Hostile–Helpless states of mind mediate relations between childhood abuse severity and personality disorder features


Hostile–Helpless states of mind mediate relations between childhood abuse severity and personality disorder features

Brent Finger1*, Sooyeon Byun2, Sharon Melnick3 and Karlen Lyons-Ruth3

1Department of Psychology, Montana State University Billings, Billings, MT, USA, 2Harvard Graduate School of Education, Cambridge, MA, USA, and 3Harvard Medical School, Boston, MA, USA


Objective: The present study assessed whether the often reported relation between childhood abuse and the extent of borderline personality disorder (BPD) and antisocial personality disorder (ASPD) features would be mediated by Hostile–Helpless (HH) and/or Unresolved (U) states of mind on the Adult Attachment Interview (AAI).

Method: One hundred and three young adults from low- to moderate-income families were assessed for HH and U states of mind on the AAI, psychopathology on the SCID I and II, and childhood abuse experiences on three validated measures of abuse.

Results: As expected, childhood abuse was related to extent of both BPD and ASPD features and to HH states of mind. In addition, BPD and ASPD features were significantly related to HH states of mind on the AAI, and those states of mind mediated the relations between severity of childhood abuse and later BPD and ASPD features. Contrary to predictions, scores for lack of resolution of loss or trauma on the AAI were not found to mediate the relations between childhood abuse and either BPD or ASPD.

Conclusions: Findings indicate that pervasively contradictory and unintegrated states of mind regarding attachment experiences play an important role in linking past abuse to current personality pathology.

Keywords: Hostile–Helpless; attachment; borderline personality disorder; antisocial personality disorder; abuse; adult attachment interview

Citation: Translational Developmental Psychiatry 2015, 3: 28785 -

Copyright: © 2015 Brent Finger et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License, permitting all non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Received: 6 June 2015; Revised: 4 September 2015; Accepted: 21 September 2015; Published: 4 November 2015

Competing interests and funding: There is no conflict of interest in the current study for any of the authors. The research was supported by NIMH grant R01MH062030 and a grant from the Borderline Foundation to K. Lyons-Ruth and by a grant from the American Psychoanalytic Research Foundation to S. Melnick.

*Correspondence to: Brent Finger, Psychology Department, Montana State University Billings, Billings, MT, USA, Email:


Attachment theory is uniquely suited to explore the interpersonal disturbances characterizing the Axis II personality disorders because personality disorders focus on disturbances in interpersonal functioning. To date, research interest has centered on borderline personality disorder (BPD) and antisocial personality disorder (ASPD). BPD is characterized by ‘a pervasive pattern of instability of interpersonal relationships, self-image, and affects, and marked impulsivity’ (1) and is widely theorized to involve central disturbances in attachment functioning (24). This possibility is supported by the fact that patients with BPD retrospectively report high rates of adverse childhood experiences with attachment figures including physical, sexual, and emotional abuse (57). In addition, prospective studies have also related borderline features to childhood experiences involving abuse (8), maternal hostility (9), and caregiver emotional withdrawal (10).

ASPD is characterized by a ‘pervasive pattern of disregard for and violation of the rights of others’ (1). The hostility, lack of empathy, and manipulative tendencies that such individuals often display compromises their ability to engage in meaningful or lasting relationships. As with BPD, this disorder has also been theoretically and empirically linked to disturbances of attachment functioning. Prospective research supports the view that ASPD is also associated with a variety of unfavorable childhood experiences, most notably caregiver emotional withdrawal in infancy (11), attachment disorganization in middle childhood (11), and physical abuse (12).

Exactly how such earlier experiences of abuse contribute to interpersonal disturbances characterizing BPD and ASPD has been the subject of much theorizing. Attachment theory proposes that childhood experiences with caregivers provide the basis for an ‘internal working model’ or affective-cognitive mental representation of self and others that serves to organize personality development and guide expectations about interpersonal relationships. Victims of early adverse experiences with attachment figures are believed to be particularly vulnerable to the construction of distorted and maladaptive early working models of self and others that, in turn, limit their developing capacity to engage in secure relationships that might lead to more adaptive working models over time (13). In some cases, however, such individuals do encounter more favorable relationships that provide the security and insight necessary for developing more coherent working models. According to this view, children who experience abuse from attachment figures would be expected to develop interpersonal psychopathology only when abuse experiences are associated with continuing distortions in working models of attachment relationships (4).

Unresolved states of mind on the adult attachment interview

The majority of studies examining attachment representations among individuals with BPD and ASPD have used the Adult Attachment Interview (AAI). The traditional coding system for the AAI identifies four principal states of mind regarding attachment experiences: secure-autonomous (F), insecure-dismissing (Ds), insecure-preoccupied (E), and insecure-unresolved with regard to experiences of loss or abuse (U). The unresolved classification identifies speakers who display signs of disorientation (lapses of reasoning or discourse) during portions of the AAI probing experiences of loss (death of significant others during the person’s lifetime) and abuse (experiences of physical or sexual abuse) (14). This classification has been particularly important in theorizing about pathways to character pathology because the unresolved classification is the maternal classification predictive of the child’s disorganized attachment classification in the next generation, as well as of other aspects of child maladaptation (15).

Previous findings indicate that participants with BPD tend to be overrepresented within the preoccupied and unresolved categories (1618), whereas samples involving participants with antisocial tendencies reveal a more mixed pattern tending toward the dismissing classification, with a relatively modest proportion of unresolved classifications (18, 19, 20).

One limitation of using the traditional AAI classification system for these samples is that coding criteria for the unresolved classifications concern relatively limited portions of the interview and apply only to speakers who have encountered and are willing to reveal significant experiences of loss or abuse. Thus, coding for unresolved states of mind might fail to identify those speakers who do not reveal abuse experiences, as well as those who do not show lapses when discussing experiences of loss or abuse, but who display a pervasive lack of integration in discussion of experiences with central attachment figures.

Hostile–Helpless representations of attachment relationships on the AAI

To better capture individuals who display such pervasive contradictions when discussing attachment experiences, Lyons-Ruth et al. developed a coding system for Hostile–Helpless (HH) states of mind regarding attachment relationships in the AAI (21). In contrast to the unresolved classification, HH states of mind are coded based on how primary attachment relationships are represented in the interview, rather than on how loss or abuse experiences are discussed. In keeping with this emphasis, HH states of mind are evaluated across the entire interview, rather than focusing only on material related to loss or abuse as in the unresolved coding.

