MEYER ET AL.
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS
WHAT’S IN A (NEUTRAL) FACE? PERSONALITY
DISORDERS, ATTACHMENT STYLES, AND THE
APPRAISAL OF AMBIGUOUS SOCIAL CUES
Björn Meyer, PhD, Paul A. Pilkonis, PhD,
and Christopher G. Beevers, PhD
Attachment theory suggests that borderline, avoidant, and schizoid per-
attachment styles. The variations of insecure attachment in these disor-
ders, in turn, might influence how others are perceived or appraised. To
test these ideas, 176 college students completed questionnaires measur-
ing attachment styles and personality disorder features, and each rated
10 emotionally neutral faces on 18 bipolar appraisal dimensions. As ex-
pected, borderline personality was associated with anxious attachment,
avoidant personality was associated with anxious and avoidant attach-
ment, and schizoid personality was weakly associated only with avoidant
attachment. Furthermore, path analyses showed that borderline and
avoidant personality features were associated with anxious attachment,
which in turn related to negative face appraisals (e.g., tendencies to rate
faces as less friendly and more rejecting). Discussion highlights the po-
tential of simultaneously studying biases in social information
processing and attachment disturbances among individuals with
In the search for parsimonious, clinically and empirically inspiring models,
conceptual framework (e.g., Bartholomew, Kwong, & Hart, 2001; Fonagy,
Gergely, Jurist, & Target, 2002; Lyddon & Sherry, 2001; Meyer, Pilkonis,
Proietti, Heape, & Egan, 2001). Bowlby’s (e.g., 1969, 1973, 1980) theory ap-
peals to many personality disorder researchers for a variety of reasons, in-
science, control systems theories, and ethological approaches; and—per-
haps most importantly—its rapidly growing empirical base (e.g., Cassidy &
Shaver, 1999; Simpson & Rholes, 1998).
According to attachment theory, early child-caregiver interactions influ-
ence interpersonal relatedness and psychosocial adjustment throughout
Journal of Personality Disorders, 18(4), 320-336, 2004
© 2004 The Guilford Press
From the University of Surrey Roehampton (B.M.), the University of Pittsburgh School of Medi-
cine (P.A.P.), and Brown University and Butler Hospital (C.G.B.).
Address correspondence to Björn Meyer, PhD, School of Psychology and Therapeutic Studies,
University of Surrey Roehampton, Whitelands College, West Hill, London SW15 3SN, UK;
the lifespan (e.g., Cassidy, 1999). Depending on the degree to which care-
givers are sensitive and responsive to children’s needs and distress, chil-
dren are thought to develop mental representations (internal working
models) of others, of the self, and of relationships in general, which later
guide and direct the way they experience and approach others (cf.
Ainsworth, Blehar, Waters, & Wall, 1978; Bretherton & Munholland, 1999;
Cassidy, 1999). These attachmentmodels thus influence adults’ capacity to
establish stable, mutually nurturing relationships (Feeney, 1999; Hazan &
In adults, individual differences in attachment styles can be construed as
variations in two orthogonal dimensions: (1) The tendency to experience at-
tachment anxiety—to worry about rejection or abandonment—and (2) the
tendency to avoid intimacy and prefer self-sufficiency (Brennan, Clark, &
Shaver, 1998; Fraley, Waller, & Brennan, 2000). Individual differences on
these dimensions can be used to derive categorical classifications of attach-
ment, such as the secure, preoccupied, fearful, and dismissing types
(Brennan, Clark, & Shaver, 1998). For example, a combination of high at-
tachment anxiety and low avoidance corresponds to preoccupied attach-
ment, whereas low levels on both dimensions indicate secure attachment.
Insecure forms of attachment can be viewed as risk factors that interact
with other vulnerabilities to facilitate the development of enduring adult
personality pathology (e.g., Bartholomew et al., 2001). One process by
which insecure attachment may confer vulnerability to personality pathol-
ogy relates to biases in social information processing shown by people with
attachment disturbances. Several studies have found that insecurely at-
Robin, & Innes-Ker, 2002). For example, fearfully attached applicants tend
to view college examiners more negatively (Horppu & Ikonen-Varila, 2001),
tion (Niedenthal et al., 2002). However, the specific relations between per-
sonality disorders, insecure forms of attachment, and biases in social
information processing remain largely unexplored.
In theory, some personality disorders can be construed as constellations
of insecure attachment (e.g., Bartholomew et al., 2001; Lyddon & Sherry,
2001; Meyer & Pilkonis, in press). Limitingour focus hereon borderline per-
sonality, symptoms such as frantic efforts to avoid abandonment may re-
semble approach-oriented responses to chronically activated attachment
anxiety, and the borderline-characteristic unstable sense of identity can be
regarded as a reflection of conflicted or poorly integrated models of the self
(Bartholomew et al., 2001). Attachment representations of others in border-
line personality may also be negative or conflicted, which is understandable
given the abuse histories typically reported by these individuals (e.g., Yen et
al.,2002; Zanarinietal.,1997). Indeed, thepersistentmaltreatmentexperi-
enced by children at risk for borderline symptoms may interfere with their
ability to appreciate and reflect on other people’s mental states (e.g., Fonagy
et al., 2002) or may lead to tendencies to construe others’ motives as
malicious (e.g., Arntz & Veen, 2001).
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS 321
Consistent with such theorizing, empirical evidence has found that inse-
cure forms of attachmentand negative views of the self and others dominate
among people with borderline personality. For example, studies based on
self-report and interview-based assessments suggest that preoccupied and
fearful forms of attachment are common (Brennan & Shaver, 1998; Fonagy
1994). Biases in social information processing have also been documented.
