). A network specific for subjects high in grandiose narcissism consisting of key nodes in the amygdala, the lateral and medial prefrontal cortex (mPFC), and the anterior cingulate cortex could be identified. Validation analyses showed that connectivity information within this network predicts grandiose narcissism scores in an independent sample of participants. While the network identified in this study is complex, interestingly, it encompasses some of the regions that were also found to discriminate between lower and higher narcissistic individuals in experimental functional imaging studies, above all the dACC (Cascio et al., Reference Cascio, Konrath and Falk2015; Chester & DeWall, Reference Chester and DeWall2016; Jauk, Benedek, et al., Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017) as part of the salience network. Individuals high in grandiose narcissism further display altered connectivity within regions commonly implicated in emotion processing (given the role of the amygdala, as part of the limbic system, in fear conditioning; LeDoux, Reference LeDoux2000) and regions belonging to the default mode network (mPFC), for which structural changes can be observed (see section 2.5). Finally, in the already discussed combined resting-state and task-based fMRI study by Scalabrini and colleagues (Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), a power law exponent analysis, measuring long-range temporal correlations in resting-state data, was conducted in ROIs that were functionally identified in the social touch experiment (human hand vs. mannequin hand: left postcentral gyrus, right culmen, and right AI; the latter only in interaction with grandiose narcissism). A significant positive association between grandiose narcissism and long-range temporal correlation was found in the AI, which the authors interpret in terms of increased self-focus in more grandiose individuals, particularly concerning aspects of the bodily and interoceptive self. Taken together, these studies show that grandiose narcissism can be related to intrinsic functional connectivity differences in the resting brain, and that the involved regions – parts of the default mode and salience networks – overlap with those implicated in imaging studies of intra- and interpersonal functioning (see section 3.3 for integrative discussion).

Finally, one EEG study used auditory evoked potentials to investigate cerebral information processing differences in several personality disorder groups, among them a group of NPD patients (Wang et al., Reference Wang, Wang, Fu, Liu, He, Dong and Jang2006). The neural responses to auditory stimuli of different amplitude were studied. It was hypothesized that certain personality disorder groups would display an “augmenting” evoked potential pattern indicative of sensation seeking and low serotonin levels. Such a pattern was found for patients with histrionic personality disorder, but no differences emerged for NPD. As with the studies reviewed in sections 2.3.1 and 2.3.2 above, it seems that low-level sensory processing is not generally altered in narcissism.

2.5 Structural imaging studies

Four structural imaging studies on narcissism were carried out so far, investigating gray matter volume, cortical thickness, and integrity of white matter tracts in grandiose narcissism, pathological narcissism, and NPD. Also, one of the abovementioned studies on resting-state functional connectivity in grandiose narcissism identified structural correlates (Yang et al., Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015). We will first review results on gray matter, followed by results on white matter.

The first structural imaging study by Schulze and colleagues (Reference Schulze, Dziobek, Vater, Heekeren, Bajbouj, Renneberg, Heuser and Roepke2013) investigated gray matter volume in a group of NPD patients compared to matched healthy controls by means of voxel-based morphometry (VBM). Gray matter volume differences were found in the left AI and in the rostral anterior and posterior cingulate cortex, as well as the dorsolateral and mPFC. The region of gray matter volume reduction in the AI overlapped with a region identified as a structural correlate of self-reported empathy in the whole sample (though no significant empathy differences were found in the self-report measure). The other regions also overlap with those commonly implicated in empathy (Lamm et al., Reference Lamm, Decety and Singer2011), which, taken together, corroborates the notion of reduced empathic functioning in NPD at the level of brain structure. A subsequent study compared gray matter volume in six male NPD patients to controls (Nenadic et al., Reference Nenadic, Güllmar, Dietzek, Langbein, Steinke and Gaser2015). Lower gray matter volume was found in the right middle frontal gyrus, left medial prefrontal/anterior cingulate cortex, left middle occipital cortex, left fusiform/inferior temporal cortex, right superior temporal cortex, and in the left lingual gyrus. While some of these areas overlap with those reported by Schulze and colleagues (Reference Schulze, Dziobek, Vater, Heekeren, Bajbouj, Renneberg, Heuser and Roepke2013), this study did not find reduced gray matter volume in the AI. A further study investigated structural correlates of pathological narcissism (PNI) in a nonclinical sample: Mao and colleagues (Reference Mao, Sang, Wang, Hou, Huang, Wei, Zhang and Qiu2016) studied cortical thickness and volume using surface-based morphometry in a large sample of young individuals. They found lower cortical thickness and volume in the right dorsolateral prefrontal cortex. Cortical thickness was further negatively associated with pathological narcissism in the right inferior frontal cortex, cortical volume was negatively associated in the right postcentral gyrus and the left mPFC. Finally, the aforementioned resting-state connectivity study by Yang and colleagues (Yang et al., Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015) also reported a positive structural correlation between grandiose narcissism and regional gray matter volume in the right superior parietal lobe, which was evident only in women. The authors interpreted this effect in terms of increased attention seeking in women high in grandiose narcissism, as the superior parietal lobe is part of the dorsal attention network.

