Unraveling the Paradoxes of Narcissism: A Dynamic Self-Regulatory Processing Model

We propose a dynamic self-regulatory processing model of narcissism and review supporting evidence. The model casts narcissism in terms of motivated self-construction, in that the narcissist's self is shaped by the dynamic interaction of cognitive and affective intrapersonal processes and interpersonal self-regulatory strategies that are played out in the social arena. A grandiose yet vulnerable self-concept appears to underlie the chronic goal of obtaining continuous external self-affirmation. Because narcissists are insensitive to others' concerns and social constraints and view others as inferior, their self-regulatory efforts often are counterproductive and ultimately prevent the positive feedback that they seek-thus undermining the self they are trying to create and maintain. We draw connections between this model and other processing models in personality and employ these models to further elucidate the construct of narcissism. Reconceptualizing narcissism as a self-regulatory processing system promises to resolve many of its apparent paradoxes, because by understanding how narcissistic cognition, affect, and motivation interrelate, their internal subjective logic and coherence come into focus.

If you ask people whether they have ever met a narcissist, most tell you about a friend, boss, or lover who was completely self-centered. They describe a person full of paradoxes: Self-aggrandizing and self-absorbed, yet easily threatened and overly sensitive to feedback from others. The friend, boss, or lover was emotionally labile and prone to extremes of euphoria, despair, and rage. They were often charming and socially facile while simultaneously insensitive to others' feelings, wishes, and needs. Some might report that they were initially attracted to such individuals only to grow weary of their constant demands for admiration and attention.
More formally, narcissism is defined by the Diagnostic and Statistical Manual of Mental Disorders (4th ed. [DSM-IV]; American Psychiatric Association, 1994) as a pervasive pattern of grandiosity, self-focus, and self-importance. According to the DSM-IV, narcissists are preoccupied with dreams of success, power, beauty, and brilliance. They live on an interpersonal stage with exhibitionistic behavior and demands for attention and admiration but respond to threats to self-esteem with feelings of rage, defiance, shame, and humiliation. In addition, they display a sense of entitlement and the expectation of special treatment. They are unwilling to reciprocate the favors of others and are unempathetic and interpersonally exploitative. In addition, as our friends noted, they have relationships that oscillate between idealization and devaluation.
Narcissists fascinate many of us, because they appear to possess such an incongruous set of characteristics and perhaps because they seem like adult versions of infantile characteristics most people leave behind early in the course of development. Likewise, the topic of narcissism has commanded a long-standing fascination in the psychoanalytic and clinical literatures and is enjoying a recent resurgence of interest from personality and social psychologists. We suggest that a significant part of the continuing scholarly interest in narcissism stems from the fact that the syndrome is highly complex, difficult to define and measure, and linked to a number of somewhat conflicting theoretical perspectives.
Clinical theorists hotly debate narcissism's developmental antecedents, as well as the meaning and operations of its defining characteristics. For social and personality psychologists interested in the social construction of the self, the narcissists' preoccupation with building, buttressing, and defending its desired self takes on a particular opportunity and challenge. The syndrome provides an ideal prototype for examining how these self-processes are played out in both the social world and inside the mind of the narcissist. In short, we suggest that both the layperson's and the psychologist's fascination with narcissism lies in the challenges inherent in understanding the underlying psychological dynamics of narcissistic behavior. What keeps the construct alive as a research topic is that these underlying psychological dynamics turn out to be far more complex than the popular, intuitive conception of "the narcissistic friend, boss, or lover" might imply.
In this article, we illustrate the utility of an approach that focuses on narcissism more as personality process than as static individual difference. We describe a program of research that has begun to validate a model of narcissism that casts the syndrome in terms of a distinctive dynamic 1 system of social, cognitive, and affective self-regulatory processes. The model assumes that these self-regulatory processes are in the service of motivated self-construction directed at building or maintaining desired selves and meeting self-evaluative needs. We argue that underlying narcissistic self-regulation is a grandiose, yet vulnerable self-concept. This fragility drives narcissists to seek continuous external self-affirmation. Furthermore, much of this self-construction effort takes place in the social arena. Yet, because narcissists are characteristically insensitive to others' concerns and social constraints, and often take an adversarial view of others, their self-construction attempts often misfire. Thus, although narcissistic strategic efforts generally help maintain self-esteem and affect short term, they negatively influence their interpersonal relationships and in the long run ironically undermine the self they are trying to build. The result is a chronic state of self-under-construction, which they relentlessly pursue through various social-cognitive-affective self-regulatory mechanisms in not always optimal ways. Our work has concentrated on illuminating the dynamics of these self-regulatory attempts while concurrently refining our self-regulatory process model of narcissism.
In our view, an appealing feature of this research is that it illustrates the viability of integrating within a unitary framework both dispositional (trait) and processing (social-cognitive-affective) approaches to personality that historically have been thought to be competing and mutually preemptive. The model connects narcissists' mental representations of self and their social worlds, through the strategic intra-and interpersonal self-regulatory behaviors and processes aimed at constructing and maintaining the narcissistic self. In this way, it addresses both stable characteristics of narcissistic individuals, as well as the psychological dynamics and processes that interact with the situation and underlie these characteristics. At the process level, narcissists are quick to perceive (or even impose) self-esteem implications in situations that leave room for it and then engage in characteristic social-cognitive-affective dynamic self-regulatory strategies to maintain self-worth. These underlying processes are reflected at the trait level, in terms of regular patterns of self-aggrandizing arrogant behavior, hostility, entitlement, and lack of empathy toward others. Thus, these trait-like differences in overall average levels of behaviors, cognitions, and affects are understood as a result of the operations of dynamic underlying self-regulatory processes. There is relative stability in the personality system because all processes are organized around central self-goals, yet also distinctiveness due to different situational features activating slightly different (albeit interconnected) aspects of the system. This integration of traits with process helps unravel the mystery of why narcissism is expressed through such a paradoxical set of traits.
In the first part of the article, we present our self-regulatory processing model of narcissism and the research we have undertaken to validate the model, as well as relevant research by others. In this context, we also discuss some of the recurrent problems and remaining unresolved issues. In the latter part of the article, we examine some broader implications of this model, its relation to other social processing models of personality, and some of the open issues. We conclude with a discussion regarding the utility of a model that can study dispositions and processes concurrently and within the same conceptual framework, thus integrating varying levels of analysis in the study of personality.

The Paradoxical Lives of Narcissists
Our initial interest in narcissism was piqued by narcissists' apparent insatiable pursuit of affirming self-knowledge through online manipulation of their social environment. This core feature of narcissism is contained in both the DSM definition and clinical characterizations. Recall that the DSM-IV (4th ed.; American Psychiatric Association, 1994) depicts narcissists as exhibiting pervasive patterns of grandiosity and self-importance, and as invested in demonstrating their superiority. Yet, despite the grandiosity, these individuals are also described as craving attention and admiration and as particularly concerned with how well they are doing and how favorably others regard them. Although on the surface this may appear paradoxical, upon further consideration, it is really not all that surprising that narcissists would have extremely positive, but simultaneously fragile self-views. The very fact that the narcissistic self is such a grandiose and bloated structure builds in an inherent vulnerability. It is a self that cannot stand on its own, as it is not grounded in an objective reality, thus it needs constant shoring up and reinforcement. It is the attainment of stable, positive self-views that narcissists seek through their self-regulatory endeavors and, as is addressed later in this article, they get what they seek if only fleetingly.
The same goal of a constant need for self-affirmation is also apparent in various clinical writings, which in addition provide some suggestions for its origin. In one way or another, they all in essence describe narcissists as individuals whose self-needs in childhood were not met due to deficiencies in early parental empathy or neglect, and who thus seek to fulfill these needs in their adult relationships. For example, Kernberg (1975) ascribed the disorder to a rejecting mother and the child's subsequent feelings of abandonment. Kohut (1971Kohut ( , 1972 pointed to inconsistent and capricious reinforcement, highly dependent on the mother's mood; and Millon (1981) blamed constant over-valuation that is not based on any objective reality. Thus, although the clinical theorists disagree about the exact etiology, they all see the origins of the fragile but grandiose self as a response to unempathetic and inconsistent early childhood interactions. Moreover, they suggest that narcissists attempt to fill the void left in childhood in their adult relationships. It seems that there may be two aspects to filling this void, both of which contribute to narcissists' quest of a grandiose self. The first is perhaps more affective and involves seeking reassurance to allay a gnawing concern of inadequacy. The second may be a more cognitive concern directed at completing self-definitional needs. Incorporating both of these components, our focus is on the repeated self-regulatory thoughts, feelings, and behaviors aimed at obtaining validation for the grandiose self.