HH states of mind are characterized by globally devaluing and explicitly contradictory emotional evaluations of caregivers and self. In HH discourse, one or more caregivers are often described in all-encompassing globally devalued terms (either as malevolent or as helpless and abdicating the parental role), yet opposing evaluations of the same caregiver(s) are made at different points in the transcript without reflective comment on the discrepancy. Typically, the individual continues to avow or demonstrate identification with an attachment figure who elsewhere in the interview is represented in globally negative terms (e.g. ‘We were close/She was a witch’; ‘My mother was horrible/Now I do the same thing’). The contradiction between the global negative representation and the continued identification is thought to indicate that the person has not integrated the varied thoughts and feelings concerning the attachment figure into a consistent organized attachment stance. The reader is referred elsewhere for more extended description and relevant findings regarding how the classification criteria for HH states of mind differ from criteria for the categories in the traditional coding system (21, 22).

We theorize that HH states of mind represent essential underlying features of the interpersonal psychopathology characterizing both BPD and ASPD. Patients with BPD are frequently observed to segregate positive and negative feelings and to switch (often abruptly) between contrasting positive and negative attitudes toward others (23). We also speculate that the callous, unemotional, and aggressive tendencies displayed by individuals with ASPD may reflect a pervasive disavowal of attachment-related vulnerabilities resulting, in part, from identification with a hostile caregiver. Both disorders are also characterized by a lack of empathy, by a tendency to perceive others as untrustworthy and malevolent, and by impulsive behavioral tendencies, characteristics that may evolve from difficulties consolidating a consistent and integrated representation of relationships and a consistent strategy for managing close relationships.

Previous studies have shown that HH classifications on the AAI are not significantly related to unresolved classifications (21, 24). Additional work has demonstrated that HH representations are strongly associated with severity of childhood abuse but are not related to extent of family loss or disruption during childhood (21, 25). However, similar to unresolved states of mind, mothers’ HH representations are associated with their infants’ disorganized attachment behaviors, as well as with their own display of disrupted affective communication with their infants (21, 24, 26). Finally, a controlled study of women with BPD versus dysthymia revealed that women with BPD were significantly more likely than those with dysthymia to exhibit HH representations of attachment relationships (27). Individuals with ASPD have not been studied to date using the HH coding system.

The current study was designed to examine both unresolved and HH states of mind on the AAI in relation to the extent of BPD and ASPD features among a low- to moderate-income community sample of 103 young adults. The first hypothesis of the study was that both unresolved and HH states of mind would be associated with severity of childhood abuse. The second hypothesis was that HH states of mind would be associated with both BPD and ASPD features, whereas unresolved states of mind would be associated only with BPD features. The third hypothesis was that the relation between HH states of mind and BPD/ASPD features would be specific and not a function of comorbidities such as anxiety, depression, and substance abuse, whereas unresolved states of mind would be specific to BPD, but not ASPD features. The fourth hypothesis was that HH states of mind would partially mediate the relation between abuse experiences and BPD and ASPD features, whereas unresolved states of mind would partially mediate only the relation between abuse and BPD features.



The 103 participants were selected from a sample of 120 low- to moderate-income young adults (M=19.9 years; SD=1.57; 69 females) and their mothers residing in a major American city who participated in a study of adaptation and psychopathology in young adulthood. Participants were selected for having data on all study measures. One participant who completed the AAI could not be evaluated for lack of resolution so analyses involving this variable were limited to 102 participants. The household income of 59% of the families was under $40,000 per year; 12% of mothers had not completed high-school; 38% of mothers were single parents; and 66% of adolescents were Caucasian.

Of the 120 families in the study, 56 families had been followed since infancy. To increase the sample size for a larger study of attachment and adaptation in young adulthood, 64 additional young adult participants were recruited from the same communities and matched to the longitudinal participants on adolescent age, ethnicity, and mothers’ single parenthood (28). However, family income was slightly higher among the cross-sectional families than the longitudinal families [family income mean range: longitudinal $20,000–$30,000/year; cross-sectional $30,000–$40,000/year (F (1, 118)=9.63, η=0.28, p<0.01)].

The 56 longitudinally studied families were part of a cohort of 76 low-income families recruited during the first 18 months of the child’s life, yielding a retention rate of 74% (14% could not be located, 9% refused, and 3% lived overseas). Attrition was unrelated to all assessments in infancy (effect sizes ϕ or η=−0.14–0.13) and was associated with only one of eight socio-economic indices: single parenthood, χ2(1, 76)=8.66, ϕ=0.34, p=0.01. Half of the families seen in infancy were referred to the study by social service providers because of their concerns about the quality of care provided to the infant; other families seen in infancy did not exhibit problems in infant care. Families first seen in adolescence reported no referrals for parenting help in infancy. Thus, sample composition ensured a range of caregiving risk within the sample. All procedures were approved by the Hospital Institutional Review Board. Written informed consent was obtained from both parent and young adult.


Upon completion of informed consent, young adults were administered the AAI and other questionnaires tapping aspects of their relationships. Diagnostic interviews and measures of traumatic experiences were administered during a later portion of the visit.



Features of BPD and ASPD were assessed using the Structured Clinical Interview for Diagnosis (SCID) for Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV Axis II, administered in the laboratory by trained clinicians (29). Diagnoses of depression, anxiety, and substance abuse disorders were evaluated with the SCID-I (30). The depression variable was scored as positive if the participant met DSM criteria for a major depressive episode, dysthymic disorder, or other depressive disorder. The anxiety variable was scored as positive if the participant met DSM criteria for one or more anxiety disorders including panic disorder, social phobia, specific phobia, obsessive–compulsive disorder, generalized anxiety disorder, or posttraumatic stress disorder. The substance abuse variable indicated the presence of an alcohol or drug abuse diagnosis. The SCID yields reliability kappa’s of 0.61 for current diagnosis and 0.68 for lifetime diagnoses, comparable to other structured diagnostic interviews (30).

