Journal of Personality Disorders, 32(1), 17-30, 2018

© 2018 The Guilford Press

BADOUD ET AL.
ATTACHMENT AND REFLECTIVE FUNCTIONING

ATTACHMENT AND REFLECTIVE FUNCTIONING

IN WOMEN WITH BORDERLINE PERSONALITY
DISORDER
Deborah Badoud, PhD, Paco Prada, MD, Rosetta Nicastro, MSc,
Charlotte Germond, MSc , Patrick Luyten, PhD,
Nader Perroud, MD, and Martin Debbané, PhD

Insecure attachment and impairments in reflective functioning (RF) are

thought to play a critical role in borderline personality disorder (BPD). In
particular, the mentalization-based model argues that insecure attachment
indirectly accounts for increased BPD features, notably via disruption of RF
capacities. Although the mediation relationship between attachment, RF,
and BPD is supported by previous evidence, it remains to be directly tested
in adults with BPD. In the current study, a sample of 55 female adult BPD
patients and 105 female healthy controls completed a battery of self-report
measures to investigate the interplay between attachment, RF capacities,
and BPD clinical status. Overall, the results showed that BPD patients pre-
dominantly reported insecure attachment, characterized by negative internal
working models of the self as unlovable and unimportant to others, and
decreased RF abilities. Our findings further indicated that actual RF capaci-
ties mediated the relationships between adult insecure attachment and BPD
clinical status.

Mentalization, or the processes sustaining our understanding of human ac-

tion as driven by mental states, has been operationalized in terms of reflec-
tive functioning (RF; Fonagy, Gergely, Jurist, & Target, 2002). Genuine RF
implies acknowledging the opaqueness of mental states, in combination with
the capacity to form relatively accurate models of the mind of self and oth-
ers (Fonagy et al., 2016). Reflectively thinking about behaviors and inter-
personal interactions helps us experience ourselves and others as predict-
able, manageable, and meaningful, rather than puzzling and unreliable. RF
is therefore one of the cornerstones of one’s sense of agency, and sustains
self-coherence and continuity over time and across situations (Fonagy &
Target, 1997). At the same time, RF significantly contributes to the adap-
From Developmental Clinical Psychology Unit, Faculty of Psychology, University of Geneva, Switzerland,
and Office Médico-Pédagogique Research Unit, Department of Psychiatry, University of Geneva School of
Medicine (D. B., M. D.); Program TRE, Service of Psychiatric Specialties, Department of Mental Health
and Psychiatry, University Hospitals of Geneva (P. P., R. N., C. G., N. P.); Department of Clinical, Educa-
tional and Health Psychology, University College London (P. L., M. D.); Faculty of Psychology and Edu-
cational Sciences, KU Leuven, Belgium (P. L.); and Department of Psychiatry, University of Geneva (N. P.).
Address correspondence to Deborah Badoud, Ph.D., 40, Boulevard du Pont d’Arve, 1205 Geneva, Swit-
zerland. E-mail: [email protected]

17
18 BADOUD ET AL.

tive management of distressing feelings when one is facing difficult interper-

sonal events (e.g., conflicts, losses), and previous studies have demonstrated
how trauma and neglect may significantly interfere in its development (e.g.,
Fonagy & Target, 1997). These different lines of research thus frame RF as
a psychological mechanism central to the consolidation of coherent identity
and self-regulation during development. Consequently, it has been assumed
that disrupted RF processes might be one crucial variable for understanding
the core features of borderline personality disorder (BPD), namely affective
dysregulation, impulsivity, and social dysfunctions. Impairments in RF can
manifest themselves in two ways, namely hypomentalizing and hypermental-
izing. Hypomentalizing describes the inability to consider complex models of
one’s own mind and/or those of others, implying high uncertainty about self
and other mental states. The opposite tendency, namely hypermentalizing,
refers to the generation of highly certain mentalistic representations of ac-
tions without appropriate evidence available to support them (Fonagy et al.,
2016). Specifically, the mentalization-based model for BPD argues that RF
dysfunction during adulthood critically mediates the relationship between
feelings of insecurity in close relationships and BPD psychopathology (Fon-
agy & Luyten, 2009).
Although RF might constitute an innate human ability, its degree of mat-
uration and its robustness in the face of high emotional interactions appear
to be critically influenced by primary attachment relationships (Fonagy et al.,
2002, Luyten, Fonagy, Lowyck, & Vermote, 2012). Indeed, the attachment
context supplies the setting in which the infant can be sensitized to inner self-
states, through his interactions with caregivers who strive to make sense of
his signals (e.g., figuring out whether a cry means the infant is sad or angry).
A secure environment may reflect caregiving that is consistently attuned to
a baby’s mental states, thereby favoring the development of RF and self-
regulation processes (e.g., Gergely & Unoka, 2008). Conversely, an insecure
attachment may suggest difficulties in the infant-parent dyad, which can un-
dermine the development of RF and self-regulation and confer an increased
risk for expressing early features of BPD (Fonagy et al., 2002).
One basic premise of attachment theory stipulates that over repeated
interaction patterns with his caregivers, the child develops cognitive repre-
sentations of self and others in relationships (i.e., internal working models;
Bowlby, 1973) that remain relatively stable and influential across the lifespan
(Bowlby, 1979). Attachment research has provided evidence that infant at-
tachment insecurity can later translate into insecure attachment patterns in
close relationships during adulthood (e.g., Fraley, 2002). Research on adult
attachment further sustains that RF is associated to the way people con-
sciously manage actual significant relationships. Adults securely attached to
their significant others (i.e., reporting positive and soothing working models
of self and others in close relationships) benefit from a robust capacity to
explicitly consider the mental states that lie behind their own and others’
behaviors, which helps them adaptively manage interpersonal stress. Con-
versely, adults with insecure anxious or avoidant bonds with close ones (re-
porting negative and dysregulating working models of self and others) tend
ATTACHMENT AND REFLECTIVE FUNCTIONING 19