These processes are particularly interesting, because ultimately the grandiose self is an impossible goal, as narcissists encounter the reality of failures and social disconfirmations from others who do not always share narcissists' high opinion of themselves. In addition, even when narcissists manage to orchestrate confirmations, this often occurs by their heavy-handed manipulations of others to bring about the desired responses, or by distortions of those responses that are not adequately self-aggrandizing. Furthermore, pre-sumably because of narcissists' negative experiences with their significant caregivers in early childhood, their later relationships are often imbued with hostility and mistrust. As a result, contingencies between social feedback and self-inferences most always remain somewhat ambiguous, thus continuously introducing doubts (at some level) about the validity of the social feedback. This, in turn, endlessly feeds the need for more online social construction: Narcissists must continuously "ask" others whether they hold admiring opinions of the narcissists. Toward this end, they incessantly keep squeezing their relationships for the feedback they desire. However, not only are narcissists mistrusting of others due to their early negative experiences, they also do not really like or care for them and often even disdain them. This tendentious relationship between narcissists and their "social self-support networks" is reflected in the wide range of interpersonal deficits portrayed in the DSM: their lack of empathy, feelings of entitlement, and their exploitative and arrogant behaviors. This combination of an adversarial interpersonal orientation and insensitivity to others' needs and desires, contributes to narcissists engaging in interpersonal strategies that are often counterproductive and ultimately inhibit their social networks from providing the positive feedback that they seek. This may be the ultimate "narcissistic paradox": as they yearn and reach for self-affirmation, they destroy the very relationships on which they are dependent.
In sum, from both the clinical and the personality and social literatures one derives the picture of an individual who is overly dependent on social sources for affirmation of a grandiose, yet vulnerable self. In fact, one gets the sense that much of narcissists' daily action is geared toward obtaining, even creating such positive feedback, to which they then in turn respond with more intense emotions than others. Furthermore, others are somehow not real to them and not important in their own right. For narcissists, others are only of value in terms of what they can provide to help bolster the narcissistic self. However, because narcissists have little empathy for others and no genuine concern with what others really think, this often seems to translate into paradoxical, counterproductive behaviors that ultimately prevent others from responding in the way narcissists desire. These two defining features make narcissism an ideal prototype for studying online social construction of the self: Narcissists are chronically "working on" the (grandiose) self and see others primarily as a source of self-enhancement and as narcissistic supplies. Because their self-regulation attempts often go awry, this chronic process of buttressing and building the self is potentially endless labor, which has proven useful in illuminating the workings of these processes. The narcissistic self is perpetually "under construction", as if the construction site were on quicksand.

A Process Model of Narcissism
Our goal has been to understand the paradoxical elements of narcissism by viewing their function in the process of self-construction and maintenance. The model depicted in Figure 1 is the framework we have developed to capture these elements and process relationships. (An earlier version of Figure 1 can be found in Rhodewalt, 2001.) It assumes that narcissists have certain identity goals that they pursue with more or less success through their social interactions. The main focus of the model is on the inter-and intrapersonal dynamic self-regulatory processes through which narcissists actively (although not necessarily consciously) operate on their social environments to create and maintain their self-knowledge. Thus, at the theoretical level, our approach shares much in common with other social-cognitive dynamic processing models (e.g., Cantor & Kihlstrom, 1987, 1989Dweck & Leggett, 1988;Higgins, 1987;Mischel & Shoda, 1995;Schlenker, 1985;Swann, 1985). These models all have at their core the notion that people are active in structuring their social environments to bring them in line with their goals, rather than just passively reacting to these situations. In this vein, our approach shares a focus on what individuals do-behaviorally, cognitively, or affectively to bring particular situations and tasks (that they confront, select, or create) in line with these goals. Furthermore, although some models tend to place more emphasis on social and others on intrapersonal processes, all embrace an explicit attempt to understand the interplay of the cognitive, affective, social units and their (joint) mediation of an individual's behavior. Within a particular person or personality type, these units are thought to be organized into relatively stable configurations. Dynamic self-regulation, then, is understood in terms of this system of person units interacting with situational demands and affordances in the pursuit of goals.
In our model, we employ a somewhat more specified definition of self-regulation that focuses specifically on instances when the individual is regulating contents of the self that define that individual's identity (rather than global self-evaluation or various attempts at self-control). In this sense, it shares close resemblance to Schlenker's work on self-identification, which delineates the processes and means through which individuals fix and express their identities for themselves or others through social interaction (Schlenker, 1985;Schlenker & Wiegold, 1992). Importantly, self-identifications are not simply faithful expressions or retrievals of the self-concept but rather are constructed at the time they occur in a dynamic transaction between the individual and the social context. Our use of self-regulation encompasses these strategic interpersonal attempts of individuals to bring about their desired identities. These interpersonal processes occur at the level of actual social behavior, in which narcissists strategically interact with their social worlds to construct and regulate their desired selves. For the narcissist, social interactions are the settings for the enactment of social manipulations and self-presentations designed to engineer positive feedback or blunt negative feedback about the self.
However, in addition, we also include in self-regulation the intrapersonal adjustments and fine-tuning that occurs when these self-identifications are not perfectly achieved. Thus, these intrapersonal processes are the cognitive, affective and self-evaluatory activities that underlie, or motivate this behavior, or are a response to the results of this behavior. Among these are biased interpretations of social feedback and performance outcomes (to regulate self-esteem); selective attention to particular features of tasks and social environments, as well as selective or distorted recall of past outcomes or events.
In our view, these intra-and interpersonal self-regulatory processes are the essence of personality, in that they give a distinctive form to the self's underlying mental system (including cognitions, emotions, needs, and motives). It should be noted that the boundary be-tween the two is fuzzy and indeed fluid, because interpersonal maneuvers are often enacted to serve intrapersonal needs and as is shown in later sections of this article, many of the intrapersonal strategies have interpersonal consequences. Moreover, the interpersonal transactions are continuously encoded, interpreted, and evaluated internally (intrapersonally), triggering a cascade of cognitive-affective responses and further self-regulatory scripts that ultimately are played out interpersonally. In other words, consistent with other contemporary cognitive-affective processing models of personality (e.g., Mischel & Shoda's CAPS model, 1995), intra-and interpersonal self-regulation involves reciprocal interaction. The narcissistic self obtains its being through these dynamic intra-and interpersonal transactions that link the narcissists' self-knowledge systems to their social relationships.
The self-knowledge component both drives and is a result of these self-regulatory processes. It represents a summary statement of the narcissist's current view of self and its social context. This includes both the cognitive self, as well as a valence statement. The cognitive self entails mental representation of the actual self (self-ascribed traits and competencies), reflected appraisals, as well as possible future selves, ideals, and goals. The valence aspect reflects one's general sense of value but also captures momentary state self-esteem. As previously discussed, the content of the narcissistic self tends to be overly grandiose, yet simultaneously vulnerable and fragile. It appears they are unable to convince themselves of their presumed grandiosity, hence the fragility, reflected in transient fluctuations in (state) self-esteem in response to external happenings. Thus, narcissists' self-esteem is high or low depending on preceding events, but these oscillations are deviations from their average self-esteem, an average that is high relative to others. Equally or more important than content in trying to address the nature of narcissistic vulnerability, though, may be the structural organization of self-knowledge. We will discuss some research we have conducted examining the suggestion that narcissists may possess self-concepts that are simplistically structured (Emmons, 1987;Kernberg, 1980).
The social relationships component reflects the larger social context within which these self-regulatory processes are played out. These relationships are affected by narcissists' strategic maneuverings aimed at shoring up the self and, in turn, have an effect on narcissists' own behaviors, their self-evaluations, and their self-knowledge. As already implied, narcissists likely prefer relationships with people who offer the potential for enhancing the narcissists' self-esteem and sustaining their inflated self-image but likely have trouble maintaining relationships as soon as the other becomes a real (i.e., imperfect, even flawed) person to them (W. K. Campbell, 1999).
In short, this model depicts the narcissistic self as shaped by the interplay of dynamic self-processes and the larger social system within which it functions. As we will show, the coherent narcissistic dynamic is a chronic goal orientation aimed at getting continuous self-affirmation, while being relatively insensitive to social constraints, especially when the self is threatened. This dynamic is in part the result of narcissists' underlying self-conceptions (grandiose, yet fragile) and their view of others (inferior), which both in turn are maintained via various social-cognitive-affective self-regulatory mechanisms. We now turn to the research we have conducted in an effort to provide support for this model. In line with Cronbach and Meehl (1955), we took as a starting point what we believe to be the key characteristic of the narcissistic dynamic, namely the goal of constantly receiving online self-affirmation of the grandiose self-images, and proceeded to uncover the conditions under which it occurs. Building an interconnected system of such functional relations contributes to the validation of our self-regulatory process model of narcissism in form of a nomological net.

Interpersonal Self-Regulation
Clinical and object-relations theory and description make clear that various aspects of narcissism should lead narcissists to use interpersonal relations as a primary means through which to transact self-regulation and bolster the self (see Reich, 1960 for an early discussion). Because of their deficient early interactions, narcissists never completed their self-definitional work and thus try to make up for this in their adult relationships. We begin our discussion on research, with the interpersonal aspects because it is here-in the interpersonal arena-that the dynamics of the narcissist become most visible and open to systematic study.