Turning now from gray matter to white matter, two studies investigated the integrity of fiber tracts using diffusion-tensor imaging: the aforementioned VBM study by Nenadic and colleagues (Reference Nenadic, Güllmar, Dietzek, Langbein, Steinke and Gaser2015) additionally investigated fractional anisotropy, a marker of integrity of white matter tracts (Basser et al., Reference Basser, Mattiello and LeBihan1994), in NPD. Fractional anisotropy was lower in the right frontal lobe the right anterior thalamic radiation, the right anterior temporal lobe, the left anterior/lateral temporal lobe, and the right brain stem. The authors state that “white matter pathology might contribute to the structural cortical deficits [but] obviously, these findings are in need of replication” (p. 185). Such a replication can be seen in the findings of Chester and colleagues (Reference Chester and DeWall2016), who also investigated the integrity of white matter tracts using fractional anisotropy as an indicator, and grandiose narcissism as a predictor. In a sample of young women and men, the a priori assumed negative relationship of grandiose narcissism and fractional anisotropy in the frontostriatal pathway, connecting the ventral striatum and the medial prefrontal cortex, was confirmed. The authors interpret this finding in terms of a neural disconnect between regions involved in reward (striatal) and intrapersonal (mPFC) processing, which might underlie narcissistic strivings for admiration from the social environment (Back et al., Reference Back, Küfner, Dufner, Gerlach, Rauthmann and Denissen2013; Morf & Rhodewalt, Reference Morf and Rhodewalt2001).

Summarizing the structural findings, the first thing that catches the eye is that all studies find negative associations between grandiose narcissism, pathological narcissism, or NPD and different indicators of brain structure or integrity (gray matter volume, cortical volume, and thickness, fractional anisotropy of fiber tracts), with the sole exception of Yang and colleagues (Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015), who find a positive association between grandiose narcissism and gray matter volume in women. Though it could be hypothesized that increased self-focus – a core aspect of narcissism – might be accompanied by positive structural correlates in self-related areas such as key nodes of the default mode network, or also the anterior cingulate cortex, this does not seem to be the case, based on the existing evidence. This evidence rather exclusively points to dysfunctional aspects of narcissism in terms of reduced neuronal integrity. To this end, the most consistent finding across the existing studies seems to be lowered gray matter volume in the mPFC (Mao et al., Reference Mao, Sang, Wang, Hou, Huang, Wei, Zhang and Qiu2016; Nenadic et al., Reference Nenadic, Güllmar, Dietzek, Langbein, Steinke and Gaser2015; Schulze et al., Reference Schulze, Dziobek, Vater, Heekeren, Bajbouj, Renneberg, Heuser and Roepke2013).

3.1 Intrapersonal functions in narcissism

One of the central conclusions that can be drawn from the reviewed neuroscience research on intrapersonal functions in narcissism is that individuals with high levels of grandiose narcissism display exaggerated neurophysiological and neuroendocrine stress reactions to situations involving actual or anticipated ego threat, and also display habitually heightened stress indicators. This could be seen as supporting the long-standing mask model of narcissism, which posits that outward displays of grandiosity serve to compensate for an underlying, fragile, and vulnerable self (Akhtar, Reference Akhtar1989). While clinicians working with narcissistic patients widely acknowledge the coexistence of grandiose and vulnerable aspects (Pincus & Lukowitsky, Reference Pincus and Lukowitsky2010; Ronningstam, Reference Ronningstam2009), this view is not without controversy in quantitative research on narcissism. Conflicting findings have been obtained for the relationship between explicit and implicit self-esteem in grandiose narcissism, partially supporting (Zeigler-Hill, Reference Zeigler-Hill2006) or disproving (e.g., Campbell et al., Reference Campbell, Bosson, Goheen, Lakey and Kernis2007) the mask model view. The neuroscience of narcissism might provide an alternative resolution to this long-standing controversy, as it focuses on observable neurophysiological or neuroendocrine reactions to particular stimuli, or measures established baseline indicators, rather contrasting the global positivity/negativity of implicit or explicit self-concepts all at once, such as, for instance, when comparing measures of implicit and explicit self-esteem. This line of research might allow drawing specific inferences on latent, maybe implicit processes that can be formulated in terms of situational contingencies (cf. Roche et al., Reference Roche, Pincus, Conroy, Hyde and Ram2013; Wright & Edershile, Reference Wright and Edershile2018), such as “when confronted with situations that involve actual or anticipated ego threat, individuals with high grandiose narcissism react with increased vigilance and stress responses”. Narcissism, hereby, acts as a moderating factor that can impact either the strength or also the direction of effects. This view is consistent with the threatened egotism model of narcissism (Baumeister et al., Reference Baumeister, Smart and Boden1996; Bushman & Baumeister, Reference Bushman and Baumeister1998), which posits that individuals high in narcissism are highly sensitive to potential ego threat, assign higher self-relevance to potentially threatening situations, and display exaggerated responses facing such threat. Higher subjective self-relevance can also be observed for approach-driven goals (Spencer et al., Reference Spencer, Foster and Bedwell2018), as evident in the ergometer study (Woodman et al., Reference Woodman, Roberts, Hardy, Callow and Rogers2011).

Interestingly, recent research on communal narcissism (agentic strivings expressed in the communal domain; Gebauer et al., Reference Gebauer, Sedikides, Verplanken and Maio2012) has unveiled similar discrepancies between behavioral and neurophysiological indicators, as it was found in the ultimatum game that individuals high in communal narcissism are not more likely to reject unfair offers, but display stronger electrophysiological responses (P3 amplitude), indicative of greater emotional sensitivity to these offers (Yang, Sedikides, Gu, Luo, Wang, Yang, et al., Reference Yang, Sedikides, Gu, Luo, Wang, Yang and Cai2018). The study of diverging behavioral and neural responses might thus add another layer to understanding the complex relations between self-report, behavior, and involuntary neurophysiological reactions.

