The present study was designed to assess the unique links between attachment, mentalizing, and the AMPD personality domains, as well as to explore mentalizing ability as a moderator of the relation between attachment and the maladaptive personality domains. In particular, we proposed that this moderation would be significant in relation to Negative Affectivity, Antagonism, and Disinhibition because of their inclusion in the diagnoses of BPD, ASPD, and NPD [11]. Results provided mixed support for our hypotheses.
As predicted, we found a significant moderating effect for overall mentalizing on the association between attachment avoidance and the Negative Affectivity domain, such that participants high on attachment avoidance and with less accurate mentalizing abilities endorsed higher Negative Affectivity than those individuals with similar attachment avoidance scores but more accurate mentalizing abilities. Post-hoc analyses supported the interaction of mentalizing ability and attachment avoidance for the Emotional Lability, Hostility, and Perseveration trait facets of Negative Affectivity, but not for the remaining facets (i.e., Anxiousness, Lack of Restricted Affectivity, Separation Insecurity, and Submissiveness).
These findings are broadly consistent with the wider literature supporting the mentalization theory, particularly when examining the vast body of research connecting the model to BPD [3, 10]—a disorder marked by pervasive difficulties with mood lability, hostile behaviors, and poor interpersonal relationships [11]. The results also lend support to the application of the mentalization theory to the alternative DSM-5 model trait profile for ASPD, given that Hostility is one of the pathological personality traits proposed to underlie the disorder [11], and previous research has found utility in treating ASPD patients with Mentalization-Based Treatment (MBT) [10, 13, 15, 16]. The current study thus extends this literature base to the AMPD, suggesting the mentalization theory applies to individuals with high Negative Affectivity, regardless of diagnostic classification. Additionally, our study supports prior research indicating mentalizing ability acts as a transdiagnostic mechanism and that, by providing MBT for a wide range of diagnoses (e.g., eating disorders and substance use disorders [6, 7, 13]), the mechanism of change in symptom reduction is through mentalizing ability.
Notably, although we anticipated the remaining trait facets of Negative Affectivity would also have significant moderations, it may be that these findings were not demonstrated given that the AMPD, and each individual domain, are constructed from a mix of behaviors, affects, and perceptions. That is, while mentalization [3, 5] and attachment [1] theories are driven by theory and compose a coherent concept, the AMPD’s individual domains and trait facets are empirically-based and describe different components of the higher-level construct [11]. Subsequently, the other facets that compose the Negative Affectivity domain may have limited overlap with those moderated by mentalizing.
Interestingly, evidence of significant moderation was not found for the Antagonism or Disinhibition domains, in contrast to prior research supporting the application of the mentalization theory to NPD and ASPD, disorders primarily composed of traits related to these domains [11]. Notably, however, BPD was the disorder on which the mentalization theory was originally theorized [2], and the model has only been applied to other personality pathologies more recently [9, 15, 17]. As such, the research connecting the mentalization theory to NPD and ASPD is relatively limited, and, particularly with NPD, is based more on theoretical underpinnings. The only study to our knowledge that has empirically explored the mentalization theory’s application to NPD did so in a sample of individuals with comorbid BPD [17].
Subsequently, it may be that the mentalization theory is only related to the Negative Affectivity aspect of maladaptive personality, resulting in the model being most closely linked with a disorder that is often associated with Negative Affectivity—BPD (r = .81 between BPD and Negative Affectivity in Fowler et al. [65]). Interestingly, the heterogeneity in the categorical diagnosis of BPD is quite extensive, given that there are 256 different ways to be diagnosed with BPD according to the DSM-5’s diagnostic criteria; however, the AMPD criteria for BPD primarily center on the trait facets that compose the Negative Affectivity domain [11, 66]. Furthermore, it may be that what was previously conceptualized as the mentalization theory of BPD is more accurately described as the mentalization theory of Negative Affectivity. This new conceptualization of the mentalization theory’s link to psychopathology warrants future research into the application of the model to any disorder that has a strong Negative Affectivity component, particularly those with symptoms related to Hostility, Emotional Lability, and Perseveration.
Another unexpected finding from our study is the presence of a significant moderation of mentalizing ability when examining attachment avoidance, but not dependence. Given that our study is the first to apply the mentalization theory to dimensional personality, there are no other studies that can fully support or oppose our findings; however, studies using dimensional attachment constructs when examining the mentalization theory and BPD have thus far provided mixed results. For instance, although some studies have found links only between the avoidance dimension and mentalizing [36, 67], others have demonstrated similar findings with only attachment dependence [68, 69]. One potential explanation for this difference in findings is the specific mentalizing ability tapped in these studies as compared to our project. More specifically, while the tasks used in the dependence-supporting studies used measures that examined mentalizing abilities for the self and others (i.e., the Metacognition Assessment Scale—Abbreviated [68] and the Mental States Task [69]), our mentalizing measure, the MASC, asks participants to hypothesize about fictional characters’ emotional and mental states [64]. Indeed, individuals high on attachment avoidance are inherently characterized by avoiding close contact with others, and, as such, it may be that the MASC is more sensitive to detecting mentalizing errors among those high in avoidance, but less useful in perceiving errors among high-dependence individuals (i.e., individuals who often look to others for validation [32]). Mentalizing ability should therefore be examined as a moderator again within the context of attachment and dimensional maladaptive personality, in which a self-focused mentalizing task is used.