In a first attempt to capture interpersonal self-esteem regulation, we (Morf & Rhodewalt, 1993) examined the effects of a threat to the narcissistic self from being outperformed by another person on a task that was relevant to the narcissist's self-definition. Our interest was whether narcissists tried to reduce this social comparison threat and boost themselves by devaluing or derogating the better performing other on another dimension. Based on narcissists' sense of entitlement and lack of concern for others, we expected that they would engage in efforts to maintain a positive self-image at all cost, even if this came at the expense of the other or the relationship. As predicted, and consistent with Tesser's (1988) self-evaluation maintenance model, we found that in an effort to restore self-esteem, threatened (relative to nonthreatened) narcissists re-acted by reporting significantly more negative views of the other's personality. Importantly, they did this even if they expected to have to provide these evaluations in a face-to-face interaction. This finding is consistent with the notion that narcissists exploit and use others to increase their self-worth, with little regard for others' feelings or the interpersonal conflict the narcissists may be creating.
These findings have since been replicated in an independent laboratory by South and Oltmann (1999). It is noteworthy that these researchers used a different self-report measure of narcissism that they collected between 1 and 2 years before the study. Using the narcissism subscale of the Clark Schedule for Nonadaptive and Adaptive Personality (SNAP; Clark, 1993), they report the same pattern of other derogation as found in Morf and Rhodewalt (1993). Interestingly, of the 18 other personality traits also identified by the SNAP, other derogation additionally correlated with the traits of manipulativeness, aggression, and entitlement.
Further corroborating evidence that narcissists prefer interpersonal over other means of self-regulation comes from a study conducted with a somewhat different paradigm from those just mentioned. Smalley and Stake (1996) found that individual differences in self-esteem predicted derogation of a nonperson feedback source (negative ratings of the test), following bogus negative feedback, but that only narcissism was associated with derogation of, and hostility toward, the person providing the evaluative feedback.
We turned, then, to investigate other potential interpersonal strategies employed by narcissists in the service of self-esteem maintenance. One line of research explored which self-presentational tactics narcissists would use if the goal was to get someone to like them during a conversation (Morf, 1994). We found that narcissists had a pervasive preference for self-aggrandizing statements, rather than self-effacement or social approval-seeking. For example, they chose to use statements, such as "people look up to me, because I always know the right thing to do"; rather than, "sometimes I get embarrassed, when I make a mistake." This same study also showed that independent coders who rated the audiotaped conversations had significantly more negative impressions of high than of low narcissists. Thus, it appears that when narcissists have to choose between being liked or admired, they go for admiration. This latter point was also clearly captured in a study by Emmons (1989) showing that narcissism was associated with high power and low intimacy strivings. The content of narcissists' personal strivings indicated that they were not particularly interested in establishing and maintaining warm interpersonal relations, but that they were interested in having impact on and influence over others. Themes of domination and exploitation were prevalent in high narcissists, which was in stark contrast to lows, who seemed concerned with minimizing interpersonal conflict and with getting along and helping others.
What is puzzling in all this, is that if narcissists seek attention and admiration from others as their central goal as suggested by the DSM, then one would expect them to present themselves in ways that earn maximal social approval. That is, it seems that even if narcissists do not care about other people, they should be able to use their social environment more strategically. They should be able to pursue respect in one situation, yet social approval in others, depending on what is most appropriate and what secures them the most benefit. The fact, that they do not seem to make these distinctions, leads one to wonder about the exact role of the audience. Morf, Ansara, and Shia (2001) addressed this question by manipulating the situational requirements unique to approval seeking or strategic impression management. Specifically, this study examined self-presentational behavior of high and low narcissists about to undergo an interaction with someone who was likely to become aware of one of the self-presenter's negative attributes. Strategic impression management requires modesty with regard to that attribute. However, the prediction was that narcissists would present the grandiose self regardless, because they would be more concerned with self-construction than with social approval. As expected, everyone enhanced on the attribute in question when they were not constrained by negative feedback, or when they were not likely to be found out. When the negative quality was likely to be discovered, however, high narcissistic males engaged in the usual inflated self-presentations. In contrast, low narcissists exhibited the expected modesty effect.
These findings indicate that narcissists are not particularly concerned with social approval but rather are invested in constructing and conveying a grandiose self. They are less sensitive to the requirements of the social situation and probably misunderstand how they are perceived. In a sense, although they appear to need the social environment to acknowledge their self-presentational efforts, their orientation is almost pseudo-social, because there is no genuine concern with what the audience really thinks. This implies that narcissists may engage in social interaction not primarily to manage strategically the impressions they convey to others, but rather to deceive the self into seeing its own grandiosity. Consistent with this conjecture, Paulhus (1998) found a high correlation between the NPI and the Self-Deceptive Enhancement component of his Balanced Inventory of Desirable Responding. Self-Deceptive Enhancement gauges overly positive self-evaluations that the respondent actually believes. However, the NPI had only a moderate correlation with the Impression Management subscale that taps self-enhancement efforts aimed at an audience.
As further evidence to this point, Rhodewalt, Tragakis, and Finnerty (2001) showed that narcissists engage in self-handicapping behavior more routinely than low narcissists and that this was even more true when the handicap was private than when it was public. Self-handicaps are impediments erected by the individual prior to performance, when the individual lacks confidence regarding the likely outcome. These handicaps allow for discounting of subsequent failure and potential augmentation of success. The primary motivation for this may be to protect one's public image or to regulate self-esteem. The fact that narcissists' self-handicapping behavior was greatest in private, when no one else knew about it, indicates that this behavior was performed more for self-deceptive purposes than for public impression management.
In sum, there is accumulating empirical support that narcissists use their social interactions to construct and maintain a grandiose self. These studies also show that narcissists are more concerned with garnering admiration from, and impressing and having an impact on others, than obtaining social approval or even real social feedback. For example, they derogate a better performing other to his face, they self-handicap prior to performance, and they engage in grandiose self-presentations in situations that call for modesty. It is clear that these behaviors contribute to their interpersonal difficulties. From other studies, we know that observers do not like self-aggrandizing (Schlenker & Leary, 1982) or self-handicapping targets (Rhodewalt, Sanbonmatsu, Feick, Tschanz, & Waller, 1995). Thus, these studies illuminate some of the ways by which narcissistic concerns translate into the disturbed interpersonal relationships characteristic of the narcissistic syndrome.

Intrapersonal Self-Regulation
As with the interpersonal processes, many of the characteristic intrapersonal mechanisms are aimed at self-esteem regulation. We designate as intrapersonal all cognitive, affective, and motivational self-regulatory processes that take place primarily inside the mind of the narcissists, as opposed to being transacted in interpersonal interactions. However, as noted when introducing the model, the boundary between the two is fuzzy and indeed fluid, as intra-and interpersonal processes are intertwined and reciprocally related.
To begin at the phenomenological level, it appears that narcissists experience both high and low self-esteem in alternation depending on external occurrences. This is based on clinical accounts emphasizing that narcissists' manifestly grandiose self-concepts masks an underlying, deeper sense of worthlessness and inferiority (for a review see Akhtar & Thompson, 1982). In confirmation of this duality, an investigation by Raskin et al. (1991a) found that narcissists' high reported self-esteem reflected both genuine and defensive components. Moreover, defensive self-esteem among narcissists took the form of defensive self-enhancement rather than the need for approval from others. Thus, although many studies have shown narcissism to correlate with high self-reported self-esteem, this is not particularly informative, as both genuine and defensive components are being measured by these self-esteem scales (see Baumeister, Tice, & Hutton, 1989). More interesting are studies that observe self-esteem under conditions that unsettle the narcissists' defended veneer and thus pick up the instability and fluctuations one would expect from the coexistence of grandiosity and vulnerability.
For example, Rhodewalt and Morf (1998) provided high and low narcissists successive success and failure feedback on alleged IQ tests. As expected, narcissists were more reactive to this feedback: Relative to nonnarcissists, narcissists showed significantly greater decreases in self-esteem, if success was followed by failure and greater increases if failure was followed by success. Then, in a series of daily diary studies, Rhodewalt and colleagues (Rhodewalt, Madrian, & Cheney, 1998;Rhodewalt, Tragakis, & Hunh, 2001) had high and low NPI participants provide daily descriptions of events and state self-esteem across a number of days. Narcissists not only displayed greater overall day-to-day fluctuations in their self-esteem than less narcissistic individuals, their self-esteem was also more related to the quality of their social interactions than it was for less narcissistic individuals. In particular, narcissists' daily self-esteem was more highly correlated with the extent to which the day's social interactions were positive or negative, the extent to which the interactions made them feel like themselves, and perhaps surprisingly, the extent to which they felt accepted by the audience. Although not directly examined, we speculate that "acceptance" to narcissists here meant affirmation rather than approval. The main point is that these findings further corroborate the claim that narcissists are over-reliant on social sources of evaluation.