For example, patients with borderline symptoms tended to exhibit a nega-
tive bias to emotionally neutral faces (Wagner & Linehan, 1999); to evaluate
interpersonal situations in extreme terms (Veen & Arntz, 2000); and to in-
terpret actions of people shown in film clips more negatively than others
(Arntz& Veen,2001). They also tendtobelieve thatotherscannotbe trusted
and that they themselves are needy and weak (Butler, Brown, Beck, &
line personality as a constellation of cognitive-affective-motivational pro-
cesses that arise from early adversity, persist in the form of
anxious-preoccupied attachment, and interfere with the person’s ability to
establish stable, mutually nurturingrelationships (e.g.,Bartholomew et al.,
2001; Meyer & Pilkonis, in press).
Whereas borderline personality might reflect predominantly an anxious
attachment to others, other personality disorders might be associated with
different attachment constellations. Avoidant personality disorder, for in-
stance, contrasts with borderline because of the more dominant interper-
sonal distancing. People with avoidant personality crave intimacy but are
simultaneously fearful of rejection and therefore tend to avoid interpersonal
situations thatentailthepotentialforrejectionor humiliation(e.g.,Millon &
Davis, 1996). This combination of attachment anxiety and interpersonal
avoidance fits with the fearful attachment pattern, where both attachment
anxiety and avoidance are elevated (Bartholomew et al., 2001; Lyddon &
Sherry, 2001). The negative mental representations of the self and others in
avoidant personality may also stem from adversity encountered in child-
hood (e.g., Stravynski, Elie, & Franche, 1989; Meyer & Carver, 2000), al-
borderline personality (e.g., Millon & Davis, 1996; Yen et al., 2002).
The few empirical studies thathave examined links between avoidant per-
sonality and attachment have typically confirmed the preponderance of
fearful attachment (e.g., Brennan & Shaver, 1998; Sheldon & West, 1990).
There is also evidence documenting characteristic biases in interpersonal
perception and interpretation among people with avoidant features. For ex-
ample, Dreessen, Arntz, Hendriks, Keune, & van den Hout (1999) showed
that avoidant personality disorder features related to characteristic
avoidant beliefs, which in turn predicted negative biases on an implicit in-
ference task. In this task, those with avoidant beliefs were more likely to in-
terpret ambiguous actions of others as indicative of motivated rejection
(Dreessen et al., 1999). In sum, theory and evidence suggest that avoidant
personality disorder can be understood as a constellation of cognition, af-
fect, and motivation that corresponds to fearful attachment and might arise
from early interpersonal adversity. The highly anxious and avoidant (i.e.,
fearful) attachment in avoidant personality disorder might predispose these
322 MEYER ET AL.
individuals to appraise and interpret other people as rejecting, even when
there is little evidence to substantiate such views. These biases in social in-
formation processing are thought to be among the critical processes that
maintain the characteristic pattern of active social avoidance (e.g., Meyer &
Pilkonis, in press).
A third disorder that we consider in this study—schizoid personality dis-
order—has also been interpreted from an attachment perspective
(Bartholomew et al., 2001; Lyddon & Sherry, 2001). Whereas borderline
personality might reflect predominant preoccupied attachment, and
avoidant personality predominant fearful attachment, the attachment con-
stellation in schizoid personality is less clear. Empirically, schizoid person-
ality has been linked with measures of dismissing or detached attachment
(e.g., Brennan & Shaver, 1998; West, Rose, & Sheldon-Keller, 1994) and
some theorists have similarly interpreted schizoid aloofness and disinterest
as an indicator of dismissing attachment (Lyddon & Sherry, 2001). Others,
however, have argued that schizoid personality might result from underde-
veloped attachment needs rather than specific forms of insecure attach-
ment. Bartholomew et al. (2001) noted that, “of the many dismissing
individuals wehaveassessed instudiesovertheyears,wehavenotseenone
individual whom we would consider schizoid” (p. 220). This suggests that
schizoid personalities might not hold strongly negative models of the self
and others in the way that insecurely attached individuals do, and their ap-
praisals of interpersonal cues might similarly not be strongly biased. What
characterizes and maintains their social alienation, then, might be an im-
paired capacity and motivation to form attachments, rather than negative
models of self or others that motivate defensive interpersonal distancing.