to exhibit fragile reflective functioning with maladaptive fluctuations in re-

sponse to attachment arousal (e.g., Luyten et al., 2012).
To date, several studies have reported associations between BPD and
adult insecure attachment models (e.g., Agrawal, Gunderson, Holmes, & Ly-
ons-Ruth, 2004, for a review) as well as between BPD and RF dysfunctions
(e.g., Fonagy et al., 1996; Gullestad et al., 2012; Harari, Shamay-Tsoory,
Ravid, & Levkovitz, 2010; Preissler, Dziobek, Ritter, Heekeren, & Roepke,
2010). These studies underline the relevance of adult attachment patterns
sustained by negative self internal working model while other findings have
also highlighted the importance of negative internal working model of oth-
ers. As a whole, these results lead many to consider that BPD patients may
fluctuate in the valence of internal working models of self and others (for a
review, see Agrawal et al., 2004)
In terms of mentalization, BPD patients are typically described as strug-
gling to engage cognitive resources when attributing mental states (e.g., Fon-
agy et al., 1996; Gullestad et al., 2012; Harari et al., 2010; Preissler et al.,
2010). This feature in their mentalization profile may bias the interpretation
of mental states that motivate self and others’ action towards automatic and
affectively based reasoning, which strongly reduces their emotion-regulation
success (Fonagy & Luyten, 2009).
Recently, several studies have also started to examine the RF’s putative
mediation role linking insecure adolescent or adult attachment to BPD symp-
tom expression (Carlson, Egeland, & Sroufe, 2009; Fossati et al., 2009; Fos-
sati, Feeney, Maffei, & Borroni, 2011, 2014; Sharp et al., 2016). Fossati and
colleagues provided three cross-sectional studies on community samples, ex-
amining the relationships between avoidant and anxious attachment dimen-
sions and conceptual cousins of RF processes (i.e., alexythymia; mindful-
ness; mental state attribution and awareness of one’s own emotional states;
respectively, in Fossati et al., 2009, 2011, 2014). They specifically examined
the degree to which expressions of nonclinical borderline personality fea-
tures were associated with these processes. Their results suggest that insecure
adolescent or adult relationship styles, and particularly attachment patterns
involving predominantly anxious (Fossati et al., 2011, 2014) or poorly or-
ganized strategies (Fossati et al., 2009), contribute to higher expression of
borderline personality features through their negative (i.e., decreasing) as-
sociations with mediating RF processes (Fossati et al., 2011, 2014). In an-
other study involving a sample of 54 adolescents with a diagnosis of BPD
and 50 matched healthy controls, Deborde and colleagues (2012) show that
insecure attachment, particularly attachment patterns that imply negative
internal representation of oneself (i.e., preoccupied and fearful), is related
to BPD diagnosis via alexithymia features, namely difficulties in describing
and being aware of one’s own feelings, thereby hinting at impairments in
RF. Finally, based on a multiple mediational approach, Sharp and colleagues
(2016) investigated the cross-sectional interplay between attachment, objec-
tive performance to a sociocognitive task, self-reported emotion regulation,
and borderline features in a sample of 259 adolescent inpatients. They ob-
served that, unlike emotion dysregulation, excessive RF capacities indepen-
dently mediated the relation between attachment insecurity and level of BPD
20 BADOUD ET AL.

features in adolescence. Together, the six studies reported above provide sup-
porting evidence concerning the mediating role of RF as conceptualized by
the mentalization-based model of BPD. The available literature further mo-
tivates an examination of the Fonagy and colleagues’ hypothesis in a clinical
sample of adults BPD patients, using a direct measurement of RF, rather than
variables measuring its “conceptual cousins” (i.e., mindfulness, alexythimia;
Choi-Kain & Gunderson, 2008)
In this context, the present study aims to further examination of the re-
lationships between RF, attachment, and BPD symptoms, and to address the
limitations of previous work by testing the mediation model between current
attachment relationship models, RF, and BPD diagnosis in a sample of out-
patient women with BPD and healthy controls. Using a valid questionnaire
designed to assess the ability to conceive behavior as motivated by mental
states (i.e., the Reflective Functioning Questionnaire; RFQ; Badoud et al.,
2015; Fonagy et al., 2016), the current investigation examined the follow-
ing hypotheses. As a premise, in line with previous studies, we first postulate
that, in comparison with nonclinical controls, a higher proportion of BPD
patients will report insecure attachment, particularly of the anxious-preoc-
cupied prototype; we further expect that these patients will report a higher
degree of negative internal working models of the self. Second, we expect
BPD patients will report a lower mean level of RF in comparison to healthy
controls. Finally, in order to reach beyond current available literature, we
predict that RF will significantly mediate the relationship between attach-
ment insecurity and BPD clinical status. More specifically, we hypothesize
that the relationship between negative working models of self and the likeli-
hood of having received a diagnosis of BPD will be significantly mediated by
decreased RF capacities.