We focused our analyses on the number of BPD and ASPD features exhibited rather than on categorical diagnoses, in part, because the base rates of BPD and ASPD (1–2% borderline) are too low to provide meaningful results in moderately sized community samples. In addition, the definition and assessment of personality are now moving toward a dimensional approach, because of increasing concern in personality disorder research that ‘using diagnosis as the measure can limit the information available from traits that are continuous in nature’ (31, p. 189). Hence, our approach is consistent with a number of recent genetic and behavioral studies that have opted to use the number of borderline or antisocial features as the outcome variable rather than diagnostic classification (9, 32).

Socio-demographic risk

Socio-demographic risk (0–3) was assessed as the cumulative presence of three families of origin risk conditions: annual household income under $40,000 per year, mother currently had no partner living in the home, and mother has received no education beyond high school.

Overall severity of childhood abuse

An overall 7-point rating for overall severity of abuse from birth to age 18 was assigned by reviewing scores from all measures that assessed abuse experiences in the current study, including the Conflict Tactics Scale 2nd version (CTS-2) (33), the Traumatic Stress Schedule (TSS) (34), and the Childhood Traumatic Experiences Scales-Revised (CTES-R) (35) coded from the AAI. Whether the family had been involved with state protective services in childhood around the care of the young adult was also assessed from interviews, as was information on whether the young adult had ever been placed in foster care between the ages of 0 and 18 years. This multimethod assessment was selected in the light of evidence that these measures may detect different aspects of abuse experiences (36). Each individual’s overall severity of abuse was classified into one of seven levels as follows: 1) no occurrence of abuse; 2) harsh punishment only; 3) witnessed violence only; 4) verbal abuse only; 5) physical abuse (using state guidelines for abuse), sexual abuse (using state guidelines for abuse), or protective services/foster care involvement; 6) two under level 5; and 7) all those under level 5. Reliability of the overall severity of abuse scale was ICC=0.99, with 37 participants (30% of sample) coded for reliability.

Severity of abusive experiences: TSS-short version

The TSS-short version (34) is an eight-question survey asking about traumatic experiences, including exposure to hazards, natural disasters, accidents, and assaults. Only the first three questions of the TSS focusing on experiences of sexual or physical assault were relevant to the overall severity of abuse rating. In case of a positive answer to any of the items, participants are asked to provide the identity of the assailant, a brief description of the event, and their belief as to why this had happened to them.

Severity of abusive experiences: CTES-R

The CTES-R (37) rates the severity of abuse revealed during the AAI (38). For this study, additional questions probing sexual and physical abuse experiences were added to the standard AAI protocol to gain fuller information on traumatic childhood experiences. Interviews were transcribed and coded on four 5-point scales for severity of physical abuse, sexual abuse, verbal abuse, and witnessing interpersonal violence. Inter-rater reliabilities ranged from ri =0.89 (verbal abuse) to ri =0.98 (sexual abuse). Convergent validity between total scores on the CTES-R and the CTS-2 in the current cohort was r=0.48 (38).

Severity of abusive experiences: CTS-2

The CTS-2 (33) is a widely used 78-item participant report measure of the type and frequency of conflict between family members. Participants rate each item on an eight-point scale indicating the frequency with which a particular conflict tactic occurred during the past year or during another rating period of interest. The rating period specified in the present study was ‘the year that things were worst between you and your parent.’ Participants were probed specifically about incidents involving their mothers. Tactics include physically abusive behavior and emotionally abusive behavior, for example, ‘My parent hit or kicked me; my parent insulted or swore at me.’ The overall scale has a stable factor structure, moderate test–retest reliability, and demonstrated validity (33).

Because the overall rating of severity of abuse was based in part on material from the AAI, we also analyzed participants’ self-reports of abuse on the widely used CTS-2 as a separate measure to control for any common methods variance that might occur between the HH coding system for the AAI and the multimethod abuse variable, which included abuse information gathered from the AAI.

Adult Attachment Interview

The AAI (38) is a semi-structured interview designed to elicit a participant’s current state of mind regarding attachment experiences with significant caregivers during childhood. The interviewer asks about the quality of childhood experiences with parents, the participant’s responses to experiences of rejection, separation, loss, and trauma during childhood, and the participant’s evaluation of the effects of those childhood experiences on his or her current functioning. Interviews were transcribed verbatim and then coded by separate teams of coders from different laboratories using the Adult Attachment Scoring and Classification System (14) criteria for attachment states of mind, including autonomous, dismissing, preoccupied, and unresolved classifications, and the Lyons-Ruth et al. (39) criteria for HH representations of attachment relationships. All coders were naive to other data from the study.

Unresolved states of mind

AAIs were coded for autonomous, preoccupied, dismissing, unresolved, and cannot classify classifications using the Adult Attachment Scoring and Classification System (14). Coders were trained and certified as reliable through the standard training procedures of Main and Hesse and were naive to all other data in the present study. The reliability kappa for classification between two coders on the present sample was K=0.71 (n=27). Given our focus on disorganized states of mind, scales indexing lack of resolution with respect to loss or trauma were the primary variables of interest (unresolved n=19; not unresolved n=95). Analyses involving these variables were limited to 102 participants because one participant could not be coded for lack of resolution. Unresolved status has extensive validity as a correlate of family risk and as a predictor of infant disorganized attachment (40).

HH representations of attachment

Individuals are classified as having HH states of mind regarding attachment experiences if they score 5 or above on a 9-point scale for level of HH states of mind. Transcripts-classified HH are characterized by evidence of opposing evaluations of central attachment relationships that are neither discussed nor reconciled by the participant (e.g. ‘We were friends … We were enemies’.). A HH state of mind suggests that the individual has not engaged in reflection adequate to bring contradictions to a conscious level and achieve a more consistent stance toward attachment experiences.