In this study, we aimed to examine associations among borderline,
avoidant, and schizoid personality disorder, anxious and avoidant attach-
ment, and biases in the appraisal of ambiguous social cues. We hypothe-
sized that borderline personality features would be linked primarily with
anxious attachment, avoidant personality with both anxious and avoidant
attachment, and schizoid personality with neither. Attachment styles have
been shown to be related to biases in social perception and appraisal, and
we therefore hypothesized that both anxious and avoidant attachment
would relate to negative interpersonal appraisals. We also explored whether
personality disorder features—especially borderline and avoidant—would
be directly linked withnegative interpersonal appraisals, or whetherthisre-
lationship would be indirect,perhaps via the mediating pathway of insecure
attachment. These hypotheses were tested in a nonclinical sample of young
To measure biases in social information processing, we showed partici-
pants a series of emotionally neutral faces and asked them to respond to
questions about how they viewed or appraised these faces (e.g., friendly,
mean-spirited), whether they felt positive toward them (e.g., whether the
person “could be a friend”), and whether they anticipated that the person
might like or reject them. We limited our stimulus set to emotionally neutral
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS323
faces because such faces are more ambiguous than the strongly emotional
faces that are often used in similar studies. A previous study had also sug-
gested that borderline personality disorder specifically predisposes people
to respond with biased appraisals to neutral faces (Wagner & Linehan,
PARTICIPANTS AND PROCEDURE
One-hundred seventy-six undergraduate students (mean age = 20.23,SD
= 2.37, range = 18-38) at a large university in the southern United States
participated in exchange for extra credit in their psychology courses. The
majority of participants were female (84%) and endorsed White/Cauca-
sian (82%) as their ethnic background. Nine percent endorsed Black/Af-
rican American, 7% Asian, and 2% Hispanic/Latino. Most participants
endorsed being single (96%) in terms of marital status and, in terms of
mental health history, most endorsed not having been diagnosed with a
psychiatric disorder (91%). Of the 9% who did report a previous diagno-
sis, the most common were some form of depression, anxiety, or an eating
Participants were tested in small groups of approximately 10 to 20 stu-
dents per session. After signing an informed consent form that had been ap-
proved by the university’s internal review board, participants completed a
demographic background questionnaire, several other questionnaires (see
below), and they viewed a series of 10 slides with emotionally neutral faces
and responded to 18 items about each face.
MATERIALS AND MEASURES
Face Rating Task. The picture set that was used in this study was drawn
from the Psychological Image Collection at the Department of Psychology,
University of Stirling, United Kingdom. This database contains a large col-
lection of facial photographs that is freely available for research purposes.
Images may be downloaded at no charge via the Internet from
http://pics.psych.stir.ac.uk. Images selected for this study were drawn
from the “Stirling_Faces” database and included frontal grayscale photo-
graphs of five adult men and five adult women. All faces were shown in a
“neutral” emotional condition, which contrasts with available “smiling” and
“speech” conditions that were not used in this study. All 10 photographs
suggested thatthistimeperiod allowed participantsto finishtheirappraisal
ratings without feeling rushed or having excessive time to “second-guess”
Instructions for the appraisal task were displayed on an initial slide,
which appeared for 90 seconds. Instructions were as follows:
“The following pictures show faces of different men and women. In this study, we
are interested in finding out how looks or facial features influence people’s opin-
ions and reactions. The differences in facial expressions are sometimes subtle,
324 MEYER ET AL.
but they should give you clues about each person’s character or personality. Can
you interpret what the facial expressions say about each person? Please use the
sheets we provided to make your ratings. There is a total of 10 pictures. Each face
intuition and first impressions. Do not dwell on any one question.”
After these instructions, different faces were displayed for 90 seconds each,
with face gender alternated between slides.
Appraisal Ratings. Participants were asked to rate each face on a number
of appraisal dimensions. Consistent with Veen and Arntz (2000), a series of
bipolar rating scales with opposing character qualities was used. A total of
18 bipolar appraisal items was used, and each dimension was rated on a
scale from 1 to 6. Thus, the omission of a midpoint forced participants to fa-
polar dimension. Rating dimensions included: (1) Dislike/Like, (2)
Attractive/Unattractive, (3) Unfriendly/Friendly, (4) Good-na-
tured/Mean-spirited, (5) Could be a friend/Couldn’t be a friend, (6)
Cold/Warm, (7) Inviting/Rejecting, (8) Sad/Happy, (9) Good character/Bad
character, (10) Not trustworthy/Trustworthy, (11) Shy/Sociable, (12) Re-
laxed/Nervous, (13) Active/Passive, (14) Dull/Lively, (15) Intelligent/Sim-
ple-minded, (16) Wouldn’t like me/Might like me, (17) Anxious/Calm, and
(18) Interesting/Boring. The content of these items was based on the study
by Veen and Arntz (2000), although some items were devised specifically for
this study, based on pilot testing, or were selected from circumplex models
of mood (Watson & Tellegen, 1985).
After rating each face on these items, participants were asked to rate their
own current mood. For this purpose, six mood adjectives (three bipolar
mood dimensions) were selected from circumplex models (Watson &
Tellegen, 1985). Specifically, these included sad versus happy mood, re-
laxed versus nervous mood, and annoyed versus content mood. Each bipo-
lar dimension was rated on a six-point, bipolar scale.
Personality Disorder Features. The Structured Clinical Interview for
DSM-IV Axis II screening questionnaire (SCID-II-SQ; First, Gibbon, Spitzer,
Williams, & Benjamin, 1997) was used to measure features of borderline,
avoidant, and schizoid personality disorders.1The SCID-II-SQ items corre-
spond to interview questions that are intended to probe in more depth and
detail whether a particular feature is present. Because of feasibility con-
straints, however, only the questionnaire was administered here—a proce-
dure that has been used by others in similar studies (e.g., Dreessen et al.,
1999). We used a slightly elaborated response scale in this study: Instead of
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS325
1. We also measured features of other personality disorders but restricted our focus here on
the avoidant, borderline, and schizoid scales. This restriction was motivated primarily by
our theoretical interest in attachment-related distinctions among these three disorders.