METHODS
PARTICIPANTS

Fifty-five women diagnosed with BPD (Mage = 30.63, SDage = 9.02) were re-
cruited from the University Hospitals of Geneva outpatient psychiatric ser-
vice specializing in the treatment of BPD. Participants were referred by their
physician or other medical services due to severe suicidal or self-damaging
behaviors and/or emotional dysregulation. Patients were interviewed by
a trained psychologist using the Structured Interview for Axis II Disorder
(SCID-II; First, Gibbon, Spitzer, Williams, & Smith Benjamin, 1994) BPD
part; only those fulfilling DSM-IV/5 criteria for BPD were accepted into the
program. Studies have shown that the SCID-II 2.0 has adequate inter-rater
and internal consistency reliability for diagnosing BPD (Maffei et al., 1997).
In addition, the French version of the Diagnostic Interview for Genetic Stud-
ies (DIGS; Preisig, Fenton, Matthey, Berney, & Ferrero, 1999) was used to as-
sess Axis I disorders. If needed, participants received psychopharmacological
treatment such as, for instance, antidepressant medication for a depressive
episode, as previously described (Perroud, Nicastro, Jermann, & Huguelet,
2012). Those with severe cognitive impairments, severe depressive episodes,
ATTACHMENT AND REFLECTIVE FUNCTIONING 21

mania and hypomania, and/or psychotic symptoms that required more inten-
sive care or hospitalization were not taken into the center’s program.
The control group of 161 non-referred women was recruited from the
Geneva community through written advertisements and word of mouth. The
sole inclusion criterion was age (at least 18 years); exclusion criteria com-
prised a clinical level of psychopathology, assessed by standardized scores
(t-score of 63 and above) on the French version of the Symptom Checklist-
90-Revised (SCL-90-R; Pariente & Guelfi, 1990) and the Adult Self-Report
scales (ASR; Achenbach & Rescorla, 2003) Participants who reported a clin-
ical score on the internalizing ASR subscale, on the externalizing ASR sub-
scale, or on the global severity index of the SCL-90-R were excluded from
the control group. Following these criteria, 56 participants were excluded.
The final sample encompassed 55 patients with BPD (Mage = 30.63, SDage
= 9.02) and 105 healthy controls (Mage = 23.26, SDage = 2.47). The cantonal
ethics committee for human research of Geneva and the ethics committee
of the psychology and educational sciences department of the University of
Geneva approved the study. All participants gave written informed consent
before participating.

MEASURES

All participants completed the following self-report questionnaires. To en-

sure that all subjects understood the items, trained clinical psychologists su-
pervised the process.

Relationship Questionnaire (RQ). The RQ (Bartholomew & Horowitz,

1991) was used to obtain a categorical and dimensional evaluation of the
four attachment patterns (i.e., secure, preoccupied, fearful, and dismissing).
The categorical ascription of attachment relies upon the RQ, a single-item
measure consisting of four short paragraphs corresponding to the four pro-
totypical attachment patterns listed above. The participants were instructed
to rate the four attachment descriptions on a 6-point scale (1 = “This is not
at all like me” to 6 = “This is absolutely like me”) and to choose the one that
best captured their individual attachment style. The single choice provided
the categorical value, while the continuous ratings were used to derive the
two dimensional values, namely, the valence of internal working models of
self and other. The two internal working models were chosen because they
fit with the predominant view of attachment as a dimensional construct and
because they specifically account for the fact that the individual may cor-
respond to a greater or lesser degree to different prototypical descriptions.
Moreover, the internal working models are believed to capture the mecha-
nisms behind the categories that are stable with age and may consist in mea-
sures suitable for adult participants (see Mikulincer & Shaver, 2010, for an
exhaustive view on attachment in adulthood).
According to Griffin and Bartholomew (1994), the self internal working
model was computed as follows: the rating score of patterns characterized by
a negative view of self (i.e., fearful and preoccupied) minus the rating score
of patterns characterized by positive self-models (i.e., secure and dismissing).
22 BADOUD ET AL.

Higher scores indicate consideration of the self as not deserving of help or

love from close others. The internal working model of others also consisted
of a difference score: the rating score of patterns characterized by positive
models of others (i.e., secure and preoccupied) minus the rating score of pat-
terns characterized by negative models of others (i.e., fearful and dismissing).
Higher scores designate a representation of significant others as being helpful
and reliable. Because this study was part of a larger study, the RQ was used
for time and feasibility concerns.

Reflective Functioning Questionnaire (RFQ). The RFQ (Badoud et al., 2015;

Fonagy et al., 2016) provides a valid self-reported measure of RF, which
strives to make RF assessment more readily accessible (i.e., less time-consum-
ing than a clinical interview such as the adult attachment interview; George,
Kaplan, & Main, 1985) and more straightforward than instruments devel-
oped for related constructs such as mindfulness (Choi-Kain & Gunderson,
2008). The validation studies of the RFQ showed that the factorial structure
that best accounted for the data is a two-dimension model, which trans-
lates the nature of minds (opacity) and the possible pitfalls in thinking about
minds: uncertainty about the value of information about minds and rigid
certainty about mentalistic attributes. They further supported the good psy-
chometric properties of the two subscale scores within different samples of
participants (Badoud et al., 2015; Fonagy et al., 2016). The Certainty About
Mental States (α = .74 and .70 in the control and BPD samples, respectively)
subscale consists of 6 items focusing on the extent to which individuals dis-
agree with statements such as “I don’t always know why I do what I do.”
All items are scored by participants on a 7-point Likert-type scale, ranging
from “completely disagree” to “completely agree.” Items are subsequently
rescored to capture more extreme levels of certainty, so that very low agree-
ments on this scale reflect excessive RF while some agreement reflects adap-
tive levels of certainty about mental states. The Uncertainty About Mental
States subscale (α = .65 and .75 in the control and BPD samples, respec-
tively), which in the extreme captures deficits of RF, also consists of 6 items
scored on the same 7-point Likert-type scale, with high scores reflecting a
stance characterized by an almost complete lack of knowledge about men-
tal states, and lower scores reflecting acknowledgment of the opaqueness of
one’s own mental states and those of others, typical of genuine RF.
In the current study, we then computed a difference score (i.e., certain-
ty minus uncertainty score) to bring greater clarity around the construct of
mentalization. The difference/total score merges the two subscale scores in
one dimension, which indeed would reflect “calibrated mentalization.” This
total score can be interpreted as follows: negative values (i.e., individuals
for whom the uncertainty score is higher than the certainty score) charac-
terize individuals who recognize that self and other actions are driven by
mind states. On the other hand, values in the positive range (i.e., individuals
for whom the certainty score is higher than the uncertainty score) designate
people who are predominantly confident that behaviors originate from in-
tentional mental states.   
ATTACHMENT AND REFLECTIVE FUNCTIONING 23