Before a rating is assigned, the transcript is also scored for seven indicators, theoretically related to such contradictory states of mind in prior clinical theory and observation. Although there is no simple algorithm relating these frequencies to a particular scale score, the first two indicators are especially central to the concept of HH states of mind and would be heavily weighted in assigning a classification. Those indicators include frequency of global devaluation of a caregiver, including actively hostile devaluation and ‘cool’ derogating descriptions; evidence of identification with a hostile caregiver, where the participant appears to accept or value similarities between the negatively evaluated attachment figure and the self, even though these similarities may not be explicitly acknowledged; frequency of references to fearful affect; frequency of references to a sense of self as bad, including generalized negative self-descriptors and references to feelings of shame or feelings of being undeserving of positive attention; frequency of instances of laughter at pain, in which the description of emotionally painful experiences is accompanied by joking or laughter; and evidence of ruptured attachments, when a participant refers to having deliberately terminated contact with one or more nuclear family members. Additional detail is available elsewhere describing the classification criteria and how this coding system differs from and extends the traditional coding system for the AAI (21, 22, 25). Transcripts were coded blind to all other data. Reliability coefficients on 15 randomly selected transcripts were as follows: scaled score for HH states of mind ri =0.83 and HH classification K=0.82.

Analytic strategy

All variables were first checked for skew and homogeneity of variance. Due to the skewed distribution of the data on extent of BPD and ASPD features, these variables were transformed using square root transformations. The first and second hypotheses were tested with Pearson’s correlations. Hypothesis 3 was assessed by hierarchical linear regression models calculated using SPSS Version 19 software (41), with extent of BPD or ASPD features as the dependent variable and Unresolved or HH-scaled scores as the independent variable. In relation to Hypothesis 4, the level of HH state of mind and the level of lack of resolution were separately tested as mediators of the effect of severity of childhood abuse on BPD and ASPD features. Preacher and Hayes’ (42) INDIRECT macro for SPSS Version 19 was used to estimate the indirect effects of predictor variables on outcomes through the proposed mediators and to calculate the bootstrapped confidence intervals to assess the significance of the indirect path (43).


Descriptive statistics and information on BPD features and ASPD features

Descriptive statistics on all variables are presented in Table 1. Fifty-one percent (n=52) of participants were classified as showing HH representations, and 18% (n=18) were classified as having unresolved loss or trauma. Thirty-eight percent (n=39) of the sample met criteria for one or more BPD features; 19% (n=20) for two or more features; and 3.9% (n=4) met criteria for diagnosis (five or more features). Among the 38% displaying any borderline features, 59% (n=23) exhibited impulsive self-damaging behaviors (DSM-IV Criterion 4); 41% (n=16) revealed intense and unstable relationships (DSM-IV Criterion 2); and 38% (n=15) exhibited recurrent suicidal threats or self-injury (DSM-IV Criterion 5). Other features were less frequent, so that the predominant features were impulsive, self-damaging behaviors, and unstable relationships.

Table 1. Descriptive data for study variables
Continuous variables Mean SD Range
Socio-demographic risk 1.27 0.99 0–3
Hostile–Helpless state of mind (scaled score) 4.72 1.66 2–8
Unresolved trauma and loss (scaled score) 2.44 2.04 0–8
Unresolved trauma (scaled score) 0.86 1.52 0–5
Unresolved loss (scaled score) 2.05 2.02 0–8
Severity of childhood abuse 3.30 2.06 1–7
Extent of physical or verbal abuse by mother 10.81 13.10 0–86
Number of BPD features 0.80 1.34 0–5
Number of ASPD features 0.77 1.19 0–4
Categorical variables Percent of sample
BPD diagnosis 3.9
ASPD diagnosis 5.8
Male participants 37.9
Depressive disorder diagnosis 33.0
Anxiety disorder diagnosis 41.8
Substance abuse diagnosis 37.9
Note. N=103 except unresolved state of mind (N=102).

Thirty-seven percent (n=38) met criteria for one or more ASPD features; 24% (n=25) for two or more features; and 6% (n=6) met diagnostic criteria (three or more features). Among the 37% displaying any features, 74% (n=28) displayed failure to conform to social norms (DSM-IV Criterion 1); 47% (n=18) exhibited irritability and aggressiveness (DSM-IV Criterion 4); and 37% (n=14) revealed reckless disregard for the safety of self or others (DSM-IV Criterion 5). Other features were less frequent, so that the principal antisocial features in the sample were aggressive, reckless, and illegal behavior. No significant sex differences were found for BPD features. However, extent of ASPD features did vary significantly by sex, with males exhibiting more features [F (1, 102)=10.4, p=0.002; male M=1.23; female M=0.48]. Consequently, we controlled for male gender in our analyses involving ASPD.

Relations between HH states of mind and unresolved states of mind

In the traditional Main et al.’s (14) coding system for unresolved loss or trauma, unresolved loss and unresolved trauma are grouped together in a single coding category. However, when the rating scales for extent of unresolved loss and extent of unresolved trauma were separately analyzed, they proved to be uncorrelated and represent independent dimensions of state of mind (Table 2). Ratings of unresolved loss were also unrelated to ratings of HH states of mind (Table 2), whereas ratings of unresolved trauma were only modestly related to HH states of mind, r=0.20 (Table 2). Thus, three largely independent aspects of young adult state of mind are being captured by the coding for unresolved loss, unresolved trauma, and HH attachment representations. Below, we present our analyses for the HH-scaled scores followed by corresponding analyses for the unresolved-scaled scores.

Table 2. Strength of associations among study variables
  1 2 3 4 5 6 7 8 9 10 11 12
1. Male gender                      
2. BPD features 0.14                    
3. ASPD features 0.30** 0.50***                  
4. Depression 0.09 0.50*** 0.27**                
5. Anxiety 0.15 0.19x 0.17x 0.24*              
6. Substance abuse 0.18x 0.51*** 0.43*** 0.35*** 0.23*            
7. Maternal abuse 0.01 0.24* 0.08 0.26** 0.20* 0.17x          
8. Overall abuse 0.07 0.30** 0.26** 0.10 0.13 0.13 0.22*        
9. Socio-demographic Risk −0.03 0.15 0.14 0.08 0.19x –0.03 0.04 0.28**      
10. HH state of mind (scaled) –0.01 0.33** 0.28** 0.10 0.30** 0.11 0.37*** 0.45*** 0.27**    
11. Unresolved loss or trauma (scaled) –0.01 0.09 0.16 0.18x 0.04 –0.04 0.09 0.27** 0.25* 0.12  
12. Unresolved trauma (scaled) 0.06 0.27** 0.12 0.10 0.05 –0.05 0.08 0.41*** 0.21* 0.22* 0.44***
13. Unresolved loss (scaled) –0.04 0.03 0.16 0.17x 0.07 –0.04 0.09 0.13 0.12 0.03 0.87*** 0.06
Note. N=103 except 11–13 (N=102). For two continuous variables r is shown; for one continuous variable and one categorical variable µ is shown; for two categorical variables ϕ is shown. xp<0.10. *p<0.05. **p<0.01.***p<0.001.