This focus also served the goal of parsimony—with nine or more personality disorder scales
in all analyses, it would be difficult to sensibly explain and interpret any observed associa-
tions. A more exhaustive description of the relations among attachment styles and all the
personality scales we assessed in this sample is currently being prepared (Meyer & Pilkonis,
the original “Yes/No” response options, a 4-point scale was used (0 = Never
or not at all; 1 = Sometimes or a little; 2 = Often or moderately; 3 = Very often
reported reasonable stability of the SCID-II-SQ scales over 10 weeks. That
study also found that responses were not strongly influenced by state de-
pression and tended to concur with informant reports. In the present study,
internal consistency (Cronbach’s α) of the 7-item avoidant scale was .79; of
the 15-item borderline scale, .86; and of the 6-item schizoid scale, .55. Al-
though this last coefficient indicates potentially problematic internal
consistency, it is above the minimally acceptable threshold of .50 (Nunally,
Attachment Styles. The 36-item Experiences in Close Relationships
Questionnaire (ECR;Brennanetal.,1998) was used to measure anxious at-
tachment (sample item: “I often want to merge completely with romantic
partners, and this sometimes scares them away”) and avoidant attachment
(sample item: “I don’t feel comfortable opening up to romantic partners”).
Items were rated on a 7-point scale ranging from Disagree Strongly (1) to
Agree Strongly (7). Brennan et al. (1998) found among 1,086 college stu-
dents that the anxiety and avoidance scales were nearly uncorrelated with
each other (r = .11) but correlated in expected directions with 14 other mea-
sures of attachment-related constructs, demonstrating adequate conver-
gent and discriminant validity. Both the anxiety and avoidance scales also
demonstratedgood internalconsistencyinaprevious study(α=.91and.94,
respectively; Brennan et al., 1998). In the present study, α internal consis-
scale, .92. As expected, the scales did not correlate significantly with each
other (r = .10, p = .20).
Because of the large number of ratings (10 faces × 18 ratings per face × 176
participants = 31,680 data-points), an initial task was data reduction to a
more feasible number of variables. Because we were interested in partici-
riod or differences among specific faces), we averaged the ratings across all
picture(Dislike/Like) weresummed anddivided bythetotalnumberofpho-
tographs. The resulting 18 mean appraisal items were then entered into a
principal components analysis, for the purpose of further data reduction.
This analysis suggested a four-component solution. Sixteen of the 18
itemsthatwereenteredintotheanalysisshowed aclearloading patternand
were used for the computation of subscales. The four components ac-
counted for68.75% of thevariance,and theeigenvalues of thefirstfivecom-
ponents were 7.58, 1.88, 1.65, 1.28, and .75, indicating a sharp drop-off
after both the first and fourth eigenvalue.
326 MEYER ET AL.
Based on this analysis, the first scale was labeled “Attraction” and con-
warm, might like me, trustworthy, and like (only the positive poles of the bi-
polar items are mentioned here). Internal consistency of this scale was ex-
cellent (Cronbach’s α = .90). The other three scales were of secondary
conceptual importance and were much shorter. Thus, the “Dullness” scale
(α = .75) consisted of three items: passive, boring, and simple-minded; the
“Extraversion” scale (α = .76) also included three items: sociable, lively, and
happy; and the “Timidity” scale (α = .57) included only two items: nervous
CORRELATIONS AND DESCRIPTIVE STATISTICS
Descriptive statistics and correlations are presented in Table 1. As hypothe-
sized, anxious attachment correlated with avoidant and borderline person-
ality features, but not with schizoid personality features. Also consistent
with hypotheses, avoidant attachment correlated with avoidant personality
features (and weakly with schizoid features), but not with borderline fea-
tures. The avoidant and borderline personality scales were moderately cor-
related with each other, and both correlated significantly but weakly with
schizoid features. The face appraisal scales were moderately correlated
among each other, such that participants who liked the faces also tended to
correlated with tendencies to dislike the persons shown on the slides and to
appraise them as more timid. Avoidant personality disorder features were
also weakly linked with disliking of the persons shown and rating them as
more timid. The extraversion and dullness scales were only weakly or not at
all related to the personality disorder and attachment scales, with the ex-
ception of a weak but significant link between avoidant attachment and
tendencies to rate faces as dull.2
Path analyses were performed to examine whether, as hypothesized, per-
sonality disorder features were indirectly linked with negative face apprais-
als via the mediating path of insecure attachment. One advantage of path
analysis over hierarchical regression is that it provides estimates of overall
model fit in addition to estimates of effects among variables. Several indices
are typically used to determine quality of model fit. The most commonly
used are: χ2/df, Comparative Fit Index (CFI), and the Standardized Root
Mean Square Residual (SRMR). Model fit that includes a χ2/df < 2, CFI > .90,
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS 327
2. We also tested whether this pattern of correlations differed depending on whether male or fe-
male faces were shown. Despite minor gender discrepancies, the main pattern was not sub-
stantially different. For example, anxious attachment correlated positively with the timidity
scale and inversely with the attraction scale for both male and female faces.
TABLE 1. Correlations and Descriptive Statistics
Attachment Scales (ECR)
1. Avoidant Attachment
2. Anxious Attachment
Personality Disorder Scales (SCID II Screening Questionnaire)
3. Avoidant Personality
4. Schizoid Personality
5. Borderline Personality
Face Appraisal Scales
Items Per Scale
*p < .05, **p < .01.
and SRMR ≤ .07 is generally acceptable (Kline, 1998). These criteria were
used in the present study.3
We focused here only on the attractionface appraisal scale, given that this
was the variable of primary interest. More specifically, we posited that bor-
derline and avoidant features would have a direct effect on anxious attach-
ment. We also modeled direct effects from avoidant and schizoid features to
avoidant attachment, expecting that avoidant features would be more
strongly associated with avoidant attachment than schizoid features. Fi-
nally, we examined the direct effect of anxious and avoidant attachment on
This model achieved good fit, χ2(6, N = 176) = 7.12, p = .31, χ2/df = 1.18,
CFI = .98, SRMR = .03. As shown in Figure 1, our hypothesized model was
supported. Borderline and avoidant personality both related to anxious at-
tachment (z = 4.71, p < .05, and z = 2.80, p < .05, respectively), and anxious
attachment in turn predicted negative appraisals of the emotionally neutral
faces (z = -4.26, p < .05). Avoidant personality features were also associated
with avoidant attachment (z = 2.80, p < .05), but the link between avoidant
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS 329
3. We chose to test a path model because the complexity of our model could potentially lead to
parameter instability in a Structural Equation Model (SEM). For path analysis and SEM, Kline
(1998) recommends a subjects to number of model parameters ratio that is greater than 10:1.