STATISTICAL ANALYSIS

Z-tests were calculated to compare the percentage of answers reported for

each attachment pattern (RQchoice) in the BPD and healthy control groups.
Analysis of variance (ANOVA) was performed to analyze the dimen-
sional attachment scores (RQ self/other internal working models) and level
of RF.
Based on the ANOVA results, variables that significantly differed be-
tween our two groups were introduced in the mediation analysis. Direct, in-
direct, and total effects of dimensional attachment and level of RF on clinical
status (BPD vs. healthy controls served as the dependent variable) were con-
currently estimated with an implemented script for SPSS software (Preacher
& Hayes, 2008). An alternative model that switches the mediator from the
independent variable was also tested. A bootstrap test for the indirect effect
(5,000 samples, confidence intervals set at 95%) was performed according
to Preacher and Hayes (2008) recommendations. Because the patient group
was significantly older than the control group (t = –6.12, p < .001), all statis-
tical analyses included age as a covariate.

RESULTS

ATTACHMENT IN BPD

Sixteen patients with BPD had missing data on the RQ questionnaire and
were excluded from the following analyses.
First, with regard to the forced choice of the prototypical attachment
category that best described participants in the two groups, 68.2% of the
control group could be categorized as secure, 11.2% as fearful, 7.5% as
preoccupied, and 13.1% as dismissing. In the BPD group, 25% could be
categorized as secure, 35% as fearful, 30% as preoccupied, and 10% as
dismissing. Percentage comparisons indicated significant differences between
the two groups for fearful, preoccupied (both higher in the BPD group), and
secure attachment (lower in the BPD group; all p < .05). No between-group
differences were found for the dismissing style (p > .05; Table 1).
Second, group comparisons performed on the two dimensions that un-
derlie the attachment prototypes (i.e., the valence of internal models of the
self and significant others) revealed that the BPD group overall had a more
negative self-model, F(2, 144) = 33.03, p = .00. No differences were found
between the groups for the internal model concerning others (p = .47; Table
1).

REFLECTIVE FUNCTIONING IN BPD

Mean comparisons performed on RF level indicated lower RFQ total score

in the BPD group than in control participants, F(2, 157) = 69.15, p <.001
(Table 1).
24 BADOUD ET AL.

TABLE 1. Means (and Standard Deviations) and Difference Significance Between BPD and Control
Groups for Attachment and Reflective Functioning Measures
BPD Control p value
(n = 55) (n = 105)
Age 30.63 (9.02) 23.26 (2.47) < .001
RQ secure 25% 68.2% < .05
RQ fearful 35% 11.2% < .05
RQ preoccupied 30% 7.5% < .05
RQ dismissing 10% 13.1% > .05
RQ self-model 1.62 (4.64) –2.65 (2.75) < .001
RQ other-model 0.69 (3.81) 1.19 (3.61) .463
RFQtot –4.40 (7.74) 7.47 (5.32) < .001
BPD: borderline personality disorder, RQ: Relationship Questionnaire, RFQtot: Reflective Functioning Questionnaire
total score.

CONTRIBUTION OF ATTACHMENT AND REFLECTIVE FUNCTIONING

TO BPD

Mediation analysis from the bootstrap analysis showed a significant indirect

effect (M = .25, SE = .08), with a 95% bias corrected confidence interval ex-
cluding zero (.14, .43), indicating that the association between the valence of
the internal working model of self and BPD diagnosis acts through the level
of RFQ total score. The direct effect and the other coefficient paths of the
model were also significant (see Figure 1 for path coefficients and p values).
Concerning the alternative reverse model (self-model as mediator, RFQ
as independent variable), the results revealed a significant indirect effect of
the RFQ in the relationship between the attachment measure and the clinical
status (M = –.08, SE = .07, 95% CI [–.21, .02]). The other coefficient paths
of the model were also all significant as summarized in Figure 2. 

DISCUSSION

In the present study, attachment, RF, and their relationship were investigated
in a clinical sample of women with BPD and a healthy control group. Our
results replicate prior studies showing an increased prevalence of insecure at-
tachment and impaired RF in BPD patients compared to the control group.
We provide the first direct evidence for RF capacity as a mediator in the rela-
tionship between attachment insecurity (internal working model of self) and
BPD diagnosis in a clinical sample. We will discuss the results sequentially in
light of the evidence concerning the different associations between insecure
attachment, RF, and BPD.
Consistent with our first hypothesis, the present results highlight the
prevalence of preoccupied-anxious attachment in BPD and underlie that
BPD might not be characterized by a unitary attachment style. Indeed, as
reported by previous studies based on the RQ (Brennan & Shaver, 1998;
Choi-Kain, Fitzmaurice, Zanarini, Laverdiere, & Gunderson, 2009; Dut-
ton, Saunders, & Starzomski, 1994; Hoermann, Clarkin, Hull, & Fertuck,
ATTACHMENT AND REFLECTIVE FUNCTIONING 25

FIGURE 1. Unstandardized path coefficients (and standard errors)

for the mediation model with attachment as independant variable,
reflective functioning as mediator and borderline personality disor-
der diagnosis as dependant variable c = total effect of RQ on clinical
status; c′ = direct effect of RQ on clinical status.