Neurophysiological research on the narcissistic self has investigated grandiose narcissism in relation to social rejection and found that imagined or actually experienced rejection goes along with higher systolic blood pressure as a stress indicator (Sommer et al., Reference Sommer, Kirkland, Newman, Estrella and Andreassi2009) and stronger activation in the dACC, AI, and subgenual ACC, regions overlapping largely with the salience network and referred to by the authors as the social pain network (Cascio et al., Reference Cascio, Konrath and Falk2015). These findings complement self-report research on social exclusion in grandiose narcissism, in which individuals high in grandiose narcissism do not report feeling threatened by social exclusion (Besser & Priel, Reference Besser and Priel2010). Neurophysiological research on social exclusion in narcissism thus reveals involuntary stress reactions that might either not be reported, or also not be accessible to the narcissistic individuals themselves. Stronger physiological responses in response to social status threats can already be observed in children (Grapsas et al., Reference Grapsas, Denissen, Lee, Bos and Brummelman2020). Importantly, these reactions are not without social consequences, as Chester and DeWall (Reference Chester and DeWall2016) report that those who are habitually high in grandiose narcissism and experience social rejection, as indicated by dACC activity, also display aggressive behavior toward others, as predicted by the threatened egotism model (Bushman & Baumeister, Reference Bushman and Baumeister1998). Similarly, as Lobbestael and colleagues (Reference Lobbestael, Baumeister, Fiebig and Eckel2014) reported, individuals high in grandiose narcissism have a higher tendency to behave aggressively toward others, which is accompanied by testosterone increase. This shows that a neuroscience approach to understanding narcissism may shed light on mechanisms underlying highly socially relevant phenomena, such as aggressive behavior, which may not be apparent in self-report research. Finally, men high in grandiose narcissism also display exaggerated stress responses as evident in high cortisol levels following performance-related ego threat (as induced by the TSST; Edelstein et al., Reference Edelstein, Yim and Quas2010), which is in line with self-report findings (Besser & Priel, Reference Besser and Priel2010). Whether or not potentially stressful situations induce stress reactions likely depends on the degree of ego threat (in terms of the possibility to fail), as lying about the self is accompanied by effects in the opposite direction (decreases in cortisol, increases in testosterone; Dane et al., Reference Dane, Jonason and McCaffrey2018). More generally, stimuli that are aversive and self-related, but not self-relevant in any way, such as the loud tones used by Kelsey and colleagues (Reference Kelsey, Ornduff, McCann and Reiff2001), do not induce stress responses, but rather evoke “cool” and prepared reactions, which reminds of fearlessness traditionally associated with the psychopathy construct (Lykken, Reference Lykken1995; for contemporary model, see Hare & Neumann, Reference Hare and Neumann2008).

While the abovementioned experimental paradigms directly challenge the self via exposure to threatening situations, the self-face viewing paradigm used in Jauk, Benedek, and colleagues (Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017) cannot be considered intrinsically threatening. Nevertheless, self-face viewing also induced saliency-related activation in the dACC (overlapping with the results reported by Cascio et al., Reference Cascio, Konrath and Falk2015), which indicates that presumably neutral, or potentially even rewarding stimuli can induce higher vigilance to potential ego threat in highly narcissistic men if they are both self-related (in this case also self-referential) and self-relevant (in terms of subjective value). A recent EEG study showed that sensory processing of self-related stimuli is modulated by antagonistic and agentic aspects of narcissism, which further underpins the notion of increased self-relevance for individuals high on narcissism (Mück et al., Reference Mück, Ohmann, Dummel, Mattes, Thesing and Stahl2020). Finally, regarding pathological narcissism, which also includes narcissistic vulnerability at an explicit level, even situations that might be expected to induce feelings of power might be perceived as threatening (Stenstrom et al., Reference Stenstrom, Dinsmore, Kunstman and Vohs2018).

Regarding the interpretation of studies showing activation differences in regions of the salience network – particularly the dACC – for self-relevant stimuli in narcissism, it is important to note that salience network activation cannot per se be attributed to a unitary psychological process (Seeley, Reference Seeley2019). According to the adaptive control hypothesis, the dACC integrates signals from multiple systems under conditions of uncertainty to enable rapid action (Shackman et al., Reference Shackman, Salomons, Slagter, Fox, Winter and Davidson2011). This adaptive control is likely more relevant facing potential threat than potential reward, also from an evolutionary perspective (de la Vega et al., Reference de la Vega, Chang, Banich, Wager and Yarkoni2016; Shackman et al., Reference Shackman, Salomons, Slagter, Fox, Winter and Davidson2011), but the implication of the dACC in processes with positive valence has also been discussed (Shackman et al., Reference Shackman, Salomons, Slagter, Fox, Winter and Davidson2011). While the observed differences can thus not unequivocally be interpreted in terms of valence, one plausible interpretation of dACC activity in the reviewed studies seems to be that individuals high in narcissism show stronger saliency-related responses facing self-relevant stimuli or situations. On the basis of the understanding of dACC function and the reviewed literature, this seems to be more likely for potentially threatening than rewarding situations, but interpretations might also vary between studies.

The notion of exaggerated physiological reactions in grandiose narcissism is further supported by baseline studies showing that cortisol levels are habitually higher in narcissistic men (Pfattheicher, Reference Pfattheicher2016; Reinhard et al., Reference Reinhard, Konrath, Lopez and Cameron2012) and also women who score high on the entitlement/exploitativeness factor of the NPI (Reinhard et al., Reference Reinhard, Konrath, Lopez and Cameron2012), which most closely marks the core of the narcissism construct (Krizan & Herlache, Reference Krizan and Herlache2018; Weiss et al., Reference Weiss, Campbell, Lynam and Miller2019). Women high in grandiose narcissism also display elevated stress levels (cortisol and alpha-amylase) when negative affect is high (Cheng et al., Reference Cheng, Tracy and Miller2013). Recent research shows that stress levels are even higher on a cellular level in NPD (Lee et al., Reference Lee, Gozal, Coccaro and Fanning2020). Taken together, this is indicative of narcissistic men, and to some extent women, experiencing higher chronic stress levels, presumably due to higher sensitivity to potentially threatening self-relevant situations.

