Introduction
Personality disorders (PDs) share a general factor (g-PD, McCabe et al., Reference McCabe, Oltmanns and Widiger2022, Sharp et al., Reference Sharp, Wright, Fowler, Frueh, Allen, Oldham and Clark2015; Wright et al., Reference Wright, Hopwood, Skodol and Morey2016) greatly overlapping with the general factor of psychopathology (p-factor) that has emerged from models of common mental disorders (Gluschkoff et al., Reference Gluschkoff, Jokela and Rosenström2021; Laceulle et al., Reference Laceulle, Vollebergh and Ormel2015; Lahey et al., Reference Lahey, Applegate, Hakes, Zald, Hariri and Rathouz2012; McCabe et al., Reference McCabe, Oltmanns and Widiger2022). This notion is parallel to the idea of a p-factor that “influences present/absent performance on hundreds of psychiatric symptoms, which are typically aggregated into dozens of distinct diagnoses, which further aggregate into two overarching externalizing versus internalizing domains, which finally aggregate into one normally distributed dimension of psychopathology from low to high” (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel, Meier, Ramrakha, Shalev, Poulton and Moffitt2014, p. 14). High p-factor scores imply, among many others, worse outcomes for severity, duration of disorder, and comorbidity, i.e., factor underlying all PDs.
The construct of g-PD predominantly revolves around the self-interpersonal impairment scales (McCabe et al., Reference McCabe, Oltmanns and Widiger2022). Sharp et al. (Reference Sharp, Wright, Fowler, Frueh, Allen, Oldham and Clark2015) made a noteworthy observation that all borderline personality disorder (BPD) criteria exhibited substantial and exclusive loadings on the g-PD factor. Wright et al. (Reference Wright, Hopwood, Skodol and Morey2016), while acknowledging the ambiguity inherent in the core composition of the g-PD, proposed that it might align with borderline personality organization (rather than borderline personality disorder per se), a concept described by Kernberg (Reference Kernberg2004). Within this framework, core impairments revolve around maladaptive self and other representations, along with identity formation (Wright et al., Reference Wright, Hopwood, Skodol and Morey2016). Building on these insights, Sharp et al. (Reference Sharp, Wright, Fowler, Frueh, Allen, Oldham and Clark2015) speculated that the key to understanding this phenomenon might be found within Criterion A of the DSM-5-III General Criteria of Personality Disorder. This criterion underscores impairments in self and interpersonal relatedness, which are a central facet across all personality disorders (Bender et al., Reference Bender, Morey and Skodol2011; Kernberg, Reference Kernberg2004). This interpretation of g-PD aligns well with clinical practices within psychotherapy and psychiatry, especially from a psychodynamic perspective (Lingiardi & McWilliams, Reference Lingiardi and McWilliams2015; Millon et al., Reference Millon, Millon, Meagher, Grossman and Ramnath2012). To comprehend personality disorders fully, we initially establish the level of personality dysfunction, which is mirrored in self-interpersonal impairment. Subsequently, we need to outline the stylistic features that a person exhibits, namely pathological personality traits. This could suggest that such stylistic attributes should exhibit stability over time, even in the face of diminishing severity of personality pathology. Actually, Wright et al. (Reference Wright, Hopwood, Skodol and Morey2016) found that g-PD displayed lower absolute stability and held stronger associations with broader indicators of psychosocial functioning, while specific features displayed greater mean stability and demonstrated more confined relationships with functioning. Such understanding of personality disorders is also present in both ICD-11 (WHO, 2019) and Diagnostic and Statistical Manual of Mental Disorders (APA, Reference Allen, Fonagy and Bateman2013) manuals where it is recommended that both the extent of the core deficits in self and stylistic traits distinguishing variants of PDs should be diagnosed simultaneously.