2004), our data supports the fact that a fearful attachment pattern is also
relevant to BPD. Preoccupied and fearful patterns both imply negative work-
ing models of self, namely a representation of oneself as being unworthy and
unacceptable that goes along with excessive anxiety and dependency in close
relationships. Nevertheless, preoccupied and fearful patterns are different
in terms of internal working models of others. Preoccupied adults maintain
positive working models of others and actively seek approval to validate
their own fragile sense of self-worth, while fearful individuals exhibit a per-
vasive sense of interpersonal distrust (Bartholomew, 1990; see Mikulincer
& Shaver, 2010, for an exhaustive view on attachment in adulthood). As
a consequence, fearfully attached individuals exhibit approach/avoidance
conflicts in relation to significant others (Dutton et al., 1994; Pietromonaco
& Feldman Barret, 2000). Unsurprisingly, these categorical results preclude
our dimensional investigation of attachment; whereas no difference between
groups in terms of working models of others was found, the BPD sample
reported negative working models of self in comparison to controls, imply-
ing views of self as unimportant and undesirable in the eyes of significant
others. Interestingly, these results might be consistent with recent literature
about shame-proneness in BPD (Gratz, Rosenthal, Tull, Lejuez, & Gunder-
son, 2010). Indeed, shame-proneness refers to the individual tendency to
easily feel ashamed due to a global sense of self as a “bad” person (Lewis,
1971). This description echoes the negative internal working model of the
self observed in the current BPD sample. Shame is a predominant emotion in
BPD and has been linked to the most serious symptoms of BPD (e.g., suicidal
behaviors or nonsuicidal self-injury), lower quality of life and self-esteem,
and increased hostility (Rüsch et al., 2007). Specifically, we might specu-
late that one path to exaggerated shame-proneness in BPD arises from nega-
tive internalized expectations about oneself in relationships. This hypothesis
could potentially be a fruitful avenue of research in BPD psychopathology.
Second, the observed RF impairments in our sample of women with
BPD suggest that these participants might experience the inherent relation-
ship between actions and mental states more tenuously than control par-
ticipants. The present data contributes to previous work on RF capacities
26 BADOUD ET AL.

FIGURE 2. Unstandardized path coefficients (and standard errors) for

the reverse mediation model with reflective functioning as indepen-
dant variable, attachment as mediator and borderline personality
disorder diagnosis as dependant variable c = total effect of RQ on
clinical status; c′ = effect of RQ on clinical status.

in two ways. First, they provide a direct investigation of subjective (self-

reported) RF capacities in individuals with BPD. So far, studies about RF in
BPD have mostly relied on proxy measures, such as experimental tasks and/
or self-reports, primarily designed to assess constructs overlapping with RF
(Choi-Kain & Gunderson, 2008). Second, they add compelling evidence for
the existence of significant impairments in RF in BPD. The current scientific
and clinical literature reports impairments in RF processes that may express
themselves through seemingly reduced (notably in the present study) or over-
active mentalization (e.g., Fertuck et al., 2009; Franzen et al., 2011; Frick
et al., 2012; Krohn, 1974; Preissler et al., 2010; Sharp, 2014; Sharp et al.,
2011). A number of variables may moderate the relationship between BPD
and RF dysfunction, such as measurement methods (subjective self-report
vs. objective experimental task), level of participant’s arousal (high vs. low),
or the developmental period (adolescent vs. adult sample). For instance, a
recent study carried out in a community sample showed higher mean levels
of self-reported RF capacities in the adult group compared to the adolescent
group (Badoud, Menghetti, Eliez, & Debbané, 2016). The current results
suggest that women with BPD, when filling out a self-report in non-arousing
conditions, conserve a degree of awareness of their difficulties to link behav-
iors with mental states. However, the relative influence of within-individual
and situational variables on RF functioning in BPD should be systematically
examined in future assessments to bridge the potentially contradictory per-
spectives on RF processes (hypo- vs. hyper-functioning) in BPD.
Our final analysis examined how attachment insecurity and RF are as-
sociated with a BPD diagnosis in women. Consistent with our hypothesis, we
observed that the representation of the self as undesirable and insignificant
to others significantly relates to the degree of uncertainty when inferring
mental states to understand self and others’ behaviors. In turn, uncertainty
in the value of mental state knowledge increased the probability of belong-
ing to the BPD group. The present results are in line with previous studies
that emphasize the mediating role of RF in the specific relationship between
ATTACHMENT AND REFLECTIVE FUNCTIONING 27

attachment disturbances and BPD features expression (Carlson et al., 2009;