Concerning sex differences in relation to narcissism, available evidence suggests that men display stronger maladaptive reactions than women (Edelstein et al., Reference Edelstein, Yim and Quas2010; Jauk, Benedek, et al., Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017; Reinhard et al., Reference Reinhard, Konrath, Lopez and Cameron2012). However, there is also evidence that women display strong reactions when perceived stress is high (Cheng et al., Reference Cheng, Tracy and Miller2013). Sex differences were additionally observed at the level of intrinsic brain network connectivity and might point to different experiential and behavioral tendencies between men and women high in grandiose narcissism (Yang et al., Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015). While some of the reviewed findings support the notion that narcissism might be more maladaptive in men than in women (cf. Jauk, Freudenthaler, et al., Reference Jauk, Freudenthaler and Neubauer2016; Morf & Rhodewalt, Reference Morf and Rhodewalt2001), it is currently unknown whether and to which extent this is due to biological or social factors. Though some research points to associations between sex-specific hormones (testosterone; Noser, Reference Noser2017; Pfattheicher, Reference Pfattheicher2016) and narcissism, other research also demonstrates the dynamic nature of these associations, which is dependent upon social factors (Dane et al., Reference Dane, Jonason and McCaffrey2018; Mead et al., Reference Mead, Baumeister, Stuppy and Vohs2018). In section 4, we provide suggestions on how future research might address the question of sex and gender differences in neuroscience studies of narcissism.

The notion of an implicit negative self-view in grandiose narcissism receives partial support from structural imaging studies: Chester, Lynam, and colleagues (Reference Chester, Lynam, Powell and DeWall2016) reported lower integrity of the frontostriatal pathway connecting the ventral striatum and the mPFC. These findings point to a neural disconnect between rewarding and intrapersonal brain areas, which the authors interpret in the way that reduced intrinsically rewarding activity might lead to admiration seeking in the social environment, as posited by self-regulatory models of grandiose narcissism (Back et al., Reference Back, Küfner, Dufner, Gerlach, Rauthmann and Denissen2013; Morf & Rhodewalt, Reference Morf and Rhodewalt2001). This interpretation is consistent with neuroscience research on self-esteem, demonstrating that individuals with high self-esteem display positive rather than negative associations with white matter integrity in the frontostriatal pathway (Chavez & Heatherton, Reference Chavez and Heatherton2017). This might point to a neural dissociation between narcissistically inflated self-esteem and genuinely high self-esteem. This dissociation is particularly interesting given that self-esteem and narcissism are moderately positively related in self-report studies (Campbell, Reference Campbell2001). Neuroscience can thus add a level of analysis that might inform about the underlying psychological processes.

Further structural imaging studies were carried out in NPD patients and groups differing in grandiose and pathological narcissism. The most striking parallel of those studies is that all but one report negative associations between narcissism and different indicators of brain structure; positive associations are hardly evident in the literature. Schulze and colleagues (Reference Schulze, Dziobek, Vater, Heekeren, Bajbouj, Renneberg, Heuser and Roepke2013) found lower gray matter volume in the left AI (for a discussion that also includes other personality disorders, see Schulze & Roepke, Reference Schulze, Roepke, Mulert and Shenton2014) and in the anterior and posterior cingulate cortex, as well as the dorsolateral and mPFC. Nenadic and colleagues (Reference Nenadic, Güllmar, Dietzek, Langbein, Steinke and Gaser2015) found lower gray matter volume in the right middle frontal gyrus, left medial prefrontal/anterior cingulate cortex, left middle occipital cortex, left fusiform/inferior temporal cortex, right superior temporal cortex, and in the left lingual gyrus. Mao and colleagues (Reference Mao, Sang, Wang, Hou, Huang, Wei, Zhang and Qiu2016) found negative relations between pathological narcissism and cortical thickness as well as volume in the right dorsolateral prefrontal cortex. Cortical thickness was further negatively associated with pathological narcissism in the right inferior frontal cortex, cortical volume was negatively associated in the right postcentral gyrus and the left mPFC. Though most of these regions are nonoverlapping, all three studies find effects in the mPFC, which was also implicated in one of the white matter integrity studies discussed above (Chester et al., Reference Chester, Lynam, Powell and DeWall2016). The mPFC, as part of the default mode network (Gusnard et al., Reference Gusnard, Akbudak, Shulman and Raichle2001), is considered central to intrapersonal processing (Wagner et al., Reference Wagner, Haxby and Heatherton2012), specifically to modulating self-referential processing within the default mode network (Davey et al., Reference Davey, Pujol and Harrison2016). Lower gray matter volume in the mPFC is associated with cumulative lifetime adversity and might reflect general disintegration of intrapersonal functions at a brain structural level (Ansell et al., Reference Ansell, Rando, Tuit, Guarnaccia and Sinha2012). These findings are further consistent with multivariate connectivity differences among the mPFC, dACC, and amygdala in grandiose narcissism (Feng et al., Reference Feng, Yuan, Geng, Gu, Zhou, Wu and Luo2018). Though the exact nature of these differences is yet to be uncovered, they point to alterations regarding self-referential processes within the default mode network (mPFC), saliency-related processes (dACC), and emotional processes (fear processing in the amygdala; LeDoux, Reference LeDoux2000). One structural imaging study, however, did not find differences in the mPFC, but reports a positive association between grandiose narcissism in the right superior parietal lobe for women (Yang et al., Reference Yang, Cun, Du, Yang, Wang, Wei and Qiu2015). On the basis of the current research, the findings of this study appear hard to reconcile with the rest. The main differences between this study and the others are that it investigated grandiose narcissism (NPI), whereas the other studies investigated either NPD patients or pathological narcissism (PNI), and that structural correlates were studied in interaction with sex.