Although the g-PD makeup is still an open question, the latest research (McCabe et al., Reference McCabe, Oltmanns and Widiger2022) affirms that g-PD was defined in large part by self-interpersonal impairments which showed the highest loadings. At the same time, there were weaker associations with specific factors, such as PID-5 Perceptual Dysregulation, which falls under the category of psychoticism, encompassing experiences of marginal or subthreshold psychosis. The question however still remains as to what is the core the self-interpersonal impairment. Individuals diagnosed with various Personality Disorders (PDs) exhibit self and interpersonal impairments, which can be attributed to a deficiency in cognitive flexibility and reflexivity, as well as a broader deficiency in social cognition, including aspects such as empathy and theory of mind (ToM), along with challenges in regulating their emotions (Millon & Davis, Reference Millon, Davis, Clarkin and Lenzenweger1996). Social psychologists (Crick, Reference Crick1995) also point to hostile attributions (HAs) as factors that relate to the difficulties in dealing with ambiguity, which is a source of distress in PDs. Importantly, these factors may lead to many interpersonal dysfunctions. Hostile attributions, a tendency to believe that the actor did something with hostile purpose, intentionally, and is to be blamed for that, even though the social cues may be ambiguous or even neutral (Dodge, Reference Dodge2006; Tuente et al., Reference Tuente, Bogaerts and Veling2019; Combs et al., Reference Combs, Penn, Wicher and Waldheter2007; Bodecka-Zych et al., Reference Bodecka-Zych, Jonason and Zajenkowska2021) are an important factor explaining interpersonal difficulties in several PDs, e.g., antisocial, borderline, paranoid PDs (Millon & Davis, Reference Millon, Davis, Clarkin and Lenzenweger1996), but are also closely related to vulnerable narcissism (Bodecka-Zych et al., Reference Bodecka-Zych, Jonason and Zajenkowska2021) and to depression, often comorbid with PDs (Gasse et al., Reference Gasse, Kim and Gagnon2020; Smith et al., Reference Smith, Summers, Dillon, Macatee and Cougle2016). Hence, a tendency for hostile attributions may be explanatory for all PDs common factor, i.e., interpersonal problems and difficulty building long-lasting and satisfying relationships of all kinds (APA, Reference Allen, Fonagy and Bateman2013; Millon & Davis, Reference Millon, Davis, Clarkin and Lenzenweger1996).
Relating g-PD and hostile attributions
Preschool children typically initially perceive actions as aligning with intentions (Lee et al., Reference Lee, Chang, Ip and Olson2019). For example, if they witness someone taking a toy away from them or hitting them, they are likely to assume it was intentional. However, as part of their socialization process, they eventually learn that actions do not always perfectly match intentions (Feinfield et al., Reference Feinfield, Lee, Flavell, Green and Flavell1999; Guntrip & Sutherland, Reference Guntrip and Sutherland2018; Steiner, Reference Steiner1988; Winnicott, Reference Winnicott1986). This transformation in social cognitive abilities can lead from a tendency toward hostile attributions to more benign attributions. It has also been suggested that children’s hostile attribution bias can become stable and resistant to change over time (Dodge, Reference Dodge2006), underscoring the importance of early childhood as a critical period for the socialization of hostile attributions and an opportune window for potential intervention. A very important aspect that impacts attributional styles is the relation with the primary caregiver and the way they interpret the behavior of the child. The child of a mother who believes that the child misbehaved due to some internal state, like some emotional arousal or temperament characteristics, will likely have lower levels of hostile attributions. In contrast, the child of a mother who believes that the child is manipulative or has some controlling intentions will likely present higher levels of hostile attributions (Lee et al., Reference Lee, Chang, Ip and Olson2019). Failure to form a secure attachment style with primary caregivers may result in hostile attributions that undermine social functioning and proper personality development, which eventually may lead to disorders (Dodge, Reference Dodge2006). Some children consistently exhibit a stable hostile attributional style, which, in certain cases, can serve as a mechanism associated with reactive anger and reactive aggression (Dodge, Reference Dodge2006). However, in other instances, hostile attributions are related and may lead to outcomes such as depression, anxiety, somatic symptoms, and various stress reactions (Dodge, Reference Dodge2006, Zajenkowska et al., Reference Zajenkowska, Ulatowska, Prusik and Budziszewska2017). It’s essential to acknowledge that hostile attributions can manifest as a stable bias across different socio-relational contexts. Still, they can also be triggered, for example, by the nature of a specific relationship (Zajenkowska et al., Reference Zajenkowska, Prusik, Jasielska and Szulawski2021). It is possible that hostile attributions are an effect of problems in the course of socialization, such as insecure attachments with primary caregivers, and may lead to psychological problems. Thus, we aim to examine whether the tendency to make hostile attributions (HAs) may be an underlying factor contributing to g-PD.
HAs encompass intentionality and blame ascribed to the potential harm doer (Combs et al., Reference Combs, Penn, Wicher and Waldheter2007). To infer whether the behavior of an actor was intentional, a perceiver evaluates the social interaction using Theory of Mind (ToM) or broader mentalizing ability (Allen et al., Reference Allen, Fonagy and Bateman2008; Moses & Chandler, Reference Moses and Chandler1992), which also develops in the interaction with primary caregivers (Fonagy & Target, Reference Fonagy and Target1996). Poor mentalizing is hypothesized to underlie the core features of borderline personality disorder (Fonagy & Luyten, Reference Fonagy and Luyten2009). Difficulties in mentalizing may also result in greater hostility and blame (Bodecka et al., Reference Bodecka, Jakubowska and Zajenkowska2021). Finally low mentalizing ability and self-interpersonal impairment (measured with LPFS) have a considerably conceptual and operational overlap (Zettl et al., Reference Zettl, Volkert, Vögele, Herpertz, Kubera and Taubner2020).