Although our primary aim related to the mentalization theory (and moderation analyses), we also sought to examine main effects of attachment insecurity on the Negative Affectivity, Antagonism, and Disinhibition domains, expecting that higher rates of attachment dependence or avoidance would be positively associated with these domains. Our results supported this hypothesis when examining attachment dependence and each of the personality domains. Significant, positive associations were also found when examining the trait facets underlying the Negative Affectivity domain. These main effects of attachment dependence are consistent with previous research [33, 34], including the AMPD domains and 24 of 25 trait facets [36]. Our study also extends these findings to a diverse sample of American undergraduate students, as previous studies used samples of Swedish students [34] and Italian adults [36].
Contrary to our expectations, our findings did not provide evidence for main effects of attachment avoidance on Negative Affectivity, Antagonism, or Disinhibition domains. Previously, only one study has demonstrated a link between attachment avoidance and the AMPD Criterion B domains, and did so with a sample of Italian adults [36]. Although attachment is considered to be stable across cultures [70, 71] and the lifespan [56], it may be that these factors moderate the relation between attachment avoidance and personality, such that the level of avoidance is different during young adulthood than later in adulthood. Because many of our participants were enrolled in a fall-semester introductory psychology class and the median age was 18, it is likely this was their initial semester of college and first time away from home. Importantly, past research has indicated this time of life is marked by increased levels of separation-individuation, a normative developmental process during which young adults begin to separate themselves from parents to form a more coherent and autonomous self-identity [72]. As such, our sample’s attachment avoidance distribution may have been higher on average and had less variability than the Fossati et al. [36] sample because of their differing developmental stages. Nevertheless, additional research is needed to confirm this.
Lastly, we did not find support for our hypothesized main effects of mentalizing ability on Negative Affectivity, Antagonism, and Disinhibition, in contrast to prior research linking mentalizing errors to personality pathology, particularly BPD [40, 47]. Still, the previous literature demonstrating relations between mentalizing and dimensional personality is much more limited [49, 50], especially when examining maladaptive domains and trait facets [39, 51]. Previous studies utilized correlational methods [39, 51] when examining mentalizing pathological personality, rather than multivariate analyses like our study. Although our study conducted a priori power analyses, we utilized a more complex model that may have failed to detect small effects found in prior research, particularly given that many of our analyses were exploratory in nature. Subsequently, replication is needed to confirm the findings reported herein and avoid conclusions that have inadvertently capitalized upon Type I error or sample anomalies.
Nevertheless, it may be that mentalizing is indeed related to personality pathology but in relation to Criterion A of the AMPD. Indeed, such a link was recently demonstrated by Zettl and colleagues [73], in which all domains of the Level of Personality Functioning Scale (LPFS; a self-report tool that measures the level of impairment one is experiencing in their personality functioning) were significantly correlated with mentalizing abilities, suggesting a strong overlap between these constructs. Other studies have also demonstrated links between constructs of mentalizing and Criterion A of the AMPD [45, 52], as well as severity of borderline traits [68]. Future studies exploring level of impairment, as well as the four aspects of impairment (i.e., Identity, Self-Direction, Empathy, and Intimacy) should therefore be undertaken to determine if the mentalization theory in fact moderates one’s severity of impairment related to attachment security, rather than level or presence of traits.
This study need not be considered without limitation. Notably, these analyses were conducted from cross-sectional data and causal inferences cannot be made. Though a longitudinal study would still not be able to support causal relations (given our inability to manipulate variables like attachment security and mentalizing ability), longitudinal analyses using these same constructs could be helpful in determining if mentalizing moderates the relation between attachment and personality across the lifespan. Furthermore, response style biases and shared method variance cannot be eliminated as a possibility for self-report measures (e.g., the PID-5-SF, RQ). Future studies should attempt to collect data via non self-report approaches, such as the Adult Attachment Interview (i.e., the gold standard in assessing adult attachment style) or observational methods, in order to reduce potential sources of statistical noise. Lastly, although our sample of undergraduate students displayed adequate variability on the personality measures, several of the domains and trait facets, such as Psychoticism and Antagonism, were negatively skewed (i.e., most participants reported themselves to have low levels of these traits). Subsequently, our hypotheses should also be tested within a clinical sample, a setting wherein maladaptive personality traits are observed more often and could provide greater variability in personality-related variables. Notwithstanding these limitations, the present study expands the current evidence base regarding relations between the AMPD Criterion B constructs, attachment, and mentalizing ability to a diverse sample of undergraduate students.