Overall severity of abuse and HH states of mind

Consistent with our first hypothesis, overall severity of abuse was robustly related to level of HH state of mind (Table 2). Maternal abuse alone, as assessed on the CTS-2, was also related to HH state of mind (Table 2). Similar relations were obtained for the HH classification.

Axis I and Axis II psychopathology and HH states of mind

Consistent with the second hypothesis, HH-scaled scores were significantly related to BPD and ASPD symptoms (Table 2). HH-scaled scores were also significantly related to the presence of an anxiety disorder and to demographic risk (Table 2). Similar correlations were observed for the HH classification; however, the HH-scaled score was used in subsequent analyses to maximize power.

Specificity of relations between HH and BPD/ASPD

The third hypothesis posited that the relations between HH and BPD/ASPD would be independent of Axis I comorbidities. Both BPD features and ASPD features were significantly correlated with the presence of depressive and substance abuse disorders (Table 2). To test the third hypothesis that the relation between HH and BPD was not a function of associated Axis I comorbidities, separate stepwise linear regression analyses were conducted on BPD features and ASPD features. In relation to BPD features, depressive and substance abuse disorders were entered in the first step of the equation, and HH-scaled scores were entered second. Anxiety disorders were not related to BPD features so were not included as covariates. In the second equation for ASPD features, sex was included in the first step, depression and substance abuse were entered in the second step, and HH scores were entered in the final step. As shown in Table 3, even after controlling for the effects of sex and comorbid psychopathology, both BPD and ASPD features were independently associated with HH states of mind.

Table 3. Hierarchical regressions of BPD and ASPD features on Hostile–Helpless state of mind scale
  F change df R2Δ B SE B B t p
BPD features: regression model                
  Step 1 30.53*** 2, 100 0.38          
   Depression       0.58 0.13 0.39 4.62*** 0.000
   Substance abuse       0.56 0.13 0.36 4.32*** 0.000
  Step 2 11.33** 1, 99 0.06          
   HH states of mind (scaled)       0.11 0.03 0.26 3.37** 0.001
ASPD features: regression model                
  Step 1 10.09** 1, 101 0.09          
   Male gender       0.44 0.14 0.30 3.18** 0.002
  Step 2 10.93*** 2, 99 0.16          
   Depression       0.51 0.14 0.35 3.74*** 0.000
   Substance abuse       0.19 0.14 0.13 1.39 n.s.
  Step 3 7.42** 1, 98 0.05          
   HH states of mind (scaled)       0.10 0.04 0.23 2.72** 0.008
Note. N=103. **p<0.01. ***p<0.001.

HH states of mind as mediators of effects of abuse on BPD/ASPD features

As displayed in Table 2, the overall severity of childhood abuse was related to both BPD and ASPD features as expected from previous literature. Maternal abuse assessed by CTS-2 was also related to BPD features, but not to ASPD features (Table 2). Given the relations between abuse and HH, and abuse and BPD/ASPD, we then examined our fourth hypothesis that HH states of mind would mediate the relations between abuse experiences and extent of BPD and ASPD features. Bootstrapped confidence intervals confirmed the presence of an indirect effect of overall severity of abuse on borderline features through HH state of mind scores, because the bootstrapped confidence intervals for the point estimate of the indirect effect did not contain zero (Table 5). A Sobel test, further, confirmed that the reduction in the magnitude of the relation between overall severity of abuse and BPD features was significant when HH states of mind were included in the equation (Z=2.91, p=0.004), providing additional support for the mediating role of HH states of mind. We then conducted an additional mediation analysis to determine whether HH states of mind also mediated the relation between maternal abuse only and BPD features. Bootstrapped confidence intervals confirmed the presence of an indirect effect linking maternal abuse to BPD features via HH states of mind (point estimate=0.0057; SE=0.0027); bootstrapped bias corrected and accelerated (95% CI: 0.0015–0.0125). Zero was not included in the confidence interval, thus providing support for mediation. A Sobel test, further, confirmed that this drop in the contribution of maternal abuse severity to BPD features was significant once HH state of mind scores were entered (Z=2.51, p=0.01), providing additional support for mediation.

A similar equation was used to assess the role of HH states of mind in mediating the relation between overall severity of abuse and ASPD features. This equation also controlled for sex because ASPD symptoms were significantly elevated among male participants. Bootstrapped confidence intervals for the point estimate of the indirect effect of severity of abuse on ASPD features through HH states of mind confirmed the presence of an indirect effect (Table 5). A Sobel test, further, confirmed a significant drop in the contribution of overall abuse severity to ASPD features once HH state of mind scores were entered (Z=2.52, p=0.01), providing additional support for mediation.

Overall severity of abuse and unresolved states of mind

Consistent with our first hypothesis, the scale for overall unresolved loss or trauma was related to overall severity of abuse (r=0.26). However, notably, it was not related to severity of maternal abuse (r=0.10; Table 2). Unresolved trauma alone was also related to overall severity of abuse (r=0.40) but not to severity of maternal abuse only (r=0.08). The scaled score for unresolved loss was not related to either form of childhood abuse. It should be noted that unresolved trauma cannot be coded in the AAI coding system unless the participant gives evidence of abuse in the interviews, so some relation between presence of abuse and lack of resolution of abuse is built into the coding system.

Axis I and Axis II psychopathology in relation to unresolved states of mind

Contrary to our second hypothesis that unresolved loss or trauma would be related to BPD features, only unresolved trauma was significantly related to BPD features (Table 2). Also, unexpectedly, none of the unresolved indicators were related to anxiety disorders, depressive disorders, or substance abuse disorders (Table 2). As expected, none of the unresolved scales were related to ASPD features (Table 2). Thus, not only were unresolved loss and trauma scales less strongly and consistently related to abuse experiences, they were also less consistently related to BPD and ASPD outcomes and associated Axis I psychopathology than were the HH scores (Table 2).