Interpersonal difficulties that hostile attributions may lead to also include distrust. An extreme form of interpersonal distrust is paranoia, in which people are convinced that others have an intention to harm them. Yet, paranoia is not restricted to a clinical context as studies have revealed that up to 28% of in the general population occasionally present paranoia-like thoughts (Bebbington et al., Reference Bebbington, McBride, Steel, Kuipers, Radovanovic, Brugha, Jenkins, Meltzer and Freeman2013). Importantly, both PDs (Niemantsverdriet et al., Reference Niemantsverdriet, van Veen, Slotema, Franken, Verbraak, Deen and van der Gaag2022) and hostile attributions (Combs et al., Reference Combs, Penn, Michael, Basso, Wiedeman, Siebenmorgan and Chapman2009) have been linked to paranoia-like thoughts. However, whether paranoia-like thoughts are an extreme point of a tendency for hostile attributions or is a separate factor in PDs is unclear.
Current study
Traditionally, mental disorders have been viewed as distinct, episodic, and categorical conditions; however, current research clearly shows that they are sequentially comorbid, recurrent/chronic, and exist on a continuum (Caspi et al., Reference Caspi, Houts, Belsky, Goldman-Mellor, Harrington, Israel, Meier, Ramrakha, Shalev, Poulton and Moffitt2014). Personality disorders share a common core, i.e., g-PD, and its interpretation, although vague, relates to the common variance of self-interpersonal impairments (e.g. Criterion A of the fifth edition of the DSM-5 Section III Alternative Model of Personality Disorder. AMPD; APA, Reference Allen, Fonagy and Bateman2013) or pathological traits (e.g. PID-5; Krueger et al., Reference Krueger, Derringer, Markon, Watson and Skodol2012; McCabe et al., Reference McCabe, Oltmanns and Widiger2022). Some also posit that the primary indicator of g-PD is borderline personality organization (Wright et al., Reference Wright, Hopwood, Skodol and Morey2016), which is related to problems with mentalizing (Bateman et al., Reference Bateman, Campbell, Luyten and Fonagy2018) that might manifest in hostile attributions (Abate et al., Reference Abate, Marshall, Sharp and Venta2017; Gutz et al., Reference Gutz, Roepke and Renneberg2016; Lee & Hoaken, Reference Lee and Hoaken2007).
Our objective was to advance our understanding of these impairments by investigating the concept of hostile attribution. In the current study, we co-examined the effect of hostile attributions on g-PD extracted from self and interpersonal impairments scales, as well as pathological traits scales representing stylistic traits distinguishing the variants of PDs.
Hostile attributions constitute a complex constellation within general hostility biases, referring to the a priori tendency to perceive and interpret social cues (Smeijers, Reference Smeijers, Martin, Preedy and Patel2022). They are not only related to interpretational processes (Hostile Interpretation Bias, e.g. interpreting facial expressions as angry) and anticipating hostility (Hostile Expectation Bias, e.g. assuming from others hostile reaction to potential conflicts), but also involve ascribing intentions and blaming individuals for presumed hostile behaviors (Hostile Attribution Bias, Smeijers, Reference Smeijers, Martin, Preedy and Patel2022), along with feeling anger in response to such events (Combs et al., Reference Combs, Penn, Wicher and Waldheter2007). Moreover, hostile attributions may be linked to different types of harmful events. For example, they could involve relational harm (e.g., a friend stood me up) or physical harm (i.e., somebody pushed me on the bus) (Combs et al., Reference Combs, Penn, Wicher and Waldheter2007; Smeijers, Reference Smeijers, Martin, Preedy and Patel2022). Finally, they encompass different socio-relational contexts; for instance, one may be more prone to attribute hostility to peers or more to authorities (Zajenkowska et al., Reference Zajenkowska, Prusik, Jasielska and Szulawski2021). In other words, hostile attributions are a heterogeneous construct, and therefore, in the current project, we explore it in various circumstances, including situations involving relational or physical harm and harm done by peers versus male/female authority figures. As reviewed above, because problems with mentalizing have been linked to hostile attributions that could stem from insecure attachment with the primary caregiver, often a mother, we predict that hostile attributions involving authority figures, and possibly females, might have greater predictive value for g-PD compared to hostile attributions in situations involving peers.
