Psychiatry Research Communications 4 (2024) 100189

Contents lists available at ScienceDirect

Psychiatry Research Communications

journal homepage: www.sciencedirect.com/journal/Psychiatry-Research-Communications

Interpersonal trauma Dissociates borderline from other personality

disorders in social orienting
Corine van Heusden a , Barbara Montagne a,b, Jack van Honk b,c , David Terburg b,d,*
a
GGz Centraal, Amersfoort, the Netherlands
b
Department of Experimental Psychology, Utrecht University, the Netherlands
c
Institute of Infectious Diseases and Molecular Medicine, University of Cape Town, South Africa
d
Department of Psychiatry & Mental Health, University of Cape Town, South Africa

A R T I C L E I N F O A B S T R A C T

Handling Editor: Dr. Leonardo Fontenelle Patients with Borderline Personality Disorder (BPD), particularly with comorbid trauma-disorders, show an
attentional bias towards angry facial expressions. This is often interpreted to reflect increased anxiety and
Keywords: sensitivity to social threats. Given BPDs severe problems in reacting to and interpreting social communication,
Emotion regulation we investigated whether this threat bias extends to social orienting. Using a gaze-cueing task, we assessed
Borderline personality disorder
whether centrally presented dynamic fearful and happy gaze stimuli promote the detection of peripherally
Social threat
presented targets. Groups with BPD (N = 50) and other personality disorders (OPD, N = 51) were compared to
Gaze-cueing
Trauma healthy controls (HC, N = 46), and evaluated on the independent influence of traumatic experience, trait anxiety
Anxiety and trait anger. Across groups we find reliable gaze-cueing. In line with earlier evidence, trait anxiety predicts
Anger faster detection of targets signaled by a fearful gaze in HCs. This threat bias is however not present in BPDs and
OPDs, thus the threat bias in BPD does not extend to social orienting. Instead, self-experienced trauma predicts
amplified gaze-cueing in BPDs, but reduced gaze-cueing in OPDs. This not only emphasizes the importance of
evaluating trauma exposure in personality disorders, but also suggests that the childhood adversity typically
associated with the development of BPD promotes increased social orienting.

1. Introduction differences in design, presentation time, emotions displayed and

whether participants fully meet the BPD diagnosis. Recently, Kaiser
Borderline Personality Disorder (BPD) is a serious mental illness, et al. (2018) followed-up on their meta-analysis and found an atten­
characterized by a pattern of intense and unstable emotions, interper­ tional bias in BPD towards angry faces, compared to non-patient and
sonal difficulties, problems with self-image and impaired impulse con­ patient (Cluster-C personality disorders) control groups. Interestingly,
trol (American Psychiatric Association, 2013). Emotional dysregulation this effect was most pronounced for the group of BPD with comorbid
is proposed to be a central feature (Carpenter and Trull, 2012; Glenn and PTSD, which is in line with a meta-analysis from Lazarov et al. (2019)
Klonsky, 2009), causing substantial suffering. In recent years much showing evidence for sustained attention on threatening stimuli in
research has been done to investigate underlying mechanisms of dys­ PTSD. Kaiser et al. (2018) therefore pose the question whether the
regulation in BPD. Several maladaptive cognitive processes were found, observed threat-related attention bias in BPD could specifically reflect
such as a negative interpretation bias of neutral and ambiguous facial PTSD rather than a maladaptive emotion-related process in BPD pa­
expressions (Baer et al., 2012; Bertsch et al., 2017; Daros et al., 2013; tients. There is, however, also evidence for the reverse, as Klanecky Earl
Domes et al., 2009; Izurieta Hidalgo et al., 2016; Mitchell et al., 2014) and colleagues (2020) found that the relationship between attentional
and an attentional bias towards negative emotional stimuli such as bias and PTSD was mediated by emotion regulation difficulties.
words or faces (Baer et al., 2012; Kaiser et al., 2017, 2018). Furthermore, Witthöft et al. (2015) found that patients with both PTSD
Research on attentional biases in BPD shows a mixed outcome (Jovev and BPD showed the strongest attentional bias for trauma-related
et al., 2012; Kaiser et al., 2018; Von Ceumern-Lindenstjerna et al., stimuli on an Emotional Stroop Task. Taken together, it thus seems to
2009), which Kaiser et al. (2017), in their meta-analysis, attribute to be the case that trauma experience and emotion regulation difficulties,

* Corresponding author. Heidelberglaan 1, 3584CS, Utrecht, the Netherlands.