Specificity of relations between unresolved state of mind and BPD

To evaluate our third hypothesis that unresolved state of mind would be related to BPD independent of associated Axis I diagnoses, a stepwise linear regression analysis was conducted with depressive disorders and substance abuse disorders entered in the first step of the equation, and unresolved trauma was entered in the second step. Consistent with the hypothesis, unresolved trauma continued to be associated with the extent of BPD features after controlling for comorbid Axis I psychopathology (Table 4).

Table 4. Hierarchical regression of BPD features on unresolved trauma scale
  F change df R2Δ B SE B B t p
BPD features: regression model                
  Step 1 30.14*** 2, 99 0.38          
   Depression       0.59 0.13 0.39 4.64*** 0.000
   Substance abuse       0.54 0.13 0.35 4.13*** 0.000
  Step 2 10.99** 1, 98 0.06          
   Unresolved trauma       0.12 0.04 0.25 3.32** 0.001
Note. N=102. **p<0.01. ***p<0.001.

Unresolved states of mind as mediators of effects of abuse on BPD/ASPD features

Our final analysis was designed to test our fourth hypothesis that lack of resolution would mediate the relations between abuse and BPD features but not ASPD features. Bootstrapped confidence intervals were evaluated following Preacher and Hayes (43) to assess the indirect effects of each unresolved variable in linking childhood abuse to both types of personality disorder features. Results indicated no mediation by any of the three unresolved variables for either BPD or ASPD features (Table 5).

Table 5. Evaluating the indirect effects of overall severity of abuse on BPD and ASPD features through unresolved states of mind and Hostile–Helpless representations of attachment relationships
    Bootstrapped bias corrected and accelerated 95% CI  
Potential mediators Point estimate (SE) Lower Upper Mediation
BPD featuresa,b        
  Unresolved loss or trauma –0.0391 (0.0423) –0.1510 0.0252 No
  Unresolved trauma 0.0484 (0.0340) –0.0085 0.1299 No
  Unresolved loss 0.0149 (0.0245) –0.0117 0.1001 No
ASPD featuresb,c        
  Unresolved loss or trauma –0.0374 (0.0424) –0.1568 0.0215 No
  Unresolved trauma 0.0224 (0.0280) –0.0286 0.0854 No
  Unresolved loss 0.0217 (0.0271) –0.0060 0.1184 No
BPD featuresa,d        
  Hostile–Helpless representation 0.0393 (0.0181) 0.0070 0.0791 Yes
ASPD featuresc,d        
  Hostile–Helpless representation 0.0339 (0.0176) 0.0037 0.0745 Yes
Note. 5,000 bootstrapped samples.
aNo control variables. bN=102. cGender controlled. dN=103. If the confidence intervals contain zero, mediation is rejected.


Behavioral genetics studies have found that BPD and ASPD share common genetic and environmental contributors (44). The current work indicates that young adults with BPD and ASPD features also share certain aspects of their mental representations of attachment relationships. Young adults with ASPD or BPD features were more likely to show an overall pattern of unintegrated, contradictory, and globally devaluing representations of childhood attachment relationships on the AAI. The relation between these aspects of mental representation and BPD/ASPD features cannot be fully accounted for by comorbid diagnoses of substance abuse and depression. Instead, HH states of mind appear to tap aspects of young adult attachment representations associated with personality pathology. These findings add to several other reports indicating that the HH coding system for the AAI provides an important tool for identifying pathological states of mind associated with BPD (27) and other disturbances in personality functioning (24).

The results also illuminate the role of mental representations in mediating the effects of abuse on personality functioning. Abuse severity was robustly related to HH representations of attachment relationships in this cohort, replicating previous findings in other samples (24, 25). Findings are also consistent with previous evidence indicating that abuse experiences are associated with both BPD and ASPD (57, 45). However, the relation of abuse severity to the young adult’s BPD or ASPD features was mediated by whether or not abuse was further associated with HH states of mind regarding attachment figures. Thus, BPD or ASPD features may be more likely to emerge among abuse victims who maintain unintegrated representations of their attachment experiences into adulthood.

As we had expected, unresolved states of mind were not significantly related to ASPD features, consistent with previous studies reporting low rates of lack of resolution in criminal populations (19, 20). However, contrary to prediction, both extent of overall unresolved state of mind and extent of unresolved loss specifically were unrelated to BPD features. In addition, all unresolved scales were unrelated to Axis I psychopathology. Only unresolved trauma was significantly related to borderline psychopathology. Thus, previously reported associations between BPD and unresolved status may have stemmed largely from an association between BPD and unresolved abuse. However, indicators of unresolved abuse on the AAI did not mediate the relation between abuse and BPD features, suggesting that the discussion of abuse on the AAI does not account for which abused individuals will develop BPD features. Consequently, the HH coding system may be more capable of capturing the representational features of working models of attachment relationships that are essential to borderline psychopathology.

Interestingly, we did not find a significant association between ASPD features and maternal abuse specifically, whereas BPD features were related to maternal abuse as well as to overall abuse. This pattern of results may stem from the joint finding that boys are more likely to develop ASPD features, as noted earlier, and that fathers are more likely than mothers to implement harsh physical discipline with boys (46). It may be that the majority of parental abuse experienced by participants with ASPD features was carried out by paternal attachment figures and, thus, was captured by the measure of overall abuse but was not captured by the measure that only assessed extent of maternal abuse. When the pattern of correlations in (Table 2) is examined, it is also apparent that maternal abuse alone was associated with other psychiatric symptoms that were not associated with overall abuse, including anxiety disorders, depressive disorders, and substance abuse. The current study was not designed to assess differences in types of abuse or perpetrators of abuse, but these differences in psychopathology associated with maternal abuse, in particular, suggest that further research on the specificity of effects of maternal versus paternal abuse might be important to pursue.