In their observations, McCabe et al. (Reference McCabe, Oltmanns and Widiger2022) noted a substantial loading (.89) of pathological traits, specifically PID-5 Perceptual Dysregulation, on g-PD, a factor closely linked to the psychoticism scale. In light of this, we incorporated paranoia-like thoughts into our study to explore their relation to g-PD. Paranoia is characterized by an intense and irrational distrust or suspicion of others, often accompanied by a belief in being targeted, persecuted, or conspired against, even in the absence of supporting evidence (Freeman et al., Reference Freeman, Loe, Kingdon, Startup, Molodynski, Rosebrock and Bird2021). Individuals experiencing paranoia may interpret ordinary situations as threatening, perceiving others as malevolent or intending harm. Paranoia can be conceptualized as comprising ideas of self-reference and persecution (Freeman et al., Reference Freeman, Loe, Kingdon, Startup, Molodynski, Rosebrock and Bird2021). Ideas of self-reference are more closely associated with a heightened sense of personal relevance or significance in everyday events. Individuals endorsing high levels on this aspect of paranoia may interpret neutral or unrelated events as having specific relevance to themselves. On the other hand, ideas of persecution involve experiencing thoughts related to the belief that others are intentionally causing harm or conspiring against them. Considering that paranoid thoughts can be perceived as threatening, they are primarily be associated with fear and anxiety (Bennetts et al., Reference Bennetts, Stopa and Newman-Taylor2022). For individuals with a pronounced hostile attribution bias, feelings of anger may be more prominent as these individuals tend to respond aggressively to perceived provocations (Bondü & Richter, Reference Bondü and Richter2016; Zajenkowska et al., Reference Zajenkowska, Rajchert and Lawrence2020).
Previous work has linked increased paranoia to more hostile attributions (Klein et al., Reference Klein, Kelsven and Pinkham2018), and given this conceptualization, it is crucial to examine whether ideas of persecution, in particular, are distinct from hostile attributions in predicting g-PD factors.
Method
Participants
The sample was representative for the general population of Poland aged 18-65, according to data from the national Polish statistical office from 2022, except for data on education, which was available as most up-to-date from 2021 (Statistics Poland, 2021; 2022). One thousand thirty people completed the survey and were taken into account in the current study (541 females, 489 males, M age = 43.39; SD age = 13.06). Further detailed demographics of the survey participants are presented in Table 1.
Table 1. Descriptive statistics and zero-order correlations between variables under the study
Note. Bonferroni correction applied. Correlations higher than .09 are significant at p ≤ .004. Positive values of the t-test indicate higher scores in females.
AIHQ: hostile attributions in situations involving relational harm, HA-Peers, HA-Males or HA-Female: hostile attributions in situations involving physical harm between peers or between subordinate and a male or female authority figure; LFPS-BF: self-interpersonal impairments scale, PiCD: pathological traits scale, ICD-11
Measures
Hostile attributions were measured with the AIHQ questionnaire (involving relational harm) and visual scenes (physical harm done by a peer or male/female authority figures):
Ambiguous intentions and hostility questionnaire (AIHQ; Combs et al., Reference Combs, Penn, Wicher and Waldheter2007, Zajenkowska et al., Reference Zajenkowska, Prusik and Szulawski2018). Participants were asked to assess whether the other person/s acted on purpose from 1 (definitely no) to 6 (definitely yes), how angry it made them feel from 1 (not angry), 5 (very angry), and how much they blamed the other person/s from 1 (not at all) to 5 (very much) with five hypothetical, ambiguous situations, that involve a range of social relationships. The scoring of the AIHQ often consists of averaging blame, intentionality, and anger, resulting in one index of hostile attribution across the five scenarios (Combs et al., Reference Combs, Penn, Wicher and Waldheter2007). This approach reflects the strong inter-correlation between blame, intentionality, and anger (all r’s > .70; Combs et al., Reference Combs, Penn, Wicher and Waldheter2007). Cronbach’s alpha for the AIHQ general score was .88.
Ambiguous visual scenes (Wilkowski et al., Reference Wilkowski, Robinson, Gordon and Troop-Gordon2007; Zajenkowska & Rajchert, Reference Zajenkowska and Rajchert2020). There were 27 scenes presented on a computer monitor. Eleven of them presented situations where a peer harms another peer (HA-Peers) – for women the peers presented in the pictures were women; for men – men. The other 16 portrayed authority figures (e.g., doctors, police officers, businessmen, of which 8 portrayed females; HA-Female Authorities; and 8 portrayed male authority figures; HA-Male Authorities), harming a subordinate (of a gender congruent with the participant’s gender). Harming included pushing, slamming, hitting, or threatening with, for example, a knife. Each scene included some aspects (e.g., hand/leg direction) indicating hostile behavior and some indicating unintentional behavior (e.g., facial expression). Participants were asked to imagine that the harm was done to them;, therefore, there were separate set of scenes for women, where the harmed person was a woman, and for men, where the harmed person was a man. After the scene was presented for 6s, participants were then asked on subsequent screens to make first, intentionality and then blame judgments: “Please rate to what extent the depicted harm was intentional” on a Likert scale ranging from 1 (not intended at all) to 9 (intended) and “Please rate to what extent you would blame the person for that” from 1 (not at all) to 9 (very much; see Zajenkowska & Rajchert, Reference Zajenkowska and Rajchert2020). To create a total score of hostile attribution bias we averaged the responses for intentionality and blame in ambiguous scenes. Cronbach’s alphas for female participants were: .82 for HA-peers, .78 for HA-Female Authorities, .82 for HA-Male Authorities; for male participants: .79 for HA-peers, .74 for HA-Female Authorities, .75 for HA-Male Authorities.