When one's self-esteem is so closely tied to social feedback, one also expects greater mood fluctuations in general, as corresponding findings by Kernis and his colleagues show for individuals who have high but unstable self-esteem (e.g., Kernis, Cornell, Sun, Berry, & Harlow, 1993). Unstable high self-esteem individuals are especially sensitive to social feedback, react to it with more extreme emotions, and find ways of attenuating the impact of negative feedback. Similarly, clinical theorists have long noted extreme emotional reactivity as a central element of the narcissistic personality (e.g., Kernberg, 1975;Kohut, 1972); and Emmons (1987) found some correlational support for this narcissistic affective extremity and variability in daily mood reports.
In the earlier mentioned study by Rhodewalt and Morf (1998), we observed this narcissistic emotional reactivity, though not globally, but specifically for self-esteem and anger. In addition, we ascertained that this reactivity was mediated by attributional processes. High narcissists, relative to lows, had a much greater propensity to attribute an initial success to ability, which in turn led to more extreme anger and greater decreases in self-esteem in response to subsequent failure. Thus, it appears that narcissists give themselves a self-esteem boost by ascribing positive outcomes to their internal, stable, and global qualities, thus taking greater credit for success. Unfortunately, this backfires when they cannot sustain success and thus, attributional self-aggrandizement contributes to overall lability. This narcissistic tendency toward making self-aggrandizing attributions had also been found in an earlier study (Rhodewalt & Morf, 1995). The focus appears to be on self-aggrandizement rather than excessive self-protection, because in neither study did we find evidence for greater discounting of negative events by narcissists. In both studies, narcissists externalized bad outcomes to the same extent as did others. Thus, as in the interpersonal realm, there is evidence that narcissists are more focused on assertive self-promoting behavior, at the risk of greater loss or threat in the event of failure or rejection.
Narcissists support their grandiose selves not only by taking credit for positive outcomes, but also simply by viewing themselves and their accomplishments as superior to others. John and Robins (1994), for example, had participants engage in a group-discussion task and subsequently rate the relative performance of all members in the group, including their own. Narcissists significantly overestimated their own contribution relative to other group-member ratings, as well as to ratings made by independent judges (see Gosling, John, Craik, & Robins, 1998, for a replication). Other studies have shown that narcissists overestimate their own intelligence and general attractiveness (Gabriel, Critelli, & Ee, 1994), and their attractiveness to others (Rhodewalt & Eddings, 2001); they overestimate their final course grades (Farwell & Wohlwend-Lloyd, 1998), and exaggerate their positive personality characteristics (Paulhus, 1998). There is also evidence that narcissists are likely to gravitate toward tasks that support their desire to demonstrate superiority over others. Narcissists were found to persist longer and enjoy more a task that was framed in a way to involve interpersonal competition and doing better than others, as opposed to one framed as a self-referent game, done just for fun (Morf, Weir, & Davidov, 2000). Thus, they are likely to seek out evaluative contexts that offer opportunities to demonstrate competence relative to others, and when unconstrained by explicit performance or ability indicators, narcissists self-enhance by claiming superiority.
When directly confronted with failure, however, narcissists find ways of undoing it. They respond to negative feedback, for example, by derogating the evaluator or the evaluation technique (Kernis & Sun, 1994;Morf & Rhodewalt, 1993;Smalley & Stake, 1996). Alternatively, they might even distort and restructure past events to soften the blow. In a particularly interesting study (Rhodewalt & Eddings, 2001), narcissists were led to experience romantic rejection, upon which they recalled personal romantic histories that were more self-aggrandizing than the histories they had reported on an earlier occasion. Furthermore, the more narcissists distorted their recall, the more their self-esteem was buffered from the rejection. This was in contrast to less narcissistic individuals, for whom rejection led to recall of a more humble past and lower self-esteem.
Before leaving this topic, it may be worth a brief sidestep to note that contrary to what the reader may think, narcissists generally do not appear to think that negative feedback is a priori less accurate. In experiments that have manipulated the positivity of the feedback, manipulation checks have shown that narcissists accept their negative performance score as accurate, but then post hoc find ways of lessening its impact (e.g., Kernis & Sun, 1994;Morf et al., 2001).
To summarize, narcissists find endlessly inventive means of casting feedback and social information in ways to reinforce their grandiose self: They view their personal attributes and accomplishments as superior, they make self-aggrandizing attributions to augment positive feedback, they restructure their past to be more favorable, and they derogate the source and validity of negative feedback. However, the fact that they keep looking for more self-validation suggests that they fail to convince themselves of their own adequacy.
These intrapersonal processes contribute to deleterious interpersonal processes in at least three ways. First, in its most benign form, the overly positive view of self promotes a negative view and disdain for other people. Second, because narcissists seek evaluative feedback, they constantly interpret tasks as being opportunities to compete with others and to demonstrate their superiority. Finally, in the extreme, when their mental constructions do not prevent them from failure, they are prone to anger (Rhodewalt & Morf, 1998) and even interpersonal aggression (Bushman & Baumeister, 1998). Although there is suggestive evidence for these postulated connections between intraand interpersonal processes, to date, research directly examining links between internal events and behavior has been sparse. More work is needed to connect narcissists' internal representations, maneuvers, and distorting processes to interpersonal behavior and its consequences.
These findings also provide some insight into the phenomenology of narcissists. They seem to experi-ence their world as a place in which they must continually seek (or are called upon) to prove their self-worth. Their emotional responses fluctuate greatly, as they are organized around a positive but fragile sense of self-worth. Finally, as a result of their dependency on external validation, their self-images are likely highly context dependent. We now turn to research on the narcissists' self-concept, within which this general instability of the self should be reflected in some form in its structure, valence, or both.

Narcissistic Self-Concept
As we noted, research by ourselves and others clearly indicates that the self-concepts of narcissists are extremely positive and grandiose. However, the fact that they appear to be simultaneously fragile and unstable has led us to address questions about the representation of narcissistic self-knowledge to help us better understand the nature of the vulnerability. If at its core narcissism is a "cognitive-affective preoccupation with the self" (Westen, 1990, p. 227), it is ironic that the clinical literature portrays the narcissistic self as "empty" or else lacking in coherence. More specific, if as clinical theorists contend, narcissists' preoccupation with self is the consequence of disturbed early relationships and self-development, their self-concepts are expected to differ from others in theoretically meaningful ways. Moreover, these features of the self should relate to narcissists' self-esteem instability and the narcissistic self-regulatory processes described earlier.
Following Bach's (1977) observation that narcissism reflects a deficit in the perception of self and Emmons's (1987) suggestion that narcissists may suffer from the consequences of having self-representations that are low in complexity, we have investigated the narcissistic self-concept from a cognitive perspective. This work can be organized around two questions: 1. Do narcissists differ from others in that their self-concepts are less clearly formed and accessible-what has been termed the deficit model? 2. Are the self-conceptions of narcissists organized differently from others-what has been termed the structural model? (Rhodewalt, 2001) The deficit approach postulates that narcissists' self-conceptions are poorly formed, unstable, not as automatically accessible, and, as a consequence, not as confidently held as they are in less narcissistic individuals. Tschanz and Rhodewalt (2001) explored the accessibility question by measuring response latencies to trait self-descriptiveness judgments some of which were "primed" by recall of past behaviors or social rep-utations and some were unprimed. Narcissists were neither faster nor slower than less narcissistic individuals to make these judgments nor were they differentially more or less responsive to the primes. In other words, this study found no evidence for enhanced or impaired accessibility of self-knowledge.
We have also attempted to assess the clarity of and confidence with which narcissists hold their self-views. Two studies found no evidence that narcissists report less self-concept clarity (J. D. Campbell, 1990) than do nonnarcissists (Rhodewalt & Regalado, 1996;Rhodewalt, Tragakis, & Hunh, 2001). Moreover, we have assessed self-concept content and confidence in a variety of ways and consistently find that narcissists report more positive and more confidently held self-views than do others (Rhodewalt & Morf, 1995).
In short, we have been unable to detect evidence that the narcissistic self-concept is empty, inaccessible, or held in low confidence. If anything, narcissists tell us that they have very high opinions of their abilities and traits, and that they are very certain about those opinions although their behavior may at times indicates otherwise.
The structural model contends that narcissists differ from others not in the content or accessibility of self-knowledge but rather they differ in how knowledge is organized. For instance, if narcissists should be low in self-complexity, as has been suggested by Emmons (1987), this accounts for their emotional hyper-responsiveness to self-relevant feedback. This reasoning follows from the work of Linville (1985), Showers (1992), and others who have connected features of self-knowledge organization to emotional lability.