Deborde et al., 2012; Fossati et al., 2009, 2011), as well as those reporting
on the role of RF in the more general association between childhood adver-
sity and adult functioning (e.g., RF as mediating the relation between child-
hood maltreatment and personality disorder; Bouchard et al., 2008; Chiesa
& Fonagy, 2014; for other examples, see also Macintosh, 2013, for a review;
or see Stein, Fonagy, Wheat, Kipp, & Gerber, 2004, and Taubner & Curth,
2013, for specific examples).
Developmental studies suggest a prospective link between early attach-
ment, adolescent RF, and adult BPD symptoms (Carlson et al., 2009). Our
results further suggest that negative views of the self in actual relationship
with close persons contribute to one’s clinical status through his or her pre-
vailing level of RF. Put another way, our results may lead to the belief that
impaired RF processes are not only involved in the BPD developmental path
but also in the maintenance of adult BPD difficulties of clinical significance.
Of note, the present study also emphasized that the reverse indirect model
(i.e., with attachment as mediator) is equally significant to the original de-
velopmental model. This emphasizes the need for a longitudinal prospective
design to illuminate the directionality of the link.  
Some limitations to this study warrant consideration. First, like most
studies in the field of BPD, the sample consisted exclusively of females, which
constrains the generalization of the results. The SCID-II interviews were not
subjected to inter-rater reliability procedures; however, to guarantee the reli-
ability of the assessment, the SCID-II was administered only by researchers
who were also trained as clinical psychologists, with several years of practice
in the assessment and care of BPD patients. We further need to mention that
the mediation analyses were performed on only a subsample of participants,
as a part of the BPD patients did not report attachment data. The current
study offers preliminary results that need to be longitudinally replicated in
integrating other measurement methods (e.g., using objective reflective func-
tioning performances) to assess the role of RF as a maintenance factor. It
should also include additional clinical samples to test the specificity of the
relationships highlighted here to BPD, as compared to regular community
controls (i.e., including a group from the general population that is wider
than just students).
Despite these limitations, the current investigation opens promising em-
pirical questions that can be summed up as follows. The study of RF and
attachment in adults with BPD might benefit from studies that aim to specify
the role of RF in the maintenance of BPD across adulthood. BPD symp-
tomatology is thought to decrease over time, with a much higher percent-
age of remission than was previously expected (Gunderson et al., 2011). A
better understanding of the factors involved in the persistence of BPD and
how these factors participate in chronic self-image disturbance, interpersonal
relationship instability, or emotion dysregulation may be potential mecha-
nisms of change and therefore clinically relevant.
28 BADOUD ET AL.

REFERENCES

Achenbach, T. M., & Rescorla, L. A. (2003). Man- distress. Personality and Mental Health,
ual for the ASEBA adult forms and profiles. 8(1), 52–66. doi:10.1002/pmh.1245
Burlington, VT: University of Vermont. Choi-Kain, L. W., Fitzmaurice, G. M., Zanarini,
Agrawal, H. R., Gunderson, J., Holmes, B. M., & M. C., Laverdiere, O., & Gunderson, J.
Lyons-Ruth, K. (2004). Attachment studies G. (2009). The relationship between self-
with borderline patients: A review. Har- reported attachment styles, interpersonal
vard Review of Psychiatry, 12(2), 94–104. dysfunction, and borderline personality
doi:10.1080/10673220490447218 disorder. Journal of Nervous and Mental
Badoud, D., Luyten, P., Fonseca-Pedrero, E., Eliez, Disease, 197(11), 816–821. doi:10.1097/
S., Fonagy, P., & Debbané, M. (2015). NMD.0b013e3181bea56e
The French version of the Reflective Func- Choi-Kain, L. W., & Gunderson, J. G. (2008).
tioning Questionnaire: Validity data for Mentalization: Ontogeny, assessment, and
adolescents and adults and its association application in the treatment of border-
with non-suicidal self-injury. PLoS One, line personality disorder. American Jour-
10(12), e0145892. doi:10.1371/journal. nal of Psychiatry, 165(9), 1127–1135.
pone.0145892 doi:10.1176/appi.ajp.2008.07081360
Badoud, D., Menghetti, S., Eliez, S., & Debbané, Deborde, A. S., Miljkovitch, R., Roy, C., Dugré-
M. (2016). Evaluating the links between Le Bigre, C., Pham-Scottez, A., Speranza,
mentalization, empathy and perspective- M., & Corcos, M. (2012). Alexithymia as
taking and their relevance to borderline a mediator between attachment and the
personality features. Manuscript submitted development of borderline personality dis-
for publication. order in adolescence. Journal of Personality
Bartholomew, K. (1990). Avoidance of intimacy: Disorders, 26(5), 676–688. doi:10.1521/
An attachment perspective. Journal of So- pedi.2012.26.5.676
cial and Personal Relationships, 7, 147– Dutton, D. G., Saunders, K., & Starzomski, A.
178. (1994). Intimacy-anger and insecure at-
Bartholomew, K., & Horowitz, L. M. (1991). At- tachment as precursors of abuse in inti-
tachment styles among young adults: A test mate relationships. Journal of Applied
of a four-category model. Journal of Per- Social Psychology, 24(15), 1367–1386.
sonality and Social Psychology, 61(2), 226– doi:10.1111/j.1559-1816.1994.tb01554.x
244. doi:10.1037/0022-3514.61.2.226 Fertuck, E. A., Jekal, A., Song, I., Wyman, B.,
Bouchard, M.-A., Target, M., Lecours, S., Fongay, Morris, M. C., Wilson, S. T., . . . Stanley,
P., Tremblay, L.-M., Schachter, A., & Stein, B. (2009). Enhanced “Reading the Mind
H. (2008). Mentalization in adult attach- in the Eyes” in borderline personality dis-
ment narratives: Reflective functioning, order compared to healthy controls. Psy-
mental states, and affect elaboration com- chological Medicine, 39(12), 1979–1988.
pared. Psychoanalytic Psychology, 25(1), doi:10.1017/S003329170900600X
47–66. doi:10.1037/0736–9735.25.1.47 First, M. B., Gibbon, M., Spitzer, R. L., Williams,
Bowlby, J. (1973). Attachment and loss (vol. 2). J. B. W., & Smith Benjamin, L. (1994).
New York, NY: Basic Books. Structured Clinical Interview for DSM-IV
Bowlby, J. (1979). The making and breaking of Personality Disorders (SCID-II). Washing-
affectional bonds. London, UK: Tavistock. ton, DC: American Psychiatric Association.
Brennan, K. A., & Shaver, P. R. (1998). Attach- Fonagy, P., Gergely, G., Jurist, E. L., & Target, M.
ment styles and personality disorders: Their (2002). Affect regulation, mentalization,
connections to each other and to parental and the development of the self. New York,
divorce, parental death, and perceptions NY: Other Press.
of parental caregiving. Journal of Person- Fonagy, P., Leigh, T., Steele, M., Steele, H., Ken-
ality, 66(5), 835–878. doi:10.1111/1467- nedy, R., Mattoon, G., . . . Gerber, A.
6494.00034 (1996). The relation of attachment status,
Carlson, E. A., Egeland, B., & Sroufe, L. A. psychiatric classification, and response
(2009). A prospective investigation of the to psychotherapy. Journal of Consulting
development of borderline personality and Clinical Psychology, 64(1), 22–31.
symptoms. Development and Psychopa- doi:10.1037/0022-006X.64.1.22
thology, 21(4), 1311–1334. doi:10.1017/ Fonagy, P., & Luyten, P. (2009). A developmen-
S0954579409990174 tal, mentalization-based approach to the
Chiesa, M., & Fonagy, P. (2014). Reflective func- understanding and treatment of border-
tion as a mediator between childhood ad- line personality disorder. Development
versity, personality disorder and symptom and Psychopathology, 21(4), 1355–1381.
doi:10.1017/S0954579409990198
ATTACHMENT AND REFLECTIVE FUNCTIONING 29