E-mail address: [email protected] (D. Terburg).

https://doi.org/10.1016/j.psycom.2024.100189
Received 3 May 2024; Received in revised form 2 August 2024; Accepted 28 August 2024
Available online 29 August 2024
2772-5987/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

both characterizations of BPD (Porter et al., 2020), together promote an 39 female, 12 male; mean age 29.96 years, SD = 9.97) were recruited
attentional bias to social threat. from several departments for Personality Disorders of a large mental
The present study examines whether this increased sensitivity for health institution (GGZ Centraal, the Netherlands). The third group
social threat in BPD and in relation to trauma extends towards social consisted of 46 non-patient (healthy) controls (HC; 36 female, 10 male;
orienting in a gaze-cueing paradigm. Gaze-cueing is the automatic mean age 26.35 years, SD = 9.66). They were recruited either on campus
inclination to attend towards the direction that is signaled by the gaze (Utrecht University), via a database consisting of interns of GGZ Cen­
direction of someone else. Such spontaneous orienting of attention to­ traal, or via social media, and received a small fee for their participation.
wards the direction signaled by someone else’s gaze is a basic form of Participants were excluded if they had an intellectual disability, were in
social attention (McCrackin and Itier, 2019), i.e. attention driven by or a psychotic state, pregnant, colorblind, or if they could not see well in
towards cues associated with other individuals, which is believed to play general. All participants therefore had normal visual acuity, which
an important role in complex social tasks (Armstrong and Olatunji, included wearing contact lenses and/or glasses during the experiment.
2012). Indeed, people reflexively imitate the gaze of others, and this The three groups did not differ significantly in age (F(2, 144) = 1.75,
unconsciously gets priority over other tasks (Ricciardelli et al., 2002). p = .18), but they did differ on years of education (F(2, 144) = 6.14, p =
Both positive and negative emotional expressions enhance gaze-cueing, .003). Education in the HC group (M = 14.67, SD = 1.35) was signifi­
but the latter to a greater extent (Bannerman et al., 2009; McCrackin and cantly higher than in both the BPD (M = 13.24, SD = 2.52) and OPD
Itier, 2019; Putman et al., 2006; Terburg et al., 2012). Recent research group (M = 13.67, SD = 2.06), but the BPD and OPD group did not differ
suggests that BPD can increase sensitivity for averted gaze as recognition significantly. No significant difference was found between the three
of neutral faces with averted gaze in a 2-back working-memory task is groups for gender (X2(2, n = 147) = 4.94, p = .085, Cramer’s V = 1.83).
faster in individuals with BPD (Berchio et al., 2017). Gaze-cueing has to With regard to current medication and drug use (see Table 1),
our knowledge however not yet been studied in BPD. The current study significantly more BPD patients used medication (p < .001) and nar­
therefore investigates whether patients with BPD, particularly those cotics (p = .04) on the day of the test and the day before than partici­
with high trauma experience, show increased gaze-cueing towards pants from the OPD and HC group. The groups did not differ in
threat compared to a group with other personality problems and a percentage participants who used alcohol (p = .31).
healthy control group. Participants from both patient groups were diagnosed by trained
To test these hypotheses, we use a gaze-cueing-task (Terburg et al., psychologists with the Dutch version of the Structured Clinical Interview
2012) which closely resembles a situation wherein someone actively for DSM-IV axis II personality disorders (SCID-II; Weertman et al.,
shifts gaze towards a rewarding or threatening location. Participants 2000). Next to not meeting the required five borderline characteristics,
watch video-clips of faces shifting gaze in a happy or fearful manner but the OPD group also showed significantly less BPD characteristics (p <
must respond by gazing as quickly as possible to a target (asterisk) .001) (see Table 2). There were no significant differences in number of
appearing in the gaze-signaled (valid), or opposite location (invalid). criteria of other personality disorders, apart from more paranoid char­
The latency difference between shifting gaze towards a valid and invalid acteristics in the BPD group (p = .01). The number, however, is very low
trial is the gaze-cueing effect, and we expect gaze-cueing to be stronger and therefore not clinically relevant.
in the fear condition, especially in BPD patients, and further amplified Severity of borderline symptoms was assessed with the Borderline
by trauma experience. Additionally, we evaluate the influence of trait Personality Disorder Checklist (BPDC-47; Bloo et al., 2017). The BPD
anxiety and anger on gaze-cueing as the former has been shown to group showed significantly higher scores on all the BPDC scales (see
strengthen gaze-cueing effects (Putman et al., 2006), while the latter Table 2).
represents one of BPD’s diagnostic criteria (American Psychiatric As­ In the HC group, presence of a personality disorder was checked by
sociation, 2013) associated with affect instability (Lubke et al., 2015). administering a questionnaire for the screening of DSM-IV personality
More insight into these processes and their influence on emotion disorders (The Assessment of DSM-IV Personality Disorders
regulation is important because excessive vigilance to gaze-signaled
threat could result in an increase of symptoms through prolonged
threat exposure and less attention to signs of safety (Armstrong and Table 1
Olatunji, 2012; Jovev et al., 2012; Kaiser et al., 2018). Furthermore, this Use of medication and drugs.
study can provide valuable information for enhancing treatments by BPD OPD HC Analysis p-value
specifically targeting (social) attentional processes (Armstrong and
Medication (n) 76% (38) 53% (26) 22% (10) Fisher’s
Olatunji, 2012; Fodor et al., 2020; Penton-Voak et al., 2013; Von
<0.001
Exact test
Ceumern-Lindenstjerna et al., 2009), and extending these towards social Antidepressants (21) (16) (1)
orienting. (n)
Mood stabilizers (11) (9) (0)
(n)
2. Methods
Stimulants (n) (4) (1) (0)
Antipsychotics (n) (4) (0) (0)
2.1. Ethics statement Benzodiazepines (7) (3) (1)
(n)
The research reported in this article does not utilize any invasive Other anxiolytics (0) (1) (0)
(n)
techniques, substance administration or psychological manipulation.
Opiods (n) (0) (1) (0)
Therefore, compliant with Dutch law, this study only required, and Narcotics (n) 10% (5) 2% (1) 0% (0) Fisher’s 0.04
received approval form our internal science board at GGZ Centraal. Exact test
Furthermore, this research was conducted according to the principles Alcohol (n) 10% (5) 10% (5) 20% (9) Fisher’s 0.31
Exact test
expressed in the Declaration of Helsinki and written informed consent of
Number alcohol M = 0.13 M = 0.22 M = 0.39 One way 0.23
each participant was obtained. The datasets analyzed in the current (SD = (SD = (SD = ANOVA
study are available on request. 0.41) 0.83) 0.93) F(2, 146)
= 1.48
2.2. Participants and procedure Note: Medication = % participants using medication on the day of the test and
the day before. Narcotics = % participants using narcotics on the day of the test
A total of 50 patients with BPD (46 female, 4 male; mean age 28.00 and the day before. Alcohol = % participants using alcohol on the day of the test
years, SD = 8.92) and 51 patients with other personality disorders (OPD; and the day before. Number alcohol = mean number of glasses of alcohol.