When considering all of the available evidence of functional, baseline, and structural studies together, the following main points can be summarized from the available evidence:

1. (1) Grandiose narcissism is related to neurophysiological stress reactions indicative of vigilance to actual or anticipated ego threat, which points to vulnerable aspects of grandiose narcissism that are not commonly apparent in self-report research. These reactions manifest in increased salience network activation in the dACC, AI, and partly, also in the ventral/subgenual ACC, as well as increased autonomic (sympathetic; systolic blood pressure) and neuroendocrine (cortisol) indicators. Stressful responses to stimuli occur only for self-relevant material; reactions to aversive stimuli that are self-related but not self-relevant can be “cool”.

2. (2) Grandiose narcissism and NPD are related to altered baseline functioning. This manifests in higher cortisol or 8-OH-DG levels, and distinguishable resting-state connectivity patterns among regions that are also implicated in experimental research (dACC, salience network) as well as the mPFC (default mode network) and the amygdala (limbic system).

3. (3) Grandiose narcissism, pathological narcissism, and NPD are related to brain structural changes, which almost exclusively manifest in negative associations. The most consistent finding is lower cortical volume/gray matter volume in the mPFC for pathological narcissism/NPD. White matter in the frontostriatal pathway, connecting the ventral striatum and the mPFC, displays lower integrity related to narcissistic grandiosity. Prefrontal white matter tracts are also weakened in NPD.

3.2 Interpersonal functions in narcissism

The study of interpersonal functions in narcissism mostly targets the long-lasting notion of empathy deficits in narcissism, as included in the diagnostic criteria for NPD since its first mention in the DSM-III (American Psychiatric Association, 1980). While lowered empathic functioning, in terms of reduced sharing of others’ affective states, is well documented for all expressions of narcissism (e.g., Lannin et al., Reference Lannin, Guyll, Krizan, Madon and Cornish2014; Pincus et al., Reference Pincus, Ansell, Pimentel, Cain, Wright and Levy2009; Ritter et al., Reference Ritter, Dziobek, Preißler, Rüter, Vater, Fydrich and Roepke2011; Wai & Tiliopoulos, Reference Wai and Tiliopoulos2012), the exact nature of reduced empathy in narcissism is still a matter of debate. Specifically, it has been questioned whether reduced interpersonal functioning is more indicative of reduced capacity or reduced propensity (Baskin-Sommers et al., Reference Baskin-Sommers, Krusemark and Ronningstam2014), with some evidence pointing into the latter direction (as empathic reactions can be increased by interventions; Hepper et al., Reference Hepper, Hart and Sedikides2014). The neuroscience study of narcissism might contribute to an understanding of empathic alterations, as it provides objective assessments of the respective activations complementary to self-report data.

From the literature reviewed above, it can be concluded that individuals with higher levels of pathological and grandiose narcissism generally display lowered activation in brain regions typically implicated in empathizing, particularly the AI, when using high-level social affect paradigms (Fan et al., Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011; Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017). These findings were, on the one hand, explained along the line of greater alexithymia in pathological narcissism (Fan et al., Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011), and on the other hand , as greater self-related attention in grandiose narcissism (Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017).

Regarding the first of these interpretations, alexithymia, in terms of a general incapacity to identify emotions, was also observed to mediate empathy deficits in aggressive offenders (Winter et al., Reference Winter, Spengler, Bermpohl, Singer and Kanske2017), and is currently being discussed as a general, transdiagnostic functional mechanism of reduced empathy, which could also play a role in borderline personality disorder or psychopathy (Valdespino et al., Reference Valdespino, Antezana, Ghane and Richey2017), depression (Hoffmann et al., Reference Hoffmann, Banzhaf, Kanske, Gärtner, Bermpohl and Singer2016), and autism (Bird et al., Reference Bird, Silani, Brindley, White, Frith and Singer2010). With respect to the question of reduced propensity or capacity for empathic reactions in narcissism, the findings of Fan and colleagues (Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011) could, tentatively speaking, be seen as more in line with the capacity than the propensity hypothesis (see also Baskin-Sommers et al., Reference Baskin-Sommers, Krusemark and Ronningstam2014). Participants were explicitly instructed to empathize with the faces they saw, and high versus low narcissistic groups gave similar empathy ratings. Yet, they differed in their neural activation, which means that different levels of brain activation were observed under similar conditions. This notion is substantiated by the observation that empathy ratings and activation in the respective neural network are usually positively associated (Kanske et al., Reference Kanske, Böckler, Trautwein and Singer2015). However, alternative explanations cannot be ruled out. For instance, different activation strengths might be accompanied by similar affective experiences (for a parallel from cognitive research, see Neubauer & Fink, Reference Neubauer and Fink2009), or individuals with higher narcissism levels might provide biased empathy ratings, as this is the case for self-perceptions of general emotional competence (Jauk, Freudenthaler, et al., Reference Jauk, Freudenthaler and Neubauer2016; Mota et al., Reference Mota, Leckelt, Geukes, Nestler, Humberg, Schröder-Abé, Schmukle and Back2019). While these interpretative problems are partially epistemological in nature, future research could compare spontaneous and deliberate empathic reactions at a neurophysiological level to further elucidate the nature of reduced empathy in narcissism.