We have examined directly relations among NPI-defined narcissism, self-complexity (Linville, 1985), evaluative integration (Showers, 1992), and emotional reactivity. Self-complexity describes the extent to which aspects of one's self-concept are differentiated. Moreover, those with complex self-representations display relatively stable moods compared to those with simplistic representations (Linville, 1985). Evaluative integration (Showers, 1992) describes the extent to which self-knowledge is compartmentalized along positive and negative evaluative dimensions. We have obtained very mixed and inconclusive data regarding narcissism and the organization of self-knowledge. Although Rhodewalt and Morf (1995, Study 1) found the predicted high NPI/low self-complexity association, five other independent investigations have failed to replicate this association Rhodewalt & Morf, 1998;Rhodewalt & Regalado, 1996). Likewise, other studies have failed to find an association between NPI-defined narcissism and low evaluative integration in three independent samples Rhodewalt & Regalado, 1996). However, evaluative integration but not self-complexity was found to moderate the relation between narcissism and emotional responsiveness to social feedback . Narcissists who were also low in evaluative integration, that is, had highly compartmentalized self-concepts, displayed the greatest day-to-day self-esteem instability.
The self-complexity perspective specifies that greater complexity is beneficial. In contrast, Donahue, Robins, Roberts, and John (1993), building on the earlier work of Block (1961), found that self-concept differentiation-the tendency to see oneself as possessing different traits in different social roles-is related to greater intra-and interpersonal distress. Rhodewalt and Regalado (1996) have some preliminary evidence that narcissism is related to greater self-concept differentiation. If this finding replicates and is subsequently linked to narcissistic emotional and self-esteem instability, it would suggest that narcissists possess selves that are highly differentiated along roles or social contexts, a fact that may explain the apparent incoherence of the narcissistic self-concept. In sum, although there appears to be no evidence for the deficit model, there is some tentative preliminary evidence that narcissistic selves may be disorganized. This is consistent with the clinical notion that narcissists may have fragmented selves.
Before leaving our discussion of the narcissistic self-concept, we wish to speculate on one additional way in which the self-knowledge of these individuals may be distorted and perhaps more dependent on external social validation. As we have already noted, narcissists possess what we have termed a self-aggrandizing attributional style (Rhodewalt & Morf, 1995Rhodewalt, Tragakis, & Finnerty, 2001; see also Emmons, 1987;John & Robins, 1994). That is, they take more credit for positive outcomes than do less narcissistic individuals while externalizing negative outcomes (although no more so than do less narcissistic people). One consequence of this tendency is that narcissists should have rapidly changing views of self, others, and social context as enhancements and threats to the self come down the road. Kernberg (1975) contends that narcissists frequently engage in "splitting" or dramatic shifts in self-evaluation and, thereby avoid the conflicts of dealing with one's strengths and weaknesses. Thus, the narcissists' ongoing, shifting, self-serving attributional analysis may appear as "splitting," indicating to the clinician an incoherent sense of self. Also, as mentioned previously, the narcissist's attributional hubris sets the stage so that threatening feedback results in greater impact on feelings of self-worth and anger than it does on these feelings in less narcissistic individuals. It may be that the narcissistic propensity to interpret any positive feedback as self-defining leads to highly context-bound self-images that, in turn, contribute to an overall lack of integration and coherence of the self.

Narcissists' Interpersonal Relationships
Much of narcissistic identity goal-striving is transacted in the social arena. As we saw in the preceding sections, many of narcissists' self-regulatory strategies either directly or indirectly involve other people, and thus, should be expected to structure and affect their interpersonal relationships. Clinical accounts (such as Kernberg & Kohut's) describe narcissism as marked by poor interpersonal functioning and a lack of ability to experience true intimacy. This comes as no surprise, if narcissists are willing to self-enhance at the cost of others and simultaneously are continually vying for others' admiration. However, there is little direct empirical evidence, as of yet, that speaks to the nature of narcissists' significant relationships or their unfolding over time. In one of the few such studies, W. K. Campbell (1999) demonstrated that narcissists are attracted to individuals who have highly positive and valued qualities, particularly when these individuals are also admiring, and narcissists are relatively less attracted to caring individuals. Mediation analysis revealed that this preference reflected a self-enhancement strategy. Apparently, these preferred individuals were seen as a source of self-esteem, because they provided the narcissists with a sense of popularity and importance. Implications of such choices are far-reaching. For example, narcissists are likely to become disenchanted with and respond negatively to their relationship partners when flaws become apparent, and they are revealed to be less than perfect. They are also likely to have relationships that are short lived, if they disengage from such relationships when the source no longer provides the self-esteem benefits, for example, due to losing a prestigious job or becoming chronically ill.
In addition, to affecting the types of affiliations narcissists seek and when they discontinue them, narcissists' heightened concern with self-affirmation also is expected to impact their relationships, through their behaviors, cognitions, and emotions during interactions. In a study by Buss and Chiodo (1991), individuals who described narcissistic acquaintances reported that the narcissists act in ways to impress others, such as bragging about accomplishments, showing off money and possessions, as well as insulting others' intelligence and putting them down. Although perceptions and reactions to these behaviors were not assessed in that study, these are clearly not endearing behaviors, as has been verified by Colvin, Block, and Funder (1995) in other research not involving narcissism.
In a particularly interesting study, Paulhus (1998) directly tested the impact narcissists have on their interaction partners. He found that people viewed narcissists especially favorably on an initial encounter, but these impressions became reversed over repeated interactions. At the first meeting, narcissists were rated agreeable, competent, intelligent, confident, and entertaining, but by the seventh interaction, they were seen as arrogant, overestimating their abilities, tending to brag, and hostile. Thus, it appears that although narcissistic self-regulatory strategies lead to attention and admiration in the short run, they lead to rejection and interpersonal failure in the long run.
Narcissists' own perceptions of their relationships also attest to greater instability. Narcissists, not surprisingly, view themselves as being quite "successful" in their past relationships. Rhodewalt and Eddings (2001) found that compared to less narcissistic men, narcissists relate histories of finding it easy to meet women, have women attracted to them, and having women be receptive to their invitations to date. At the same time, narcissists report having had a greater number of serious relationships and more frequently dating more than one woman at a time than did less narcissistic men. These later results suggest greater instability in the romantic relationships of narcissists compared to low narcissists. Rhodewalt and Shimoda (2000) included the NPI and a lengthy questionnaire concerning narcissists' most serious romantic relationship in a replication of Hazan and Shaver's (1987) love quiz study. Narcissists reported experiencing greater emotional extremes, jealousy, obsession, and sexual attraction, than did low narcissists. Thus, even, by their own acknowledgment, narcissists characterize their relationships (particularly romantic ones) by emotional turmoil and instability.
Research is still needed to elucidate more fully the interpersonal component of the model. There is much to understand about the relatively enduring patterns of dynamic interplay between narcissists' self-regulatory behavior and their significant interpersonal relationships. There are many interesting possible avenues to pursue, involving not only narcissists' negative and hostile attitudes toward others, but also their general lack of empathy and responsivity to others' needs or viewpoints. For example, Westen (1990) described narcissistic patients engaging in an egocentric communication style, in which they fail to adapt the communication to the listener's perspective and carry on a kind of "collective monologue." This implies there may be deficits in relationship-maintenance mechanisms, such as perspective and role taking, empathic accuracy, and accommodation processes-the ability to inhibit the impulse to respond in kind to a destructive act by one's partner and react in a constructive manner instead (e.g., Rusbult, Verette, Whitney, Slovik, & Lipkus, 1991). For all of these interaction patterns, it will be particularly important to examine how they unfold and transform over time, in order to understand the role of significant relationships in the social construction (or deconstruction) of narcissists' self-concepts.

The Narcissism Model in Perspective:
Unraveling the Paradoxes Thus far, we have laid out the model and the research relevant to it. The model's focus is on the trans-actions between narcissists' self-concerns, goals, and motives, and their interpersonal context. Clearly there is ample evidence that narcissists' construct and maintain their selves in interaction with their social worlds. Importantly, although narcissists in their efforts to gain stable positive views of the self, create social domains that are in part self-sustaining, they operate on their social environments in ways that often may be ultimately deleterious to this goal. Thus, consistent with the Cantor and Kihlstrom's (1987Kihlstrom's ( , 1989; see also, Cantor, 1990Cantor, , 1994 problem-solving perspective of personality and social behavior, we view the social behavior patterns that characterize narcissists as a form of social intelligence, which is effective and adaptive in some ways and contexts, albeit self-defeating in others, as we elaborate in the following section.

Narcissism as a Form of Social Intelligence
The social intelligence framework specifies that regularities in social behavior are connected to people's problem-solving efforts in everyday life. Individuals' personalities are defined by their central problems, which are rooted in the persons' self-concept and autobiographical knowledge, and by the means and processes through which they choose and shape situations to provide strategic "solutions" to these personal problems. As in other processing models, the idea is inherent that various situations provide different affordances for particular problems to be pursued and that individuals' personalities are characterized by differential responsiveness to the particular affordances. In other words, the psychological or functional significance of situations depends on the needs and goals of the individual. Thus, the problems, responses, strategies, and solutions are all joint constructions of the person and the situation. Individuals are socially intelligent to the extent that they make optimal use of situational affordances to move toward their personal goals.