Green paranoid thoughts scale - Revised (R-GPTS) (Freeman et al., Reference Freeman, Loe, Kingdon, Startup, Molodynski, Rosebrock and Bird2021) is a self-report 18 item-scale that measures the level of paranoia-like thoughts (based on the past month’s experiences) on two subscales – ideas of self-reference (8 items) and ideas of persecution (10 items). Participants rate the items on a scale from 0 (totally disagree) to 4 (totally agree). In addition to the subscale scores, the total score can be computed, and it ranges from 0 to 72, where higher scores indicate higher levels of paranoia-like thoughts. Cronbach’s alphas for the subscales in our study were: .91 for Self-Reference subscale, .96 for Persecution subscale.
Level of personality functioning scale-brief form 2.0. (LPFS-BF) (Łakuta et al., Reference Łakuta, Cieciuch, Strus and Hutsebaut2022; Weekers et al., Reference Weekers, Hutsebaut and Kamphuis2019). The LPFS-BF 2.0 is a brief self-report questionnaire, which assesses the LPFS as described in Section III of the DSM-5.1 It consists of 12 items, divided into two higher order domains: self-functioning (Intrapsychic) and interpersonal functioning (Interpersonal). Participants are asked to rate the 12 items on a 4-point Likert scale from 1 (completely untrue) to 4 (completely true). Cronbach’s alphas for the subscales in our study were: .88 for the Intrapsychic subscale and .80 for the Interpersonal subscale.
Personality inventory for ICD-11 (PiCD) (Cieciuch et al., 2021; Oltmanns & Widiger, Reference Oltmanns and Widiger2018 ). The PiCD is a 60-item self-report measure designed to assess five broad personality domains of the ICD-11. Each domain contains of 12 items rated on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). Cronbach’s alphas for the subscales in our study were: .91 for Negative Affect, .88 for Disinhibition, .87 for Detachment, .81 for Dissociality, and .79 for Anankastia.
Procedure
The study was anonymous and conducted online through a ReaktorOpinii® research panel, based on a survey prepared in Qualtrics®. All participants were registered users of the panel. The survey was designed so as to be filled out on virtually any electronic device (computer, smartphone, tablet). All participants provided informed consent prior to beginning the questionnaire, and participants were allowed to discontinue at any time without penalty. The last screen of the survey was an information screen with contact information for helplines which support people in mental health crises. After finishing the whole set of questionnaires, participants received a remuneration – points from the research panel, which they could later exchange into money. The study materials and procedure were approved prior to conduction by the Maria Grzegorzewska University Research Ethics Committee.
Statistical procedure
To test our hypotheses, we analyzed five linear regression models for each of the p-factor indices (i.e., for LFPS-BF self-interpersonal impairments scale and for PiCD pathological traits scale, ICD-11 separately). In Models 1-4 to account for the zero-order relations, we included only one independent variable per model (i.e., AIHQ: hostile attributions in situations involving relational harm, HA-Peers, HA-Males or HA-Female: hostile attributions in situations involving physical harm). In Model 5, we have analyzed a model with two indicators of paranoia entered simultaneously (i.e., GreenSR and GreenPE). In Model 6, we have analyzed all four variables related to hostile attributions (i.e., all except for the GreenSR and GreenPE) to assess which of these effects are unique. Finally, in Model 7, we additionally included estimates of paranoia, to assess whether the identified effects of hostile attributions would remain significant after accounting for the shared variance with paranoia.
Results
To extract the g-PD factors out of the LFPS-BF (self-interpersonal impairments scale) and PiCD (pathological traits scale, ICD-11) scores, we conducted two separate factor analyses with a forced-choice one-factor solution using the principal axis factoring method. For LFPS-BF (self-interpersonal impairments scale), given there were only two variables (i.e., inter and intra), the extracted factor explained 83.97% of the variance and the strength of the factor loadings of both variables equaled .82. For the pathological traits scale (ICD), the extracted factor explained 47.59% of the variance. Except Anankastia, which appeared as a non-significant indicator of the PiCD (pathological traits scale, ICD-11) p-factor (i.e., λ = -.01), the factor loadings were all good (i.e.., Negative Affect λ = .68, Disinhibition λ = .83, Detachment λ = .61, Dissociality λ = .60). Next, within a stepwise linear regression model these p-factors were sequentially regressed on 1) AIHQ, 2) HA-Peers, HA-Male Authorities, and HA-Female Authorities, and 3) paranoia. The results of these analyses are presented in Table 2 and 3 (LFPS-BF, self-interpersonal impairments scale) and Table 4 and 5 (PiCD, pathological traits scale, ICD-11).