Our path model exceeds this 10:1 criterion (176 subjects: 15 parameters = 11.73:1). However,
thesamemodelusinglatent indictorsofconstructs(i.e., SEM)wouldrequireatleast29param-
eters, which falls substantially short of the 10:1 criterion (i.e., 6.06:1). Therefore, we chose to
use path analyses because model results are likely to be more reliable than those derived from
FIGURE 1. Path Model: Personality Disorder (PD) Features, Attachment Styles, and
Face Appraisals (Attraction Scale).
attachment and negative face appraisals was not significant (z < 1, ns).
Schizoid features were not significantly related to avoidant attachment (z =
We tested a second model to determine if adding direct effects from each of
the three personality disorder features to negative face appraisals improved
model fit (i.e., partial mediation model). Doing so did not significantly im-
prove model fit, ∆χ2(1, N = 176) = 1.59, ns. Furthermore, none of the direct
paths from personality disorder features to face appraisals were significant
(zs < 1 and βs < |.08|). We thus retained the previous model. It appears that
borderline and avoidant personality features have an indirect influence on
face appraisals via attachment.
RELATIONS AMONG PERSONALITY FEATURES, ATTACHMENT STYLES,
AND AFFECT RATINGS
The correlations involving participants’ affect ratings (Table 2) showed that
borderline and avoidant personality features as well as anxious attachment
were associated with relatively worse affect during the face appraisal task.
tive face appraisals might be better explained by current negative affect. To
test this possibility, we computed three partial correlations between anx-
ious attachment and the main negative appraisal scale (attraction), control-
ling for each of the three negative affect items. In each case, the inverse
association between anxious attachment and positive appraisals remained,
even when controlling for concurrent affect (ps < .01 in each case). Even
when controlling for all three affect items simultaneously, anxious attach-
CORRELATIONS WITH APPRAISAL VARIABILITY AND EXTREMITY
Subsidiary analyses were conducted to test whether personality disorder
features, attachment styles, or affect correlated with the variability or ex-
tremity of face appraisals. This was based on previous research suggesting
that borderline patients tend to rate others in extreme terms (Veen & Arntz,
330MEYER ET AL.
TABLE 2. Correlations Among Attachment Styles, Personality Disorder Features, and
Personality Disorder FeaturesAttachment Styles
I feel happy
-.24**-.33** -.07-.21** -.11
I feel nervous
.29**.28** .05.30** .06
I feel content
-.25** -.21**-.11-.23** .03
*p < .05, **p < .01.
2000) and on the idea that instability in affect and cognition are core ele-
ments of the borderline syndrome (e.g., Koenigsberg et al., 2001, 2002).
Two sets of variables were computed for this purpose. First, we con-
structed five scales measuring the average (mean) standard deviation of
ing of 4 on all 8 attraction items, across all 10 faces. This participant’s at-
be zero. Participant Y, by contrast, might show a more variable response
pattern, with ratings on the attraction items fluctuating between 1 and 6
across the 10 faces. This participant’s attraction variability score would be
higher. The five variability indexes tended to be normally distributed (Skew-
ness and Kurtosis < .50).
The second set of variables measured the total number of extreme re-
sponses (i.e., responses of either 1 or 6 on the 6-point scale) per participant.
For example, one participant might never endorse strong liking or disliking
on any of the eight attraction items, across all faces. This participant’s at-
traction-extreme score would be zero. Another participant might endorse
extreme responses on all attraction items, across all 10 faces. This partici-
pant’s attraction-extreme score would be 80 (8 extreme responses per face ×
10 faces). The extremity variables exhibited positive skew (Skewness > 1.4,
Kurtosis > 1.8) because most participants tended to make no or very few
tion and affect variability, such that participants with more borderline fea-
tures tended to fluctuate in affect and in their appraisals of whether others
seemed likable or not. Participants with avoidant personality features (like
those with borderline features) were slightly more likely to experience affect
fluctuations than those who endorsed fewer of these PD features (see Table
3). Compared to those who scored lower on the schizoid scale, those with
more pronounced schizoid features tended to be more stable in their attrac-
tion and timidity face-appraisals. Participants who scored higher on anx-
ious attachment also tended to report greater variability in affect. None of
the personality or attachment scales correlated with the extremity of face
In a nonclinical college student sample, we found that borderline and
avoidant personality disorder features both related to anxious attachment,
and anxious attachment was in turn linked with tendencies to appraise
emotionally neutral faces more negatively. Anxiously attached individuals
tended to interpret these ambiguous interpersonal cues with considerable
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS331
4. Because of this skew, we also computed normalized extremity variables by applying
square-root transformations. However, the pattern of correlations changed only minimally,
with no changes at all in the significance of any correlation coefficients. Therefore, we report
here the correlations of the untransformed variables.
suspicion; they rated them, for example, as unlikely to be a friend, as rela-
tively unfriendly, untrustworthy, and potentially rejecting. These links
could not be explained by concurrent greater negative affect among those
with anxious attachment.