One important limitation of the current study is that participants were drawn from a community sample in which few individuals met full diagnostic criteria for BPD or ASPD. Further work is needed among individuals meeting full criteria for BPD/ASPD to assess whether the effects of abuse are mediated through HH states of mind for those individuals as well. In addition, the majority of BPD features in this sample were those of impulsive, self-damaging behavior, whereas the majority of antisocial features pertained to impulsive, aggressive, reckless, and norm-violating behavior. Consequently, our results may be specific to these features. Finally, this was a low-income cohort, and replication is needed in more affluent samples.

Further work is also needed to better understand how, and for whom, abuse becomes associated with HH states of mind, and ultimately with BPD and ASPD. A number of additional variables may independently or interactively contribute to these linkages, including genetic risk, other aspects of family environment, psychosocial adversity, and peer relationships. Future studies using longitudinal designs and multimethod measures will be needed to aid our understanding of the development of these devastating pathologies.

However, the current work suggests that attachment representations may present an important focus of treatment for individuals who already display, or are at risk for developing, personality pathology. Such intervention may be particularly important during childhood and adolescence prior to the time when personality disorder features begin to consolidate during early adulthood. Attachment representations may also be more flexible and amenable to change in childhood and adolescence. Thus, future work is needed to explore the developmental periods during which problematic attachment representations begin to emerge and stabilize. To this point, few studies have examined more disturbed forms of attachment representations during middle and late childhood, and none have examined HH features. In light of strong current interest in identifying childhood precursors of BPD and ASPD features (9, 10, 37), these emerging representational distortions would be important targets for early identification. Thus, further development of assessments capable of identifying these emerging representations should also be given high priority.


The author thanks the families who have been our most important collaborators and so generously given their time to this study. The author acknowledges the invaluable contributions of Nancy Hall Brooks, the study coordinator, and many additional students and staff to the conduct of the study.