From this functional perspective that is sensitive to the individuals' idiosyncratic life problems, one can make sense of much of narcissists' behaviors, even if on the surface they look paradoxical because they are frequently associated with costs and often counterproductive. If narcissists' core concern is to foster their grandiose self-view, it is rational as a first step to try to preempt an ego threat, if at all possible. In this effort, they approach situations with high perceived control and the expectation of doing well (e.g., Farwell & Wohlwend-Lloyd, 1998). Presumably, this allows them to focus on and persist at the task; and in interpersonal tasks, their self-assured behavior likely earns them respect, at least on first encounter (Paulhus, 1998). They also manipulate their contexts to allow them to capitalize on positive events, for example, by making ability attributions for success or self-handicapping prior to performance. When an ego-threatening event occurs, they engage in cognitive reorganization of information in line with their self-schema to restore self-esteem (e.g., reframing failure as someone else's or the tasks' fault, or selectively retrieving favorable "facts"). Another potential, although as of yet untested, advantage of their self-assertive strategies may be a gradual or at least temporary phenomenological carryover to their identity.
It is important to note here that as in the other processing models, we conceive of this type of self-regulation, despite its strategic (i.e., goal-directed) nature, as largely automatic and nonreflective, triggered automatically by situational cues-situational cues that have acquired their meaning from past experiences with similar features. This seems especially likely in the case of narcissism which appears to be a disposition that is affect based because of its self-esteem maintenance driven nature. Due to this deep-seated need, the narcissists' self-system is in a chronically vigilant state to detect opportunities for self-enhancement or potential departures from self-affirmation, in response to which the system automatically mobilizes its strategies. This activation then spreads through the processing units of the self-system along previously learned connections, much like in a connectionist model in cognitive psychology, necessitating no conscious, reflective involvement (see Mischel & Shoda, 1995). Affect in a sense is the self's alarm system that sets all self-regulatory behaviors in motion.
Overall, it appears that narcissists' self-aggrandizing strategies serve their ego enhancement and maintenance pursuits reasonably well. Even at the clinical level, Kernberg (1975) emphasized the adaptive nature of narcissists' power pursuit, noting that many highly successful people have the disorder. However, the fact that they strategically target one problem domain does not preclude their incurring costs in other domains. Their indiscriminant self-promotion is likely to generate unrealistically high performance expectations, which they might have trouble living up to. Turning every event, even those that are otherwise fun and relaxing, into competitions and opportunities for self-promotion can engender constant stress and performance apprehension. However, most important, their self-aggrandizing behavior will take a toll on their interpersonal relations. By acting in an egotistic and arrogant manner, they alienate their friends and acquaintances and incur negative social sanctions. Furthermore, their tendency to assertively promote the self interferes with their ability to empathize and see the other's point of view, thus severely impairing their intimate relationships.
These deficits in the interpersonal domain may render many of their strategies ultimately self-defeating, if to preserve their positive self-views, they depend on consistent social affirmation. As with many other problems in self-regulation, the source of the poor choice in social strategies is likely to be found in an unconscious trade-off of short-term benefits for long-term costs, due to short-term benefits being more salient (for a review, see Baumeister & Scher, 1988). Conditions signaling potential threat to the self elicit an aversive affective state, from which the narcissist tries to escape as quickly and by the most immediately available means possible. It is possible that their negative emotional state narrows their attentional focus to see only the imminent threatening agent but causes them to lose sight of longer-term effects. However, it is also plausible that they simply misjudge the probabilities of long-term risks and costs, perhaps in connection with overestimating their own capabilities and resources. As a result, their self-enhancement attempts tend at times to overstep the realm of the believable and to undermine the desired effects.
Social intelligence and effective self-regulation depend on one's ability to subtly adjust one's strategies in response to ever-changing environmental contingencies. It appears that narcissists apply their favorite strategies too generally and indiscriminately across tasks and contexts. Thus, although narcissistic strategies make sense and have adaptive value for building and aggrandizing the self, their misapplication to the sphere of interpersonal relationships undermines the self they are trying to build and ultimately contributes to its demise. Sadly, though they might be oblivious to the impact their behaviors have on others, we suggest that the effects of their inability to build warm and enduring relationships are very much felt by narcissists. Although it is unclear that narcissists really want warmth and intimacy, clinical reports describe narcissists emotionally feeling cold, unhappy, empty, depressed, and meaningless (e.g., Kernberg, 1975;Kohut, 1984). Thus, while they spend their public lives engaging in self-aggrandizing behaviors that are in part successful, self-doubt and feelings of worthlessness linger just below the surface and regularly invade their private lives. Although narcissists' self-doubts emerge clearly from the clinical literature, a future challenge will be to document them with empirical research.

Narcissistic Motives and the Role of the Social Audience
If narcissism is a form of socially intelligent problem solving, then one of the most paradoxical elements of narcissistic self-construction is that their characteristic use-or abuse-of the social audience appears so unintelligent. How can it be that narcissists' grandiose self-images are so dependent on social feedback, yet they employ behaviors that engender negative re-sponses from others? Not until one considers the narcissist's subjective internal world, goals, and motivations does this apparent enigma become less mysterious. The key seems to lie in research suggesting that narcissists are more motivated to seek admiration than they are to gain approval (Raskin et al., 1991a).
If narcissists enter social interactions with the goal of seeking corroboration for their grandiose self-view, in which "winning is not only everything, but the only thing," then the specific concerns or desires of the audience are of little importance. All they need is a stage, where they try to win applause, no matter what the interpersonal costs. This is in contrast with social approval-seeking that requires one be sensitive and responsive to a particular audience's wants and preferences (Baumeister, 1982). Preliminary evidence of this insensitivity to social requirements comes from Morf et al. (2001), in which male narcissists, following negative feedback, did not make the typical adjustment of self-presenting modestly toward an expert interviewer-a person likely to detect one of the narcissists' negative attributes. Rather, unlike nonnarcissists, they engaged in as much self-promotion toward the expert as toward the layperson. If anything, there was a trend in the opposite direction, with male narcissists acting even more self-enhancingly toward the expert, perhaps implying that self-construction battles are even more important to win with certain audiences. Thus, though the exact nature of social discriminations will need further clarification by future research, it seems clear that narcissists do not make the usual distinctions between their audiences. They appear to be pursuing a maximal gain strategy, aimed at capitalizing on success, no matter how risky. Self-enhancing toward an expert entails high risk, because it is less probable one can get away with it, but there is also more to be gained, because an expert's favorable opinion is more meaningful.
The same high-risk strategy also is apparent when narcissists make internal attributions for success outcomes (Rhodewalt & Morf, 1995. This has maximal benefit if they continue to succeed but has the potential to seriously undermine the self, if they subsequently fail. Thus, in terms of self-presentational behavior, narcissists appear to employ what Arkin (1981) called the acquisitive kind. These self-presentations refer to those instances in which an individual approaches and embraces risk, treating the self-presentation as a challenge, and presenting the most positive self possible. By contrast, protective self-presentation characterizes the social conservatism of an individual trying to avoid a potential negative outcome or inference. This style involves escaping risk, and "playing it safe"; thus is characterized by avoidance and withdrawal. The motive underlying acquisitive self-promotion is gaining respect or deference, whereas protective self-presentation has avoidance of social disapproval as its goal. This strategy makes sense if narcissists' main concern is not with social disapproval, but with construction of grandiose self. Then, they can afford to not be particularly attentive to cues regarding social appropriateness and decline to properly moderate their explicit self-bolstering and aggrandizing behavior. In this sense, when narcissists' motivations are taken into account, their seeming insensitivity to the social audience is understandable.
Narcissists' lack of concern for others is likely further fueled by their overestimating their ability to deliver successful performances, although this has not been tested directly. In the same vein, we suspect that narcissists even employ self-handicapping, more typically thought of as predominantly self-protective, as an acquisitive brand of self-promotion. They probably feel they can afford the heightened risk of failure that comes with the handicap, as they expect to do well (recall that they tend to overestimate their abilities), and hope that others cannot help but admire one who achieves success in the face of adversity. Thus, they bet all their chips with the intention of capitalizing on success. To circumvent a potential confusion, it should be noted, that in our view (although not explicitly addressed by Arkin), acquisitive self-promotion though clearly not cautious or conservative may nevertheless serve a self-protective motive at a deeper underlying level.
We return to this point shortly, but first introduce Higgins's (1998) model of self-regulatory focus, which is also germane to this aspect of the nature of socially intelligent self-regulation in narcissists. In terms of the Higgins model, it appears that at least at a strategic action level, narcissists self-regulate with a promotion rather than a prevention focus. Individuals with a promotion focus are described by the model as concerned with advancement, growth, and accomplishment. Thus, their strategic inclination is to make progress by approaching matches to desired end-states. In contrast, prevention focused individuals are concerned with security, duties, and obligations, which translates into a prudent and precautionary strategy and avoidance of mismatches to the desired end-state. In short, promotion focus characterizes self-regulation according to potential positive outcomes, and prevention focus according to potential negative outcomes. Consequently, when making choices or decisions in task performance, promotion individuals are driven by a desire to accomplish "hits" and avoid "misses," as opposed to the prevention strategy of attaining "correct rejections" or avoiding making a mistake.