These findings fit with a series of previous studies suggesting that border-
line and avoidant personality disorders as well as anxious-preoccupied at-
tachment styles predispose people to process social information in biased,
negativeways(e.g.,Arntz& Veen,2001; Dreessen etal.,1999; Niedenthal et
al., 2002; Wagner & Linehan, 1999). Similar to the study by Dreessen and
colleagues (1999), who found that avoidant beliefs but not avoidant symp-
toms were linked with biased interpretations, we found that attachment
styles, which were linked with personality disorder features, were better
predictors of information-processing biases than the personality disorder
We found that borderline features were associated with the variability of
face appraisals, even though they were not linked with the valence (i.e.,
negativity) of appraisals. Those with more pronounced features of border-
line personality tended to fluctuate more in their level of attraction to the
faces, and their affect also tended to be more varied or unstable, compared
view that cognitive and affective instability are central to the borderline syn-
drome (Koenigsberg et al., 2001, 2002). Borderline features were not re-
lated, however, with tendencies to rate faces in extreme or dichotomous
treme response tendencies only emerged when borderline-related content
(emotional film clips) was presented.
332 MEYER ET AL.
TABLE 3. Appraisal Variability and Extremity: Correlations with Personality Disorder
Features, Attachment Styles, and Mean Affect Levels
Personality Disorder FeaturesAttachment Styles
Attachment Borderline AvoidantSchizoid
Appraisal Variability Scales (mean standard deviations per subject across items and faces)
.18* -.05 .18*.09
Appraisal Extremity Scales (endorsing responses of 1 or 6 on a 6-point scale)
Affect Extremity -.13
*p < .05, **p < .01.
A unique feature of thisstudy was the inclusion of a new measure of social
cognition: The 18-item rating scale of emotionally neutral faces. Many stud-
ies have linked different forms of psychopathology with biases in the ap-
praisal of faces (e.g., Davis & Gibson, 2000; Mogg, Millar, & Bradley, 2000;
Pollack & Tolley-Schell, 2003), but such investigations are rare in studies
examining personality disorders (Wagner& Linehan,1999). The method ex-
plored here seemed promising because participants engaged easily with the
task, internally consistent subscales could be derived, and theoretical pre-
dictions were at least partly supported. Along with other methods to assess
social information processing biases (e.g., Arntz & Veen, 2001; Veen &
processes creating and maintaining personality pathology.
Apart from the links between personality disorders and information pro-
cessing, our findings also support the view that borderline, avoidant, and
inant attachment pattern. Whereas borderline features related only to anx-
ious attachment, avoidant personality features were associated with both
anxious and avoidant attachment, and schizoid personality features were
weakly linked only with avoidant attachment. Because the combination of
high attachment anxiety and avoidant strategies can be interpreted as fear-
ful attachment, and the combination of high attachment anxiety and ap-
proach tendencies as preoccupied attachment (e.g., Brennan, Clark, &
Shaver, 1998), our findings support the conceptualization of avoidant per-
sonality as a variant of fearful attachment, and borderline personality as a
variant of preoccupied attachment (cf. Bartholomew et al., 2001; Lyddon &
Sherry, 2001; Meyer & Pilkonis, in press). Whether schizoid personality is
better viewed as a pattern of dismissing attachment or nonattachment (cf.
Bartholomew et al., 2001) cannot be resolved as we did not measure the
strength of attachment motivation itself.
The goal of this study was to provide a small node in the increasingly elab-
orate nomological net linking attachment theoretical constructs and per-
sonality disorders (cf. Bartholomew et al., 2001; Fonagy et al., 2002).
Bowlby’s theory has inspired an enormous amount of research with chil-
encouraging that this work is fertilizing efforts to address some of the short-
comings in current Axis II classification (cf. Clark, Livesley, & Morey, 1997).
At the same time, it is important to remember that attachment theory can
inform, but is not in itself, a theory of personality disorder. Bartholomew et
al. (2001) noted in this regard that, “attachment processes, in the past and
present, may be one important factor affecting developmental pathways to
personality disorder” (p. 211) but at the same time, “attachment problems
will be more relevant to some disorders than to others or to some aspects of
symptomatology than to others” (p. 212). The present study confirms that
even in a nonclinical sample of subthreshold personality features and mild
attachment problems, several theoretically expected links among insecure
attachment, borderline, avoidant, and schizoid personality disorder can be
documented. The magnitude of these relationships appears to be small to
moderate (cf. Hemphill, 2003).
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS333
Beyond documenting these empirical links among PD features, attach-
mentstyles,andbiased social cognition,however,itwillbeimportanttofur-
ther refine theoretical models that explain how these constructs create and
maintain personality dysfunction. We regard personality disorders as com-
plex phenomena that cannot be explained easily from a single theoretical
perspective (e.g., cognitive, behavioral, biological, social-interpersonal). In
the case of avoidant personality disorder, for example, it is reasonable to as-
sume that a set of distal (biological) and proximal (social and cognitive) fac-
tors converge to create the characteristic pattern of social avoidance. People
with a sensitive temperament might experience interpersonal rejection (or
other forms of early maltreatment) more intensely than others, which then
might lead to the formation of perceptual and evaluative biases that further
rejection lies ahead, which in turn might facilitate the formation of avoidant
motivational tendencies (Meyer & Carver, 2000). In avoidant personality
disorder, persistent tendencies to experience or interpret neutral social in-
formation as negative, to disapprove of others, or to attribute malicious in-
tent might serve to justify, confirm, and strengthen an intention to avoid
close social encounters, despite a desire for intimacy. Much more work will
be needed, however, to articulate and to document empirically the nature
and functional significance of such social-cognitive biases across the
various personality disorders.