Table 2. The effects of AIHQ, HA-peers, HA-males, HA-females, and paranoia on LFPS P-factor
*p < .05; ** p < .01; *** p < .001. Note. AIHQ: hostile attributions in situations involving relational harm, HA-Peers, HA-Males or HA-Female: hostile attributions in situations involving physical harm between peers or between subordinate and a male or female authority figure; LFPS-BF: self-interpersonal impairments scale, PiCD: pathological traits scale, ICD-11. Models 1 to 4 included a single independent variable (i.e. one of the hostile attribution variables). Model 5 included the two independent variables regarding paranoia. Model 6 included all variables related to hostile attributions as independent variables. Model 7 included all variables at once.
Table 3. Comparison of the tested regression models on LFPS P-factor
*p < .05; ** p < .01; *** p < .001.
Table 4. The effects of AIHQ, HA-peers, HA-males, HA-females, and paranoia on piCD P-factor
*p < .05; ** p < .01; *** p < .001. Note. AIHQ: hostile attributions in situations involving relational harm, HA-Peers, HA-Males or HA-Female: hostile attributions in situations involving physical harm between peers or between subordinate and a male or female authority figure; LFPS-BF: self-interpersonal impairments scale, PiCD: pathological traits scale, ICD-11. Models 1 to 4 included a single independent variable (i.e. one of the hostile attribution variables). Model 5 included the two independent variables regarding paranoia. Model 6 included all variables related to hostile attributions as independent variables. Model 7 included all variables at once.
Table 5. Comparison of the tested regression models on piCD P-factor
*p < .05; ** p < .01; *** p < .001.
In both analyzed models we found the same pattern of relations. At the zero-order level (Models 1-4), we found that AIHQ (hostile attributions in situations involving relational harm) and HA-Females (hostile attributions in situations involving physical harm) were significant and positive predictors of both of the analyzed g-PD factors, while HA-Peers (hostile attributions in situations involving physical harm) and HA-Male Authorities (hostile attributions in situations involving physical harm) were both non-significant. Both indicators of paranoia (Model 5) were significant and positive predictors of the both g-PD factors as well. After accounting for the shared variance between hostile attributions predictor variables (Model 6), we observed that the effect of HA-Peers was significant and negative, while the remaining effects were not altered (i.e., AIHQ and HA-Females were positive and significant predictors, while HA-Males were a non-significant predictor). Finally (Model 7), the inclusion of the GPTS subscales (self-reference and ideas of persecution), significantly reduced the effect of AIHQ (hostile attributions in situations involving relational harm), which, however, remained significant after the extraction of the shared variance with paranoia. The effects of HA-Male Authorities (hostile attributions in situations involving physical harm), remained non-significant, and the effect of HA-Female Authorities (hostile attributions in situations involving physical harm), turned to non-significance as well. The effect of HA-Peers (hostile attributions in situations involving physical harm), remained a significant and negative predictor of the p-factors in both models. Of note, the inclusion of the indicators of paranoia significantly improved the amount of the g-PD factors’ explained variables.
Discussion
The main objective of the current project was to expand existing knowledge about underlying contributors to the general factor of PDs and to specifically examine whether hostile attributions play a role. Mostly our hypotheses were confirmed as hostile attributions predicted both indices of the g-PD factor that were assessed here, i.e., extracted from self-interpersonal impairments scales as well as pathological traits scales representing specific traits distinguishing variants of PDs. However, the relations were more complex than we hypothesized, and thus they need careful interpretation.
In both analyzed models we found the same pattern of relations. Making more hostile and blaming attributions on the AIHQ which contains relational harm situations appeared as a significant and positive predictor of g-PD along with assigning intentionality and blame to female authorities in physically harmful situations. Hostile attributions for situations involving male authorities were a non-significant predictor. However, contrary to what was expected, hostile attributions in the case of peer physical harm negatively predicted both indices of g-PD. Thus, it appears that a greater tendency to make hostile attributions, especially in regard to relational harm, is related to higher levels of the g-PD factor.
These results align with the notion that borderline personality organization (Kernberg, Reference Kernberg2004) was the strongest marker of the general factor (Gluschkoff et al., Reference Gluschkoff, Jokela and Rosenström2021; Sharp et al., Reference Sharp, Wright, Fowler, Frueh, Allen, Oldham and Clark2015). Personality pathology embraces difficulties with self or identity and chronic interpersonal dysfunction (Clarkin et al., Reference Clarkin, Lenzenweger, Yeomans, Levy and Kernberg2007; Livesley, Reference Lingiardi and McWilliams2001; Pincus, Reference Pincus, Lenzenweger and Clarkin2005), and such interpersonal dysfunction in case of more pathological structure like borderline is not simply related to affect dysregulation, but also to particular cognitions. Clarkin et al. (Reference Clarkin, Lenzenweger, Yeomans, Levy and Kernberg2007) point out that when a patient gets angry, he/she is certain that there is a reasonable event that led to this emotion. The problem mostly relies on the lack of integration of positive and negative (often persecutory elements) of the object they interact with. Based on our study, and previous research (Buck et al., Reference Buck, Pinkham, Harvey and Penn2016; Combs et al., Reference Combs, Penn, Michael, Basso, Wiedeman, Siebenmorgan and Chapman2009), we can hypothesize that persecutory elements are closely related to hostile attributions and may be more strongly linked when females are the object of blame.