Such a promotion orientation is just what we have been observing in narcissists who push the envelope by taking credit for successes and persist in self-enhancement, even after failure, as opposed to protecting the self by withdrawing or becoming avoidant. Promotion focus is also expected to have an impact on other motivational variables, such as how people evaluate incentives and means, and on people's affective reactions to tasks and outcomes. Thus, applying the regulatory focus framework and paradigms to the study of narcissism may potentially shed more light on how narcissists interact with and experience their worlds.
As already alluded to, it is important to note here that we believe that narcissists' extreme promotion at the behavioral-expression level likely serves a failure avoidance or self-protective function at the underlying motivational level. That is, at the core is the vulnerable narcissistic self that needs to be defended. In principle, such vulnerability could be dealt with in a variety of ways, such as minimizing negative outcomes through avoidance behavior, gaining social approval and support through affiliative and friendly behavior, or maximizing positive outcomes through self-promotion. Narcissists seem to have elected to employ this last strategy. They act offensively, promoting the self at every turn, aiming to capitalize on positive events to the fullest amount possible, and preemptively discounting failure prospects or negative consequences. Thus, instead of engaging in "passive failure avoidance" in the form of mental and physical withdrawal, narcissists engage in "active failure avoidance" in the form of self-promotion-even when such self-promotion in interpersonal contexts risks-and yields-negative consequences (see Elliot & Church, 1997;and Elliot & Harackiewicz, 1996 for similar concepts in the achievement domain). Although not as of yet tested specifically for the interpersonal domain, it is likely that narcissists' positive outcome expectancies are what allow them to pursue this aggressive route to tackle their concerns regarding adequacy of the self.
To summarize, we propose that in dealing with the vulnerable self, "getting ahead" is more important to narcissists, than either minimizing damage to the self, or getting along with others (Hogan, Jones, & Cheek, 1985). Although this may be beneficial to performance outcomes, narcissists trade off maximizing short-term self-gain to the detriment of long-term supportive interpersonal relationships.

Are There Potential Gender Differences in Narcissism?
An analysis of the dynamics of narcissism as expressions of social intelligence at the interpersonal level also requires attention to the role of gender as a potential moderator of adaptive efforts. Which interpersonal behaviors are adaptive or pragmatic varies by gender depending on what is socially expected and accepted sex role behavior. Thus, even if many motivational aspects, characteristic vulnerabilities, and strengths may be shared, it seems likely that at the very least the expression of narcissism would differ by gender.
Indeed, in our recent experimental work, we have been observing greater evidence for "stereotypical narcissistic" behaviors in men than in women. For example, only male, but not female narcissists, employed heightened self-handicapping (Rhodewalt, Tragakis, & Finnerty, 2001), showed a preference for a task framed in terms of interpersonal competition (Morf et al., 2000), or self-enhanced when modesty was called for (Morf et al., 2001). Thus, a fairly clear and consistent picture of self-aggrandizing behavior is emerging for male narcissists (defined by the NPI), but not consistently for females. This raises the possibility that narcissism may not describe the same phenomenon in both genders-a question that, we believe, merits more systematic attention. 2 A possible starting point can be found in the psychoanalytic literature, which has dealt extensively with the developmental courses of sex role socializations that might lead to gender differences in narcissism and has also provided some suggestions as to what these differences might look like. Although it is beyond the scope of this article to describe this in detail, suffice it to say that the disorder is generally thought to be the result of some form of failure in empathic responding by the mother, which in turn results for both genders in a deficient internalized structure of self. Thus, both genders are concerned with "shoring up" the self. At the same time, it suggests, however, that this faulty empathy and the strategies developed to compensate for it may take on different forms for males and females. As described by Philipson (1985), mothers may be responding to boys as a significant other figure (e.g., husband), but to girls as an extension of self. As a result, each gender has different psychological resources to cope with the same lack of an internalized self. Males will more likely display grandiosity, extreme self-centeredness, and excessive need for admiration to establish their "otherness." Females, in contrast, should try to overinvest in, or overidentify with, significant others to reproduce the relationship with the mother.
In light of this, one might expect few gender differences in terms of the underlying concerns about the self, but marked gender differences in their strategic attempts at self-construction and in their reactions to results of these efforts. Indeed these strategic differences may be so distinctive that they may manifest as different clinical disorders. Haaken (1983) suggested that these early disturbances in caregiver empathy are more likely to produce borderline conditions for women and narcissistic personality disorders in men. This is quite plausible, as the borderline, in contrast to the narcissistic personality who develops an early, precarious sense of autonomy, is marked by failure to individuate (Masterson, 1981). It is also consistent with a higher reported incidence of narcissistic personality disorder among men (Akhtar & Thompson, 1982;Masterson, 1981; see also 4th ed. [DSM-IV]; American Psychiatric Association, 1994) and a higher incidence of borderline in women (Haaken, 1983).
Thus, both psychoanalytic theory and empirical observation lead to the conclusion that the excessive efforts to assert one's superiority over others may primarily be part of the male syndrome, whereas narcissistic problems may take on different forms for females. As further evidence to this point, Tschanz, Morf, and Turner (1998) demonstrated that feelings of exploitativeness and entitlement are less integrated into the construct of narcissism for females relative to males. This makes sense, as for males it is more socially acceptable to explicitly dominate and otherwise behave in line with their self-interests, whereas females reap fewer social benefits from the same behaviors. Thus, whereas male narcissists apparently perceive instrumental behaviors as viable strategies, females, due to different interpersonal beliefs, different resources, and different social constraints, likely seek other means of fortifying the self. Females presumably are forced to meet their narcissistic goals through more subtle, indirect, and affiliative means that conform with expectations of their sex role. For example, having been socialized to have a communal orientation toward relationships, one might speculate that females would be more likely to enhance their social power through means such as seeking affiliation with "glamorous" others.
In sum, narcissistic concerns might manifest differently in each gender due to gender differences in development and socialization. In social intelligence terms, stereotypical narcissistic behaviors may be more pragmatic for men than for women, because for men there 191 NARCISSISM: A SELF-REGULATORY MODEL 2 The analysis of gender differences in narcissism is complicated by the fact that the DSM's definition of narcissism is abstracted from clinical descriptions of pathological narcissism, and the majority of these case studies are based on male patients. Consequently, several theorists have raised questions about whether narcissism as defined by the DSM can be generalized to females (e.g., Akhtar & Thompson, 1982;Philipson, 1985). This question also applies to the NPI in that it was developed to be a face-valid measure of the DSM definition of narcissism. A review of the empirical research on narcissism fails to shed much light on the gender question because the data are highly inconsistent with no systematic emerging patterns. One fairly consistent finding is that males typically score on average somewhat higher on the NPI than do females (Carroll, 1987;Farwell & Wohlwend-Lloyd, 1998;Gabriel et al., 1994;Tschanz, Morf, & Turner, 1998). Beyond this, however, it has been unclear if or how gender moderates the relationship between narcissism and behavior or other important variables in systematic and predictable ways. Many studies either failed to report or find gender differences (Auerbach, 1984;Emmons, 1984Emmons, , 1987Raskin et al., 1991a;Rhodewalt & Morf, 1995; others included only one gender as participants (e.g., Kernis & Sun, 1994;Morf & Rhodewalt, 1993); and when gender differences were found, these tended to be small and of questionable meaningfulness (e.g., Buss & Chiodo, 1991;Carroll, 1987;McCann & Biaggio, 1989). are relatively fewer costs associated with these behaviors. Future research is needed to further clarify gender differences in narcissism and to map out the forms of self-construction females employ, particularly when their selves are threatened.