Several limitations of this study ought to be noted, including our reliance
on a convenience sample of undergraduates and the use of a screening
questionnaire to assess personality disorder features. Additional limita-
tions are (1) that participants were asked to rate only faces but not other
(e.g., nonsocial) stimuli, which limits our ability to estimate the extent to
which the observed evaluative biases are specific to social stimuli; and (2)
that the order of face presentation and rating items was fixed rather than
counterbalanced or randomized, which might have introduced systematic
error. Nevertheless, this study adds to the literature in that it is perhaps the
in social information processing simultaneously. A larger study in which a
similar face-appraisal method is used to examine appraisal biases among
psychiatric outpatients is currently underway. Based on the encouraging
preliminary results observed in the present study, it seems reasonable to
further pursue how attachment theory and social cognitive approaches can
334 MEYER ET AL.
Ainsworth, M. D., S., Blehar, M. C., Waters,
E., & Wall, S. (1978). Pattern of attach-
ment: Psychological study of the
Strange Situation. Hillsdale, NJ:
Arntz, A., & Veen, G. (2001). Evaluations of
others by borderline patients. Journal
of Nervous and Mental Disease, 189,
Bartholomew, K., Kwong, M. J., & Hart, S. D.
(2001). Attachment. In J. W. Livesley
(Ed.), Handbook of personality disor-
ders: Theory, research, and treatment.
(pp. 196-230). New York: Guilford.
Attachment. New York: Basic Books.
Separation. New York: Basic Books.
PERSONALITY, ATTACHMENT, AND FACE APPRAISALS 335
Bowlby, J. (1980). Attachment and loss: Vol.3.
Loss. New York: Basic Books.
Brennan, K. A., & Shaver, P. R. (1998). At-
tachment styles and personality disor-
ders: Their connections to each other
and perceptions of parental caregiving.
Journal of Personality, 66, 835-878.
Brennan, K. A., Clark, C., & Shaver, P. R.
(1998). Self-report measurement of
adult attachment: An integrative over-
view. In J. A. simpson & W. S. Rholes
(Eds.), Attachment theory and close re-
lationships (pp. 46-76). New York:
Bretherton, I., & Munholland, K. A. (1999).
Internal working models in attachment
relationships: A construct revisited: A .
In J. Cassidy & P. R. Shaver (Eds.),
Handbook of attachment: Theory, re-
search, and clinical applications (pp.
89-111). New York: Guilford.
Butler, A. C., Brown, G. K., Beck, A. T., &
Grisham, J. R. (2002). Assessment of
dysfunctional beliefs in borderline per-
sonality disorder. Behaviour Research
and Therapy, 40, 1231-1240.
Cassidy, J. (1999). The nature of the child’s
ties. In J. Cassidy & P. R. Shaver (Eds.),
Handbook of attachment: Theory, re-
search, and clinical applications (pp.
3-20). New York: Guilford.
Cassidy, J., & Shaver, P. R. (1999). Handbook
of attachment: Theory, research, and
clinical applications. New York:
Clark, L. A., Livesley, W. J., & Morey, L.
(1997). Personality disorder assess-
ment: The challenge of construct valid-
ity. Journal of Personality Disorders,
Davis, P. J., & Gibson, M. G. (2000). Recogni-
tion of posedand genuine facial expres-
sions of emotion in paranoid and
nonparanoid schizophrenia. Journal of
Abnormal Psychology, 109, 445-450.
Dreessen, L., Arntz, A., Hendriks, T., Keune,
personality disorder and implicit
schema-congruent information pro-
cessing bias: A pilot study with a prag-
matic inference task. Behaviour
Research and Therapy, 37, 619-632.
Feeney, J. A. (1999). Adult romantic attach-
ment and couple relationships. In J.
Cassidy & P. R. Shaver (Eds.), Hand-
book of attachment: Theory, research,
and clinical applications (pp. 355-377).
New York: Guilford.
First, M. B., Gibbon, M., Spitzer, R. L., Wil-
liams, J. B. W., & Benjamin, L. S.
(1997). User’s guide for the Structured
Clinical Interview for DSM-IV Axis II Per-
sonality Disorders (SCID-II). Washing-
ton, DC: American Psychiatric Press.
mentalization, and the development of
self. New York: Other Press.
Fonagy, P., Leigh, T., Steele, M., Steele, H.,
Kennedy, R., Mattoon, G., Target, M., &
Gerber, A. (1996). The relation of at-
tachment status, psychiatric classifi-
Journal of Consulting and Clinical Psy-
chology, 64, 22-31.
Fraley, R.C., Waller, N.G., & Brennan, K.A.
(2000). An item response theory analy-
sis of self-report measures of adult at-
tachment. Journal of Personality and
Social Psychology, 78, 350-365.
as an organizational framework for re-
search on close relationships. Psycho-
logical Inquiry, 5, 1-22.
Hemphill, J. F. (2003). Interpreting the mag-
nitudes of correlation coefficients.
American Psychologist, 58, 78-79.