The finding that hostile attributions toward female authority figures, but not male authority figures, predicts g-PD is somewhat unexpected but not inconsistent with current theory. Hostile attributions towards female authorities might mirror insecure attachment representations, which, in themselves, have a significant contribution to the pathological mechanism of personality organization. As the pathology of personality organization increases, primitive defense mechanisms (e.g., splitting, dissociation) tend to dominate over more mature ones (e.g., humor, rationalization) (Lingiardi & McWilliams, Reference Lingiardi and McWilliams2015). Insecure attachment can be also understood according to object-relations theory (Clarkin et al., Reference Clarkin, Lenzenweger, Yeomans, Levy and Kernberg2007). The theory emphasizes the importance of the mother (still usually a primary caregiver) and child relationship as a primary relationship, which is internalized and forms a template that shapes latter subsequent relationships. Empirical studies link the quality of a safe relationship with the mother with the quality of peer relations and less hostile attributions (Wong et al., Reference Wong, McElwain and Halberstadt2009). Individuals with PDs have higher incidences of early relational traumas caused by their mothers (e.g. Bozzatello et al., Reference Bozzatello, Garbarini, Rocca and Bellino2021, Paris, Reference Paris1998) and maternal aggressive behavior is being studied as a risk factor for developing depression, self-harm behaviors and anxiety which often correlate with PDs (Pearson et al., Reference Pearson, Campbell, Howard, Bornstein, O’Mahen, Mars and Moran2018) and for developing BPD (Zanarini, Reference Zanarini2000).
Apart from object-relations theory, the disparity between the predictiveness of attributions for male and female authority figures might be related to expectations that people generally have towards men and women. Aggression is more socially approved or expected when it is expressed by a man (Eagly & Steffen, Reference Eagly and Steffen1986) and is thus less surprising. Males are stereotypically perceived as tougher and more aggressive than females (Huddy & Terkildsen, Reference Huddy and Terkildsen1993), which may generally contribute to ascribing them violent intentions in ambiguous situations. Thus, ascribing hostility in ambiguous situations with females may be a more relevant predictor of g-PD, as the culture-related gender bias suggests the opposite intentions of women.
Greater levels of g-PD factors were also associated with lower perceived hostility when participants imagined they were physically harmed by peers. It is possible that in the case of physical aggression from peers, the greater the personality dysfunction, the more people are afraid of retaliatory reactions. Here, future studies could control for anxiety levels but also Hostile Expectation Bias. It is possible that in the case of harmful events with peers, rather than attributing blame and intent, people with greater personality dysfunctions might expect hostile reactions (Smeijer, Reference Smeijers, Martin, Preedy and Patel2022). This could imply that people at the first stage make even more benign attributions to protect themselves from potential retaliation. However, this notion needs further investigations.
Importantly when paranoia-like thoughts were included in the model it significantly reduced the effect of HA on the AIHQ, which, however, remained significant after the extraction of the shared variance with paranoia, supporting our predictions. It should be noted that self-reference ideas demonstrated a stronger link to g-PD factors as compared to persecutory ideas. Hence, our results suggest that although paranoia-like thoughts share some variance with hostile attributions, they still are an important and independent predictor of g-PD, and this is particularly so for self-referential ideas. Given that paranoia is inherently a social construct, it makes sense that it would be linked to the self-interpersonal impairment at the core of PDs. Paranoia is often conceptualized as existing on a continuum that spans healthy to pathological populations, and previous work demonstrates that it negatively impacts social functioning across this continuum (Hajdúk et al., Reference Hajdúk, Klein, Harvey, Penn and Pinkham2019). The links between paranoia and g-PD found here offer support for the continuum view and highlight an additional diagnostic and therapeutic aspect to be addressed with PDs and other psychopathological conditions that may be relevant for improving function. Of note, mean scores on both subscales of the paranoia measure indicate our sample reported “moderately severe” to “severe” levels of paranoid ideation as classified by Freeman and colleagues (2021). Additional work will be needed to determine if the relationship between paranoia and g-PD remains the same at lesser levels of paranoia and to further illuminate the shared and independent elements of paranoia and g-PD.