In the second high-level social affect study of narcissism that has been conducted to date, Scalabrini and colleagues (Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017) also observed lower activation in the right AI during anticipation of touching a human versus mannequin hand. This is generally in line with the previously discussed findings of Fan and colleagues (Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011), as both points to reduced responses in the AI to social stimuli in narcissism. The interpretation, however, differs from the previous one: Scalabrini and colleagues used complemental resting-state scans, which indicated a higher long-range temporal correlation in the AI in individuals higher in grandiose narcissism. The authors interpret this finding in terms of increased self-focus during resting state, particularly with respect to interoceptive and bodily self-aspects. This can be seen as at least partially supporting a hypothesis proposed by Jankowiak-Siuda and Zajkowksi (Reference Jankowiak-Siuda and Zajkowski2013), according to which defective function of the AI, as part of the salience network, leads to a higher likelihood of engaging default mode, self-referential processes in more narcissistic individuals. However, in the study by Scalabrini and colleagues (Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), it was the AI itself, which displayed increased long-range temporal correlation during resting state. Finally, the alexithymia interpretation by Fan and colleagues (Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011) and the interoception – interpretation by Scalabrini and colleagues (Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017) might be hard to reconcile, as alexithymia and interoceptive abilities are themselves negatively related (Brewer et al., Reference Brewer, Cook and Bird2016). Thus, for now, it can be concluded that both pathological and grandiose narcissism (grandiose factor of the PNI) are associated with relative deactivation of the AI during high-level social affect paradigms, though the exact nature of this phenomenon must remain subject to future studies.

Turning now to low level, more basic experimental paradigms, two EEG studies found that sensory processing of other related stimuli is not generally impaired in NPD (Marcoux et al., Reference Marcoux, Michon, Lemelin, Voisin, Vachon-Presseau and Jackson2014; Zhang et al., Reference Zhang, Shen, Zhu, Ma and Wang2016). The first of these studies showed that NPD patients even display an increased somatosensory gating effect to observing others’ pain, that is, a stronger reduction of own somatosensory activity (P3 component) when anticipating or observing others’ pain (Marcoux et al., Reference Marcoux, Michon, Lemelin, Voisin, Vachon-Presseau and Jackson2014). The second study found no neurophysiological differences in the processing of emotional faces between NPD patients and healthy controls, though the amount of depression reported by NPD negatively modulated P2 amplitudes to neutral and happy faces, which shows that depression amplifies self-focus in NPD (Zhang et al., Reference Zhang, Shen, Zhu, Ma and Wang2016). Taken together, these studies show that NPD patients do not generally show low-level sensory deficits in the processing of social stimuli, but comorbid symptoms must be considered when studying NPD patients.

Finally, the social–psychological line of neuroscience research on narcissism points to dynamic interactions between narcissism, testosterone, and the subjective experience of power: those with higher baseline testosterone levels were found to score higher on the NPI following a social power manipulation (Mead et al., Reference Mead, Baumeister, Stuppy and Vohs2018), which aligns with the finding that – the other way around – testosterone increases in more narcissistic men after acting deceptively (Dane et al., Reference Dane, Jonason and McCaffrey2018). Testosterone is also correlated with narcissism at baseline in men (Noser, Reference Noser2017; Pfattheicher, Reference Pfattheicher2016). Taken together, these findings point to the unbowed socially assertive and antagonistic face of narcissism, which likely arises in the absence of a subjective threat to the self.

When considering all of the available evidence together, the following main points can be summarized regarding interpersonal aspects:

1. (1) Individuals with higher levels of pathological and grandiose narcissism display lowered reactions to social stimuli in the AI in high-level paradigms. The AI also shows altered baseline activity in relation to narcissism, and is implicated in intrapersonal processing in narcissism. Though the results seem to be more in line with the notion of a reduced capacity than a reduced propensity for empathic reactions, more research will be needed to elucidate the involved mechanisms.

2. (2) Low-level social information processing, as investigated in electrophysiological studies, is likely not affected by narcissism. However, when studying patients with NPD, comorbid disorders, such as depression, might also alter these processes. This is consistent with the notion that interpersonal functioning in narcissism can substantially vary between different states.

3. (3) The assertive and antagonistic interpersonal style associated with grandiose narcissism is evident in neuroendocrine indicators such as increased testosterone levels. Research on neuroendocrine markers associated with narcissism also points to dynamic situational interactions.

3.3 Overall conclusion and conceptual integration

One of the main conclusions that can be drawn from the reviewed research is that intrapersonal functioning in grandiose narcissism is accompanied by heightened vigilance to ego threat and stress responses following subjective ego threat, and also heightened stress indicators at a baseline level (Cascio et al., Reference Cascio, Konrath and Falk2015; Cheng et al., Reference Cheng, Tracy and Miller2013; Edelstein et al., Reference Edelstein, Yim and Quas2010; Jauk, Benedek, et al., Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017; Reinhard et al., Reference Reinhard, Konrath, Lopez and Cameron2012; Sommer et al., Reference Sommer, Kirkland, Newman, Estrella and Andreassi2009). These findings are consistent with the threatened egotism model of narcissism (Bushman & Baumeister, Reference Bushman and Baumeister1998) and demonstrate that ego threat may not only have negative consequences for others (in terms of aggressive behavior) but also for the self. Importantly, stress reactions are not necessarily apparent in self-report research and point to vulnerable aspects of grandiose narcissism, which might either not be accessible to, or not reported by individuals high in narcissism. This is in line with behavioral research demonstrating overestimation of emotion-related abilities in narcissism (Ames & Kammrath, Reference Ames and Kammrath2004; Jauk, Freudenthaler, et al., Reference Jauk, Freudenthaler and Neubauer2016; John & Robins, Reference John and Robins1994; Lobbestael et al., Reference Lobbestael, de Bruin, Kok and Voncken2016; Mota et al., Reference Mota, Leckelt, Geukes, Nestler, Humberg, Schröder-Abé, Schmukle and Back2019; Zajenkowski et al., Reference Zajenkowski, Maciantowicz, Szymaniak and Urban2018). Also, it is in line with the conclusion of a recent review of stress reactivity in narcissism (Coleman et al., Reference Coleman, Pincus and Smyth2019), and shows that it extends to a broader spectrum of research paradigms and methods. The conclusion is also mostly in line with earlier chapters, which could not yet draw on a large amount of studies specific to narcissism (Konrath & Bonadonna, Reference Konrath, Bonadonna and Besser2014; Krusemark, Reference Krusemark, Campbell and Miller2012), and a recent chapter (Krusemark, Reference Krusemark, Hermann, Brunell and Foster2018). In particular, the present review shows that self-relevant stimuli, even when not necessarily stressful, can lead to similar reactions as those that are considered intrinsically stressful or threatening. Individuals high in grandiose narcissism show similar neural reactions in parts of the salience network and also similar “hard wiring” of the frontostriatal pathway as do individuals with low self-esteem (Chavez & Heatherton, Reference Chavez and Heatherton2017; Chester et al., Reference Chester, Lynam, Powell and DeWall2016; Jauk, Benedek, et al., Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017; Onoda et al., Reference Onoda, Okamoto, Nakashima, Nittono, Yoshimura, Yamawaki, Yamaguchi and Ura2010).