Horppu, R., & Ikonen-Varila, M. (2001). Are
attachment styles general interper-
sonal orientations? Applicants’ percep-
tions and emotions in interaction with
evaluators in a college entrance exami-
nation. Journal of Social and Personal
Relationships, 18, 131-148.
Kline, R. B. (1998). Principles and practice of
Koenigsberg, H. W., Harvey, P. D.,
Mitropoulou, V., New, A. S., Goodman,
M., Silverman, J. M., Silverman, J.,
Serby, M., Schopick, F., & Siever, L. J.
(2001). Are the interpersonal and iden-
tity disturbances in the borderline per-
sonality disorder criteria linked to the
traits of affective instability and
impulsivity? Journal of Personality Dis-
orders, 15, 358-370.
Koenigsberg, H. W., Harvey, P. D.,
Mitropoulou, V., Schmeidler, J., New,
A. S., Goodman, M., Silverman, J. M.,
Serby, M., Schopick, F., & Siever, L. J.
(2003). Characterizing affective insta-
bility in borderline personality disor-
336 MEYER ET AL. Download full-text
der. American Journal of Psychiatry,
Lyddon, W. J., & Sherry, A. (2001). Develop-
mental personality styles: An attach-
ment theory conceptualisation of
personality disorders. Journal of Coun-
seling and Development, 79, 405-414.
Meyer, B., & Carver, C. S. (2000). Negative
simism: A study of avoidant personality
disorder features in college students.
Journal of Personality Disorders, 14,
Meyer, B., & Pilkonis, P. A. (in press). Implica-
tions of attachment theory for person-
ality disorders. In M. F. Lenzenweger &
J. F. Clarkin (Eds.), Major theories of
personality disorder (2nd ed.). New
York: Guilford Press.
Meyer, B., & Pilkonis, P. A., Proietti, J.,
Heape, C., & Egan, M. (2001). Attach-
ment styles, personality disorders, and
response to treatment. Journal of Per-
sonality Disorders, 15, 371-389.
Mikulincer, M., & Shaver, P. R. (2001). At-
tachment theory and intergroup bias:
Evidence that priming the secure base
schema attenuates negative reactions
to out-groups. Journal of Personality
and Social Psychology, 81, 97-115.
sonality: DSM-IV and beyond. New
Mogg, K., Millar, N., & Bradley, B. P. (2000). Bi-
cial expressions in generalized anxiety
disorder and depressive disorder. Jour-
nal of Abnormal Psychology, 109,
Niedenthal, P. M., Brauer, M., Robin, L., &
Innes-Ker, A. H. (2002). Adult attach-
ment and the perception of facial ex-
pression of emotion.
Journal of Personality and Social Psychology,
Nunally, J. C. (1978). Psychometric theory.
New York: McGraw-Hill.
Ouimette, P.C., &Klein, D.N.(1995).Test-re-
test stability, mood-state dependence,
and informant-subject concordance of
the SCID-Axis II Questionnaire in a
nonclinical sample. Journal of Person-
ality Disorders, 9, 105-111.
Patrick, M., Hobson, R. P., Castle, D., Howard,
R., & Maughn, B. (1994). Personality
disorder and the mental representation
of early social experience. Development
and Psychopathology, 6, 375-388.
Pollak, S. D., & Tolley-Schell, S. A. (2003). Se-
lective attention to facial emotion in
physically abused children. Journal of
Abnormal Psychology, 112, 323-338.
Sheldon, A. E., & West, M. (1990). Attach-
ment pathology and low social skills in
avoidant personality disorder: An ex-
ploratory study. Canadian Journal of
Psychiatry, 35, 596-599.
ment theory and close relationships.
New York: Guilford.
Stravynski, A., Elie, R., & Franche, R. L.
patients diagnosed with avoidant per-
sonality disorder: A test of the
overprotection hypothesis. Acta
Veen, G., & Arntz, A. (2000). Multidimen-
sional dichotomous thinking charac-
terizes borderline personality disorder.
Cognitive Therapy and Research, 24,
Wagner, A. W., & Linehan, M. M. (1999). Fa-
cial expression recognition ability
among women with borderline person-
ality disorder: Implications for emotion
regulation? Journal of Personality Dis-
orders, 13, 329-344.
Watson, D., & Tellegen, A. (1985). Toward a
consensual structure of mood. Psycho-
logical Bulletin, 98, 219-235.
West, M., Rose, M. S., & Sheldon-Keller, A.
(1994). Interpersonal disorder in schiz-
oid and avoidant personality disorders:
An attachment perspective. Canadian
Journal of Psychiatry, 40, 411-414.
D. M., Zlotnick, C., Dolan-Sewell, R.,
Skodol, A. E., Grilo, C. M., Gunderson,
J. G., Sanislow, C. A., Zanarini, M. C.,
Bender, D. S., Rettew, J. B., &
McGlashan, T. H. (2002). Traumatic ex-
posure and posttraumatic stress disor-
der in borderline, schizotypal, avoidant
and obsessive-compulsive personality
disorders: Findings from the Collabora-
tive Longitudinal Personality Disorders
Study. Journal of Nervous and Mental
Disease, 190, 510-518.
Zanarini, M. C., Williams, A. A., Lewis, R. E.,
Reich, R. B., Vera, S. C., Marino, M. F.,
Levin, A., Yong, L., & Frankenburg, F.
R. (1997). Reported pathological child-
hood experiences associated with the
development of borderline personality
disorders. AmericanJournalof Psychia-
try, 154, 1101-1106.