The Internal Logic of the Narcissist's Processing System
We have formulated a conceptualization of narcissism as a self-regulatory processing system that is socially intelligent in that it specifies goals and strategies. Such a framework helps to explain how apparently paradoxical and contradictory components of narcissism can coexist coherently within the same person in a meaningfully organized pattern. That is, it can reveal how it is possible that narcissists on the one hand can be highly dependent on others for feedback affirming their positive self-views while simultaneously engaging in off-putting behaviors that turn others away, ultimately preventing the responses they seek. When one understands how the cognitions and affects in the narcissistic system interrelate, these apparently opposing facets become less inconsistent and can be understood as expressions of the same underlying enduring personality system. This analysis of narcissism is consistent with the Cognitive-Affective Personality System (CAPS) theory of Shoda (1995, 1998) in which a personality type consists of a subset of individuals who have a similar organization of relations among cognitions and affects that become activated in response to particular, psychologically meaningful, situational stimuli. Although for narcissism, these relations have not all been systematically examined yet, some can be inferred from the observed relationships among various variables. Narcissists appear to have a heightened chronic activation level for self-esteem implicating events-such that they readily perceive and try to take advantage of opportunities for self-enhancement, and they also have an amplified need to fend off potential self-threats. As we saw, narcissists focus on situational and task features that provide possibilities for self-promotion: They make ability attributions for positive outcomes (Rhodewalt & Morf, 1995), persist at a task framed in terms of interpersonal competition (Morf et al., 2000), and are attracted to admiring partners with positive qualities (W. K. Campbell, 1999). These behaviors seem to imply that narcissists believe others accept their assertive behaviors at face value and that these behaviors bring about positive consequences (e.g., they gain respect and be admired). They also appear to be discounting risk. When their expectations are violated (e.g., when they receive negative performance or personality feedback), they experience anger (Rhodewalt & Morf, 1998) and employ intra-and interpersonal strategies to undo its ramifications: Narcissists have been shown to devalue the source of the negative feedback (Kernis & Sun, 1994), to derogate a better-performing other to obviate social comparison (Morf & Rhodewalt, 1993), and to glorify their personal romantic histories when faced with rejection (Rhodewalt & Eddings, 2001).
One of the major advantages of the CAPS model is an explicit specification of the relevant types of psychological mediating units (encodings, expectancies, affects, goals, and competencies) important to consider. In the course of reviewing the evidence for our self-regulatory model, it became clear that elaborating narcissists' mental representations and their interactions is a major area in need of further attention, if one is to better understand when and why different cognitive-affective patterns become activated. That is, although we have a reasonably good understanding of the distinctive narcissistic behavioral signatures, it is much less clear what are their specific eliciting situational features, beyond some rather gross generalizations (e.g., failure feedback). Toward this end, there is a need for more direct and comprehensive assessments of the psychological mediating system, for example by coding narcissists' open-ended descriptions and interpretations of events.
These representations span a broad range, including narcissists' subjective phenomenology, personal constructs, implicit theories of self and others, and their social rules. For example: What are the bases on which they feel so privileged or entitled? Do they not understand social rules, or do they simply ignore them, or do they misjudge the effects of their behaviors on others, or all three? That is, what do they think they are doing and how do they justify and make sense of it to themselves? Such fine-grained knowledge would allow one to understand what are the "active" psychological ingredients that enable a narcissist in one situation to be especially charming, engaging, and entertaining, but trigger hostility, other-derogation and aggression in another. It further also would help distinguish between availability of certain competencies and whether or how they are applied. For example, it is unclear whether narcissists simply lack the ability to be empathic toward others, or whether this deficiency has more of a motivational basis. Charting these mental operations would aid in identifying the psychological features of situations that attract narcissists, as well as those that set particular narcissistic processing dynamics in motion-thus refining the assessment of narcissism in self-regulatory terms.
As is evident from the foregoing discussion, our self-regulatory processing framework facilitates seeing the internal, subjective logic and coherence in a personality disposition such as narcissism. In addition, however, framing narcissism in terms of self-regula-tory goals that are inherently hierarchical also allows predictions regarding discriminitiveness, thus incorporating both stability and flexibility in responding. Stability comes from the underlying structure and the stable activation of specific processing dynamics in response to particular psychological triggers. Flexibility results when these triggers change, thus setting different dynamics in motion. We speculate that this latter process may be relatively infrequent in narcissists, in that they may be relatively lacking in flexibility and generally behave more trait-like. Although still awaiting further examination, some of the work we have reviewed in this article provides tentative evidence that narcissists may be low in organizational complexity of the self, as well as low in social discriminativeness and responding. This brings another perplexing issue to the forefront to which, thus far, we have only hinted. Namely, what is the nature of the "self" that the narcissist is trying to construct?
Most people's self-concepts are organized in hierarchical fashion, such that some attributes are more important for them to attain and validate than are others. Do narcissists have a core set of attributes around which they build their self-images? Or, do they simply embrace whichever attributes are likely to earn them social validation and acclaim in a particular situation? Our findings thus far lead us to speculate in favor of the latter, in which the narcissist is pursuing as the ultimate desired self one that is superior and grandiose, but perhaps not wedded to any core collection of attributes and values, and thus also highly context dependent (see also Westen, 1990). Studies are required in which narcissists' self-concepts are measured across situations that vary in psychological and social demands. This investigation might also include exploring narcissists' contingencies of self-esteem (Crocker & Wolfe, 2001). Unlike most people, who have particular domains on which their self-esteem is contingent, narcissists may simply have highly contingent self-esteem across the board. Thus, whereas on the surface, it may appear paradoxical that narcissists invest so much energy in the social validation of self-images that seem to shift with the wind, it makes sense if their stable underlying goal is to be "the best" at everything.

Conclusions
Narcissism is a complex and multidimensional personality and clinical construct that has generated considerable interest across the social sciences and in literature for many decades. Chroniclers of popular culture (Lasch, 1979;Wolfe, 1976) contend that we live in a culture and at a time that promotes greater narcissism among all members of Western society. Research on the construct, however, has been hindered by limitations in its conceptualization, resulting in imprecise definition and controversial approaches to measurement. These difficulties have arisen, because the construct involves a complex pattern of characteristic cognitions, affects, behaviors, and interpersonal relationships open to diverse levels of analyses and interpretations. In this article, we provide a contemporary view of narcissism by casting it in dynamic self-regulatory processing terms. We believe that the process model of narcissism together with supporting validation research help provide a more tractable and heuristic definition of the construct. In its essence, the model views narcissism in terms of motivated self-construction, in which the narcissistic self is shaped by the dynamic interplay of cognitive and affective intrapersonal self-processes and the interpersonal self-regulatory strategies played out in the social arena.
The dynamics are in part the result of narcissists' underlying grandiose, yet fragile self-conceptions and their cynical and unempathetic view of others. These mental representations of self and the self's social context are maintained via distinctive social-cognitive-affective mechanisms. The empirical findings we presented for a variety of these mechanisms corroborate the idea that narcissists are chronically engaged in self-construction efforts and find endlessly inventive ways of reinforcing the self. Thus, they provide accumulating evidence in support of our self-regulatory process model of narcissism.
At a conceptual level, our approach bridges the gap between trait-based and process-based approaches to personality. By addressing the coherent functioning of the whole person, the self-regulatory model permits understanding of both the underlying psychological processes, as well as the regular characteristic (trait-like) patterns of cognition, affect, and behavior of narcissists. Thus, it escapes the common criticism justly leveled at process approaches in the past that they assemble seemingly disconnected lists of personality processes and risk losing the person in the process. The self-regulatory model describes what narcissists are like and can characterize them in broad dispositional terms. At the same time it addresses the psychological processes and dynamics that underlie the dispositional trait and deals with the complex interplay of situations and behavioral tendencies, thus avoiding the criticisms commonly directed at trait approaches. In this way, it provides insight into both coherence and stability, as well as flexibility and distinctiveness. This level of complexity is needed when trying to understand a personality type such as narcissism. Neither a simple combination of traits, nor of processing units will do; rather such understanding requires analysis of a dynamic information and affect processing system that takes into account the network of interconnections and how they function as a whole in interaction with particular types of psychological situations (cf. Mischel & Shoda, 1995. While our work has concentrated on developing a self-regulatory processing model for the narcissistic personality, this model may be thought of as a prototype that can be applied to understanding other dispositional or categorical conceptions of personality in terms of their characteristic self-regulatory processing dynamics. However, it is important to note that it is unlikely that all traits lend themselves equally well to such an analysis. The best candidates are dispositions with specific identifiable cognitive representations of self and other, and characteristic affective components, in terms of needs, goals, values, and emotional reactions. Moreover, these components would have to be formulated at a level of abstraction that allows them to be translated into distinctive motivational concerns and self-regulatory processes (for a similar argument, see Cantor, 1990). Thus, neither descriptive taxonomic categories, such as extraversion or conscientiousness, nor broad information processing styles, such as self-monitoring or public/private self-consciousness are likely fruitfully cast in dynamic goal-process terms, because they are framed at a level too broad for one to map unique working models of self and other, or specific processing dynamics. Such categories describe preferred response or information processing styles that are components of a variety of personality types in different combinations, but themselves do not define specific self-regulatory goals, or cognitive and affective mental representations. 3 For narcissism, the self-regulatory process model has helped clarify the definition and workings of the construct. By making explicit statements about the conditions under which narcissistic processes should take place, the model has helped disentangle some of the conflicting views of narcissism and build an increasingly coherent portrait of the relations of the individual components. Of course, many of the connections have not yet been filled in empirically, but the model and the accumulating evidence provide a basis on which to formulate further predictions. In the spirit of Cronbach and Meehl (1955) we are in the process of "learning more about" the theoretical construct of narcissism by elaborating the nomological net within which it occurs. As they so shrewdly noted, however, we can never know precisely "what the construct is" until we know all the laws that govern it. Nevertheless, it is fair to say that with regard to the construct of narcissism we are coming a lot closer.