Importantly, self-relevance is a crucial situational variable in this equation, as higher narcissistic individuals do not show stress reactivity (Kelsey et al., Reference Kelsey, Ornduff, McCann and Reiff2001; Sylvers et al., Reference Sylvers, Brubaker, Alden, Brennan and Lilienfeld2008), or even cool and prepared reactions facing potentially stress-inducing stimuli which that are not of relevance to the self (Kelsey et al., Reference Kelsey, Ornduff, McCann and Reiff2001). Individuals higher in grandiose narcissism also make riskier decisions, accompanied by weaker neurophysiological reactions (Yang, Sedikides, Gu, Luo, Wang, & Cai, Reference Yang, Sedikides, Gu, Luo, Wang and Cai2018), and show reactions indicative of lower stress or even enjoyment following deceptive behavior (Dane et al., Reference Dane, Jonason and McCaffrey2018). These latter findings might point to common mechanisms underlying certain aspects of narcissism and the psychopathy construct (traditionally characterized by “fearlessness”; Hare & Neumann, Reference Hare and Neumann2008).

Interpersonal personality functioning, particularly empathic reactions, is reduced at a behavioral and neurophysiological level when assessed by high-level paradigms and measures (Fan et al., Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011; Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017). More low-level, sensory processing of other related information appears to be intact in NPD (Marcoux et al., Reference Marcoux, Michon, Lemelin, Voisin, Vachon-Presseau and Jackson2014; Zhang et al., Reference Zhang, Shen, Zhu, Ma and Wang2016), in terms of the most extreme expression of grandiose narcissism, though comorbid disorders (depression) might also alter low-level processes (Zhang et al., Reference Zhang, Shen, Zhu, Ma and Wang2016).

Interestingly, intra- and interpersonal functioning in narcissism partially draw on similar neural networks, as evident in functional, baseline, and structural imaging studies (see Figure 2). The dACC, as part of the salience network (Bressler & Menon, Reference Bressler and Menon2010), shows relatively higher activation in intrapersonal processing of actual or anticipated ego threat in individuals with higher grandiose narcissism (Cascio et al., Reference Cascio, Konrath and Falk2015; Jauk, Benedek, et al., Reference Jauk, Benedek, Koschutnig, Kedia and Neubauer2017). The AI, which is also considered as a part of the salience network (Bressler & Menon, Reference Bressler and Menon2010) and implicated in empathic processes (Lamm et al., Reference Lamm, Decety and Singer2011), shows relatively lower activation during interpersonal paradigms in highly grandiose and pathologically narcissistic individuals (Fan et al., Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011; Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), and it shows altered baseline functioning in individuals higher in grandiose narcissism (Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017). Taken together, a very outright interpretation of these findings could be that processing of intrapersonal stimuli in narcissism is accompanied by increased saliency, particularly when these are self-relevant and thus go along with potential ego threat. Processing of other related, interpersonal stimuli, however, seems to be associated with decreased saliency. Activation differences within the salience network thus point to a double dissociation regarding the expression of narcissism and self- and other relatedness of the situation (see Figure 2). This interpretation is partially consistent with a hypothesis proposed by Jankowiak-Siuda and Zajkowski (Reference Jankowiak-Siuda and Zajkowski2013), who also ascribed the salience network a central role in relation to narcissism (though other predictions cannot be evaluated on the basis of the present research), and with a conceptual article pointing to the associations among AI function, narcissism, empathy, and general emotional awareness (i.e., alexithymia; George & Short, Reference George and Short2018), which could also be seen as one potential underlying mechanism (see above; Valdespino et al., Reference Valdespino, Antezana, Ghane and Richey2017).

Figure 2.

Schematic depiction of central findings of neuroscience studies on narcissism. Dashed lines indicating the moderating influence of narcissism (see section 3) on particular effects. Activation within central nodes of the salience network (orange) and reactions in autonomic (sympathetic) or neuroendocrine markers are higher in situations involving intrapersonal focus when self-relevance is high (potential ego threat). Reactions to aversive self-related stimuli of low self-relevance can be “cool”. Situations involving interpersonal focus (empathy, social touch) lead to down-regulations. At a brain structural and baseline function level, narcissism is related to alterations in regions of the default mode network (cyan) and in the frontostriatal pathway (purple) as well as neuroendocrine markers.