  1. American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders, fourth ed. Washington, DC: American Psychiatric Press.
  2. Bateman A, Fonagy P. Comorbid antisocial and borderline personality disorders: mentalization-based treatment. J Clin Psychol 2008; 64: 181–94. PubMed Abstract | Publisher Full Text
  3. Gunderson J, Lyons-Ruth K. BPD’s interpersonal hypersensitivity phenotype: a gene-environment transactional model. J Pers Disord 2008; 22: 22–41. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  4. Levy KN. The implications of attachment theory and research for understanding borderline personality disorder. Dev Psychopathol 2005; 17: 959–86. PubMed Abstract
  5. Herman JL. Histories of violence in an outpatient population: an exploratory study. Am J Orthopsychiatry 1988; 56: 137–41. Publisher Full Text
  6. Trull T. Relationships of borderline features to parental mental illness, childhood abuse, Axis I disorder, and current functioning. J Pers Disord 2001; 15: 19–32. PubMed Abstract | Publisher Full Text
  7. Zanarini MC, Frankenburg FR, Reich DB, Hennen J, Silk KR. Adult experiences of abuse reported by borderline patients and axis II comparison subjects over six years of prospective follow-up. J Nerv Ment Dis 2005; 193: 412–16. PubMed Abstract | Publisher Full Text
  8. Widom CS, Czaja SJ, Paris J. A prospective investigation of borderline personality disorder in abused and neglected children followed up into adulthood. J Pers Disord 2009; 23: 433–46. PubMed Abstract | Publisher Full Text
  9. Carlson EA, Egeland B, Sroufe LA. A prospective investigation of the development of borderline personality symptoms. Dev Psychopathol 2009; 21: 1311–34. PubMed Abstract | Publisher Full Text
  10. Lyons-Ruth K, Bureau JF, Holmes B, Easterbrooks A, Brooks NH. Borderline symptoms and suicidality/self-injury in late adolescence: prospectively observed relationship correlates in infancy and childhood. Psychiatry Res 2013; 206: 273–81. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  11. Shi Z, Bureau JF, Easterbrooks MA, Zhao X, Lyons-Ruth K. Childhood maltreatment and prospectively observed quality of early care as predictors of antisocial personality disorder. Infant Ment Health J 2011; 33: 1–14.
  12. White HR, Widom CS. Intimate partner violence among abused and neglected children in young adulthood: the mediating effects of early aggression, antisocial personality, hostility and alcohol problems. Aggress Behav 2003; 29: 332–45. Publisher Full Text
  13. Bowlby J. Attachment and loss, Volume II: separation, anxiety, and loss. London, UK: Hogarth Press; 1973.
  14. Main M, Goldwyn R, Hesse E. Adult attachment scoring and classification system. Version 7.2. Unpublished manuscript, University of California at Berkeley, Berkeley, CA, 2003.
  15. Lyons-Ruth K, Jacobvitz D. Attachment disorganization: genetic factors, parenting contexts, and developmental transformation from infancy to adulthood. In: Cassidy J, Shaver PR (Eds.). Handbook of attachment: theory, research, and clinical applications. New York: Guilford Press; 2008. p. 666–97.
  16. Barone L. Developmental protective and risk factors in borderline personality disorder: a study using the Adult Attachment Interview. Attach Hum Dev 2003; 5: 64–77. PubMed Abstract | Publisher Full Text
  17. Fonagy P, Leigh T, Steele M, Steele H, Kennedy R, Mattoon G, et al. The relation of attachment status, psychiatric classification, and response to psychotherapy. J Consult Clin Psychol 1996; 64: 22–31. PubMed Abstract | Publisher Full Text
  18. Patrick M, Hobson RP, Castle D, Howard R, Maughan B. Personality disorder and the mental representation of early social experience. Dev Psychopathol 1994; 6: 375–88. Publisher Full Text
  19. Levinson A, Fonagy P. Offending and attachment: the relationship between interpersonal awareness and offending in a prison population with psychiatric disorders. Can J Psychoanal 2004; 12: 225–51.
  20. van IJzendoorn MH, Feldbrugge JTTM, Derks FCH, de Ruiter C, Verhagen MFM, Philipse MWG, et al. Attachment representations of personality-disordered criminal offenders. Am J Orthopsychiatry 2009; 67: 449–59.
  21. Lyons-Ruth K, Yellin C, Melnick S, Atwood G. Expanding the concept of unresolved mental states: hostile/helpless states of mind on the adult attachment interview are associated with atypical maternal behavior and infant disorganization. Dev Psychopathol 2005; 17: 1–23. PubMed Abstract | PubMed Central Full Text
  22. Melnick S, Finger B, Hans S, Patrick M, Lyons-Ruth K, Hostile-helpless states of mind in the AAI: a proposed additional AAI category with implications for identifying disorganized infant attachment in high-risk samples. In: Steele H, Steele M (Eds.). Clinical applications of the Adult Attachment Interview. New York: Guilford Press; 2008. p. 399–423.
  23. Butler AC, Brown GK, Beck AT, Grisham JR. Assessment of dysfunctional beliefs in borderline personality disorder. Behav Res Ther 2002; 40: 1231–40. PubMed Abstract | Publisher Full Text
  24. Finger B. Exploring the intergenerational transmission of attachment disorganization. Unpublished doctoral dissertation, University of Chicago, Chicago, IL, 2006.
  25. Lyons-Ruth K, Yellin C, Melnick S, Atwood G. Childhood experiences of trauma and loss have different relations to maternal unresolved and hostile-helpless states of mind on the AAI. Attach Hum Dev 2003; 5: 330–52. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  26. Frigerio A, Costantino E, Ceppi E, Barone L. The hostile helpless state of mind among women from low-risk, poverty, and maltreatment samples. Attach Hum Dev 2013; 15: 424–42. PubMed Abstract | Publisher Full Text
  27. Lyons-Ruth K, Melnick S, Patrick M, Hobson RP. A controlled study of hostile-helpless states of mind among borderline and dysthymic women. Attach Hum Dev 2007; 9: 1–16. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  28. Brumariu LE, Obsuth I, Lyons-Ruth K. Quality of attachment relationships and peer relationship dysfunction among late adolescents with and without anxiety disorders. Journal of Anxiety Disorders 2013; 27: 116–124. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  29. First MB, Gibbon M, Spitzer RL, Williams JBW, Benjamin LS. User’s guide for the structured clinical interview for DSM-IV Axis II personality disorders (SCID-II). Washington, DC: American Psychiatric Press; 1997.
  30. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview for DSM-IV Axis I disorders–Clinical version (SCID-CV). Washington, DC: American Psychiatric Press; 1997.
  31. Horwitz AV, Widom CS, McLaughlin J, White HR. The impact of childhood abuse and neglect on adult mental health: a prospective study. J Health Soc Behav 2001; 42: 184–201. PubMed Abstract | Publisher Full Text
  32. Kendler KS, Aggen SH, Czajkowski N, Røysamb E, Tambs K, Torgersen S, et al. The structure of genetic and environmental risk factors for DSM-IV personality disorders: a multivariate twin study. Arch Gen Psychiatry 2008; 65: 1438–46. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  33. Straus M, Hamby S, Finkelhor D, Moore D, Runyan D. Identification of child maltreatment with the Parent-Child Conflict Tactics Scales: development and psychometric data for a national sample of American parents. Child Abuse Negl 1998; 22: 249–70. PubMed Abstract | Publisher Full Text
  34. Norris FH. Screening for traumatic stress: a scale of use in the general population. J Appl Soc Psychol 1990; 20: 1704–18. Publisher Full Text
  35. Dutra L, Jenei J, Long N, Holmes B, Lyons-Ruth K. Childhood traumatic experiences scale: using the AAI to assess trauma. Poster session presented at the 21st Annual Meeting of the International Society for Traumatic Stress Studies, Toronto, Canada, 2005.
  36. Bailey HN, Moran G, Pederson DR. Childhood maltreatment, complex trauma symptoms, and unresolved attachment in an at-risk sample of adolescent mothers. Attach Hum Dev 2007; 9: 139–61. PubMed Abstract | Publisher Full Text
  37. Dutra L, Bureau JF, Holmes B, Lyubchik A, Lyons-Ruth K. Quality of early care and childhood trauma: a prospective study of developmental pathways to dissociation. J Nerv Ment Dis 2009; 197: 383–90. PubMed Abstract | PubMed Central Full Text | Publisher Full Text
  38. George C, Kaplan N, Main M. Adult Attachment Interview. Unpublished manuscript, University of California at Berkeley, Berkeley, CA, 1985.
  39. Lyons Ruth K, Melnick S, Atwood G, Yellin C. Pervasively unintegrated/hostile-helpless states of mind on the Adult Attachment Interview: classification and coding manual. Unpublished Manuscript, Harvard Medical School, Boston, MA, 1995–2005.
  40. van Ijzendoorn M, Schuengal C, Bakermans-Kranenburg M. Disorganized attachment in early childhood: meta-analysis of precursors, concomitants and sequelae. Dev Psychopathol 1999; 11: 225–49. PubMed Abstract | Publisher Full Text
  41. Cohen J, Cohen P, West SG, Aiken LS. Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Lawrence Erlbaum; 2003.
  42. Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 2008; 40: 879–91. PubMed Abstract | Publisher Full Text
  43. Dearing E, Hamilton LC. V. Contemporary advances and classic advice for analyzing mediating and moderating variable. Monogr Soc Res Child Dev 2006; 71: 88–104.
  44. Torgersen S, Czajkowski N, Jacobson K, Reichborn-Kjennerud T, Røysamb E, Neale MC, et al. Dimensional representations of DSM-IV cluster B personality disorders in a population-based sample of Norwegian twins: a multivariate study. Psychol Med 2008; 38: 1617–25. PubMed Abstract | Publisher Full Text
  45. Bierer LM, Yehuda R, Schmeidler J, Mitropoulou V, New AS, Silverman JM, et al. Abuse and neglect in childhood: relationship to personality disorder diagnoses. CNS Spectr 2003; 8: 737–40. PubMed Abstract
  46. McKee L, Roland E, Coffelt N, Olson AL, Forehand R, Massari C, et al. Harsh discipline and child problem behaviors: the roles of positive parenting and gender. J Fam Violence 2007; 22: 187–96. Publisher Full Text
About The Authors

Brent Finger
Montana State University Billings
United States

Assistant professor of Psychology

Sooyeon Byun

United States

Sharon Melnick

United States

Karlen Lyons-Ruth

United States

Article Metrics

Metrics Loading ...

Metrics powered by PLOS ALM

Related Content