It has to be noted that when the p-factor from the pathological traits scales was extracted, Anankastia was the only non-significant indicator. Four of the domains which are shared by the Alternative Model of Personality Disorders (AMPD) and ICD-11 do refletc the g-PD factor in our study. The literature is currently unclear regarding whether the fifth domain should be Psychoticism (as in AMPD) or Anankastia (as in ICD-11). Our results provide support for the dubious status of the latter in terms of its contribution to the g-PD. Anankastia is characteristic of obsessive-compulsive tendencies and is to some extent opposite of Disinhibition (see correlations of both facets with conscientiousness, Mulder, Reference Mulder2021), which constitutes one of the central traits for borderline PD. Out of the PDs, borderline PD is the best known for being associated with the p-factor (Gluschkoff et al., Reference Gluschkoff, Jokela and Rosenström2021). Thus, Anankastia is not highly represented in among borderline features and its absence in our extracted general factors is likely not problematic. .
As a final consideration, the Anankastia non-significance for the g-PD might have been an artifact of the sample, given that participants were respondents from a research panel. Panelists need to engage in survey completion in order to accrue loyalty points that can be exchanged for rewards. This structure may be well suited to anankastic people who may be more accepting of non-immediately achievable goals, as anankastia links to high conscientiousness (Oltmanns, Reference Oltmanns2021) and perfectionism (Bach & First, Reference Bach and First2018). The sample might have been therefore biased in terms of Anankastia compared to other psychopathological traits, and as a result, this trait may not have differentiated the p-factor well.
Strengths, limitations and future research directions
Our study is the first to examine two parallel indicators of g-PD factors and their relationship to hostile/paranoid cognitions within a single research design. As such, it contributes to the discussion about the nature of the general psychopathology factor, as well as the mechanisms underlying it, and identifies potential treatment targets for psychotherapeutic processes of people with personality disorders. We managed to recruit a relatively large and representative sample of Polish adults, and used not only self-report, but performance-based reactions to described and depicted ambiguous situations. However, as any study, this one is not free of limitations. It is a cross-sectional study, performed only in one culture. Data collection was restricted to participants of one research panel. Moreover, we did not involve a clinical sample, although 10.7% participants declared having a diagnosis of mental/psychiatric conditions at least once in their lifetime, and 20.4% took medicines due to problems with mental health at least once in their life. On the one hand it enabled us to capture the risk factors associated with g-PDs in the general sample and investigate people who might have not received diagnosis, e.g., due to not deciding to visit a psychiatrist or psychologist. On the other hand, the results might vary among a pure clinical sample, in which symptoms are vivid and negatively impact everyday functioning. Future research should focus on involving more aspects of g-PDs, not only cognitions, but also emotionality. It is worth involving qualitative designs in patient studies in order to assess the behaviors and consequences of hostile attributions and paranoid thoughts as the participants view them. Moreover, diary studies could help understand how feelings and cognitions are shaped through the course of days and how they translate into actual behaviors.
Conclusion
We can conclude that both hostile attributions and paranoia are a crucial aspect of personality pathology. They are both related to problems with mentalizing and therefore this aspect of social cognition should be developed further in the psychotherapy process. At this stage of our knowledge, drawing from one of the latest studies on g-PD, it becomes evident that g-PD predominantly revolves around the self-interpersonal impairment scales (McCabe et al., Reference McCabe, Oltmanns and Widiger2022). In our current study, our primary aim was to delve deeper and gain a better understanding of g-PD, particularly as it relates to self-interpersonal impairments. We accomplished this by applying the concept of hostile attribution from social psychology, which is also present in psychoanalytic and psychodynamic theories. The current ICD-11 and DSM-5, along with alternative models for assessing personality pathology, closely align themselves with psychodynamic theories and research concepts (Krzysztof-Świderska & Małek, Reference Krzysztof-Świderska and Małek2022). These include object-relations concepts inspired by Melani Klein (Reference Klein1946, 1996), Kohut’s theory of self (2009) and, significantly, Kernberg’s (2004) notions regarding personality organization. Within the realm of psychodynamic and psychoanalytic theories, one of the foremost challenges lies in research and empirical validation. Melanie Klein (Reference Klein1946, 1996) differentiated between two general positions that individuals adopt during the course of personality development: the paranoid-schizoid and depressive positions. The former is characterized by splitting and hostility, with individuals perceiving the world and others as threatening and provocative. Those who adopt this position typically exhibit significant self-interpersonal impairments. We propose that hostile attributions can serve as an indicator of the typical perception of social interactions in this position. For clinicians, these results hold valuable insights for addressing issues related to hostile attributions in therapeutic settings and relationship functioning. Additionally, our findings shed light on issues like negative transference towards female therapists, which may be rooted in a more general hostile attribution tendency towards female authorities.
Acknowledgments
The studies were funded by National Science Centre (UMO-2021/42/E/HS6/00018) and partially by the Maria Grzegorzewska University internal funding.
Competing interests
All the authors declare that they have no conflict of interest.
Data Availability
Data used in this study will be stored at: http://www.healab.pl
The work was not preregistered.