Interestingly, the dACC was also among the key nodes found to display altered patterns of resting-state connectivity in individuals with higher levels of grandiose narcissism (Feng et al., Reference Feng, Yuan, Geng, Gu, Zhou, Wu and Luo2018). Among the other nodes in this study was also the dorsal and ventral mPFC (see Figure 2). The mPFC displayed lower gray matter volume in three out of the four morphometric studies (NPD and pathological narcissism), and is connected to the ventral striatum via the frontostriatal pathway (see Figure 2), which showed decreased white matter integrity in relation to grandiose narcissism (Chester et al., Reference Chester, Lynam, Powell and DeWall2016). The mPFC is considered as a part of the default mode network (Gusnard et al., Reference Gusnard, Akbudak, Shulman and Raichle2001) and central to intrapersonal processing (Davey et al., Reference Davey, Pujol and Harrison2016; Wagner et al., Reference Wagner, Haxby and Heatherton2012). Lower mPFC volume could indicate a general disintegration of intrapersonal functions, as it is also related to cumulative lifetime adversity (Ansell et al., Reference Ansell, Rando, Tuit, Guarnaccia and Sinha2012). Lower white matter integrity between the ventral striatum and the mPFC could further indicate a neural disconnect between regions associated with reward and self-processing (Chester et al., Reference Chester, Lynam, Powell and DeWall2016).

Taken together, neuroimaging findings point to functional and structural alterations of regions within the salience network (dACC, AI) and the default mode network (mPFC). Activity within the salience network points to a double dissociation of narcissism and self- versus other related processes.

Some of the conclusions presented here for intra- and interpersonal functioning in narcissism have, from different perspectives or with different emphases, also been expressed in previous integrative works. While we cannot reiterate all points addressed in previous reviews here, we want to highlight some central agreements with recent works: Coleman and colleagues (Reference Coleman, Pincus and Smyth2019) have recently reviewed studies investigating stress reactivity (psychological, biological, and behavioral) in narcissism, concluding that grandiosity, as well as vulnerability, are generally associated with increased stress reactivity. Krusemark (Reference Krusemark, Hermann, Brunell and Foster2018) came to a similar conclusion, stating that “grandiose narcissism involves cardiovascular reactivity in response to aversive stimuli and achievement-related stressors” and “vulnerable narcissism and exploitativeness/entitlement are generally associated with heightened cardiovascular reactivity to interpersonal rejection” (Reference Krusemark, Hermann, Brunell and Foster2018, p. 213). However, Coleman and colleagues further argue that “considerable nuance in these associations appears to exist (e.g., what contexts/stressors) […], and under certain stressful conditions narcissism (particularly grandiosity) may confer some level of resilience” (Reference Coleman, Pincus and Smyth2019, p. 61). We here refer to these moderating factors as self-relatedness and self-relevance, putting forward the hypothesis that increased salience network up-regulation or stress responses are only observed when both conditions are met. Potentially aversive stimuli or situations that are self-related, but not self-relevant might lead to unaffected responses, or as Coleman and colleagues (Reference Coleman, Pincus and Smyth2019) further assert, individuals high in grandiosity might have advantages in such situations. Similarly, Krusemark referred to these unaffected responses as a “hypo-reactive physiological profile” (Reference Krusemark, Hermann, Brunell and Foster2018, p. 215). Importantly, given that each of us experiences potentially ego-threatening situations, and – as we have argued – individuals high in narcissism might be even more prone to assign self-relevance to such situations, we hypothesize that this might be evident in chronic alterations of neuroendocrine systems or also brain structure in individuals high in narcissism. Krusemark (Reference Krusemark, Hermann, Brunell and Foster2018) also asserts that increased physiological reactivity associated with narcissism is accompanied by a heightened risk for cardiovascular disease, and similarly, Coleman and colleagues state that “maladaptive stress-reactivity, particularly when repeated often and over long periods, may cause wear-and-tear on bodily systems that ultimately hasten disease onset and progression” (Reference Coleman, Pincus and Smyth2019, p. 67). In addition to that, as Konrath and Bonadonna (Reference Konrath, Bonadonna and Besser2014) further note, narcissism is a “risk-positive personality style” (such as, for instance, also evident in the gambling study reviewed here; Yang, Sedikides, Gu, Luo, Wang, & Cai, Reference Yang, Sedikides, Gu, Luo, Wang and Cai2018), which is accompanied by a broad variety of potentially self-defeating, health-relevant behaviors such as sexual risk-taking, substance use, or risky driving. Thus, in addition to physical health risks imposed by increased stress reactivity, narcissism is also associated with a variety of lifestyle factors that might lead to serious health risks over the lifespan (Konrath & Bonadonna, Reference Konrath, Bonadonna and Besser2014).

Regarding interpersonal functioning in narcissism, George and Short (Reference George and Short2018) as well as Di Sarno and colleagues (Reference Di Sarno, Di Pierro and Madeddu2018) recently highlighted the role of the AI regarding empathic alterations in NPD. George and Short (Reference George and Short2018) further pinpoint that the AI is also involved in general emotional awareness, making alexithymia a possible common mechanism of self- and other related emotional deficits in narcissism. As discussed earlier (see section 3.2), while the AI – and more generally the salience network – might play a key role in self- versus other related processing in narcissism, the alexithymia interpretation (also originally put forward by Fan et al., Reference Fan, Wonneberger, Enzi, de Greck, Ulrich, Tempelmann, Bogerts, Doering and Northoff2011) might be hard to reconcile with the notion of increased interoception (Scalabrini et al., Reference Scalabrini, Huang, Mucci, Perrucci, Ferretti, Fossati and Ebisch2017), which is why the exact mechanisms of emotion-related alterations in narcissism must remain subject to future study.

Given these areas of overlap with related works, we think that the main contribution of our review is to integrate the available neuroscience literature on narcissism in a theoretical model spanning a broad variety of relevant psychological processes as well as a broad variety of indicators used in neuroscience research. Though such a model cannot adequately depict every finding obtained so far, it is an attempt to identify key topics which are evident across a wide range of studies and previous theoretical works, condense them into testable hypotheses, and put them into a larger perspective, thus moving one step beyond approaches which are either confined to a specific psychological process or a specific set of research methods. Whether or not the assumptions expressed in this model will hold true, we think that providing testable hypotheses is a crucial next step for future neuroscience works on narcissism, which we hope our work does encourage (see following section 4).