Fonagy, P., Luyten, P., Moulton-Perkins, A., Lee, Gratz, K. L., Rosenthal, M. Z., Tull, M. T., Lejuez,
Y. W., Warren, F., Howard, S., … Lowyck, C. W., & Gunderson, J. G. (2010). An ex-
B. (2016). Development and validation of a perimental investigation of emotional re-
self-report measure of mentalizing: The Re- activity and delayed emotional recovery
flective Functioning Questionnaire. PLoS in borderline personality disorder: The
One, 11(7), e0158678. doi:10.1371/jour- role of shame. Comprehensive Psychia-
nal.pone.0158678 try, 51(3), 275–285. doi:10.1016/j.comp-
Fonagy, P., & Target, M. (1997). Attachment psych.2009.08.005
and reflective function: Their role in self- Griffin, D., & Bartholomew, K. (1994). Models of
organization. Development and Psycho- the self and other: Fundamental dimensions
pathology, 9(4), 679–700. doi:10.1017/ underlying measures of adult attachment.
S0954579497001399 Journal of Personality and Social Psychol-
Fossati, A., Acquarini, E., Feeney, J. A., Borroni, ogy, 67(3), 430–445. doi:10.1037//0022-
S., Grazioli, F., Giarolli, L. E., . . . Maffei, C. 3514.67.3.430
(2009). Alexithymia and attachment insecu- Gullestad, F. S., Wilberg, T., Klungsøyr, O., Jo-
rities in impulsive aggression. Attachment hansen, M. S., Urnes, O., & Karterud, S.
& Human Development, 11(2), 165–182. (2012). Is treatment in a day hospital step-
doi:10.1080/14616730802625235 down program superior to outpatient indi-
Fossati, A., Feeney, J., Maffei, C., & Borroni, S. vidual psychotherapy for patients with per-
(2011). Does mindfulness mediate the asso- sonality disorders? 36-month follow-up of
ciation between attachment dimensions and a randomized clinical trial comparing dif-
borderline personality disorder features? A ferent treatment modalities. Psychotherapy
study of Italian non-clinical adolescents. At- Research, 22(4), 426–441. doi:10.1080/10
tachment & Human Development, 13(6), 503307.2012.662608
563–578. doi:10.1080/14616734.2011.60 Gunderson, J. G., Stout, R. L., McGlashan, T. H.,
8993 Shea, M. T., Morey, L. C., Grilo, C. M., .
Fossati, A., Feeney, J., Maffei, C., & Borroni, S. . . Skodol, A. E. (2011). Ten-year course
(2014). Thinking about feelings: Affective of borderline personality disorder: Psy-
state mentalization, attachment styles, and chopathology and function from the Col-
borderline personality disorder features laborative Longitudinal Personality Disor-
among italian nonclinical adolescents. ders study. Archives of General Psychiatry,
Psychoanalytic Psychology, 31(1), 41–67. 68(8), 827–837. doi:10.1001/archgenpsy-
doi:10.1037/A0033960 chiatry.2011.37
Fraley, R. C. (2002). Attachment stability from Harari, H., Shamay-Tsoory, S. G., Ravid, M., &
infancy to adulthood: Meta-analysis and Levkovitz, Y. (2010). Double dissociation
dynamic modeling of developmental between cognitive and affective empathy in
mechanism. Personality and Social Psy- borderline personality disorder. Psychiatry
chology Review, 6, 123–151. doi:10.1207/ Research, 175(3), 277–279. doi:10.1016/j.
S15327957PSPR0602_03 psychres.2009.03.002
Franzen, N., Hagenhoff, M., Baer, N., Schmidt, A., Hoermann, S., Clarkin, J. F., Hull, J. W., & Fer-
Mier, D., Sammer, G., . . . Lis, S. (2011). tuck, E. A. (2004). Attachment dimensions
Superior “theory of mind” in borderline as predictors of medical hospitalizations
personality disorder: An analysis of inter- in individuals with DSM IV cluster B per-
action behavior in a virtual trust game. sonality disorders. Journal of Personality
Psychiatry Research, 187(1–2), 224–233. Disorders, 18(6), 595–603. doi:10.1521/
doi:10.1016/j.psychres.2010.11.012 pedi.18.6.595.54791
Frick, C., Lang, S., Kotchoubey, B., Sieswerda, S., Krohn, A. (1974). Borderline “empathy” and dif-
Dinu-Biringer, R., Berger, M., . . . Barnow, ferentiation of object representations: A
S. (2012). Hypersensitivity in borderline contribution to the psychology of object
personality disorder during mindreading. relations. International Journal of Psycho-
PLoS One, 7(8), e41650. doi:10.1371/jour- analytic Psychotherapy, 3(2), 142–165.
nal.pone.0041650 Lewis, H. B. (1971). Shame and guilt in neurosis.
George, C., Kaplan, N., & Main, M. (1985). Adult Psychoanalytic Review, 58(3), 419–438.
Attachment Interview. Unpublished manu- Luyten, P., Fonagy, P., Lowyck, B., & Vermote, R.
script. University of California, Berkeley. (2012). Assessment of mentalization. In A.
Gergely, G., & Unoka, Z. (2008). Attachment, af- W. Bateman & P. Fonagy (Eds.), Handbook
fect-regulation and mentalization: The de- of mentalizing in mental health practice
velopmental origins of the representational (pp. 43–66). Washington, DC: American
affective self. In C. Sharp, P. Fonagy, & I. Psychiatric Publishing.
Goodyer (Eds.), Social cognition and devel- Macintosh, H. B. (2013). Mentalizing and its role
opmental psychopathology (pp. 303–340). as a mediator in the relationship between
Oxford, UK: Oxford University Press. childhood experiences and adult function-
30 BADOUD ET AL.

ing: Exploring the empirical evidence. Psi- Clinical Neuroscience, 249(4), 174–179.
hologija, 46, 193–212. doi:10.1007/s004060050084
Maffei, C., Fossati, A., Agostoni, I., Barraco, A., Preissler, S., Dziobek, I., Ritter, K., Heekeren, H.
Bagnato, M., Deborah, D., Namia, C., No- R., & Roepke, S. (2010). Social cognition
vella, L., & Petrachi, M. (1997). Interrater in borderline personality disorder: Evidence
reliability and internal consistency of the for disturbed recognition of the emotions,
structured clinical interview for DSM-IV thoughts, and intentions of others. Fron-
axis II personality disorders (SCID-II), ver- tiers in Behavioral Neuroscience, 4, 1–8.
sion 2.0. Journal of Personality Disorders, doi:10.3389/fnbeh.2010.00182
11, 279–284. Rüsch, N., Lieb, K., Göttler, I., Hermann, C.,
Mikulincer, M., & Shaver, P. R. (2010). Attach- Schramm, E., Richter, H., . . . Bohus, M.
ment in adulthood: Structure, dynamics, (2007). Shame and implicit self-concept in
and change. New York, NY: Guilford. women with borderline personality disor-
Pariente, P., & Guelfi, J. D. (1990). Version fran- der. American Journal of Psychiatry, 164(3),
çaise de la SCL-90-R: Inventaire d’auto- 500–508. doi:10.1176/appi.ajp.164.3.500
évaluation psychopathologique chez les Sharp, C. (2014). The social-cognitive basis of
adultes. Première partie: Inventaires multi- BPD: A theory of hypermentalizing. In C.
dimensionnels [French version of the SCL- Sharp & J. L. Tackett (Eds.), Handbook of
90-R: A self-preported psychopathological borderline personality disorder in children
symptom checklist for adults. Part One: and adolescents (pp. 211–228). New York,
Multidimensional inventories]. Psychiatry NY: Springer.
and Psychobiology, 5, 49–63. Sharp, C., Pane, H., Ha, C., Venta, A., Patel, A.
Perroud, N., Nicastro, R., Jermann, F., & Hugue- B., Sturek, J., & Fonagy, P. (2011). Theory
let, P. (2012). Mindfulness skills in border- of mind and emotion regulation difficul-
line personality disorder patients during ties in adolescents with borderline traits.
dialectical behavior therapy: Preliminary Journal of the American Academy of Child
results. International Journal of Psychiatry and Adolescent Psychiatry, 50(6), 563–573.
in Clinical Practice, 16(3), 189–196. doi:10 doi:10.1016/j.jaac.2011.01.017
.3109/13651501.2012.674531 Sharp, C., Venta, A., Vanwoerden, S., Schramm,
Pietromonaco, P. R., & Feldman Barret, L. (2000). A., Ha, C., Newlin, E., . . . Fonagy, P.
The internal working models concept: (2016). First empirical evaluation of the
What do we really know about the self in link between attachment, social cognition
relation to others? Review of General Psy- and borderline features in adolescents.
chology, 4(2), 155–175. doi:10.1037/1089- Comprehensive Psychiatry, 64, 4–11.
2680.4.2.155 Stein, H., Fonagy, P., Wheat, K., Kipp, M., & Ger-
Preacher, K. J., & Hayes, A. F. (2008). Asymp- ber, A. (2004). Does mentalizing capacity
totic and resampling strategies for assess- mediate the relationship between develop-
ing and comparing indirect effects in mul- mental adversity and adult social function-
tiple mediator models. Behavior Research ing? Journal of the American Psychoana-
Methods, 40(3), 879–891. doi:10.3758/ lytic Association, 52, 1217–1221.
BRM.40.3.879 Taubner, S., & Curth, C. (2013). Mentalization
Preisig, M., Fenton, B. T., Matthey, M. L., Berney, mediates the relation between traumatic
A., & Ferrero, F. (1999). Diagnostic Inter- experiences and aggressive behavior in
view for Genetic Studies (DIGS): Inter-rater adolescence. Psihologija, 46(3), 177–192.
and test-retest reliability of the French ver- doi:10.1177/0003065112450148
sion. European Archives of Psychiatry and