2
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

Table 2
Personality diagnoses and severity of BPD symptoms in the BPD and OPD group.
BPD (n = 50) OPD (n = 51) Statistics dfᵃ p-value

Mean SD Mean SD T

SCID-II
Paranoid 1.18 1.24 0.55 1.15 − 2.66 98 0.01
Schizoid 0.29 0.54 0.29 0.70 0.02 97 0.98
Schizotypal 0.75 1.10 0.55 0.86 − 1.01 97 0.31
Antisocial 0.78 1.38 0.53 1.05 − 1.02 95 0.31
Borderline 6.24 1.25 2.14 1.47 − 15.07 99 0.00
Histrionic 0.27 0.74 0.14 0.40 − 1.11 71.65 0.27
Narcissistic 0.20 0.74 0.41 1.25 1.00 98 0.32
Avoident 2.60 1.99 2.10 1.96 − 1.29 97 0.20
Dependent 1.22 1.45 0.82 1.75 − 1.25 97 0.21
Obsessive-compulsive 1.57 1.66 1.30 1.26 − 0.88 98 0.38
BPDC
Abandonment 17.40 5.68 11.51 4.51 − 5.29 84 <0.001
Relationships 8.82 3.43 5.59 2.33 − 5.15 77.92 <0.001
Self image 30.44 8.76 23.54 6.99 − 4.01 84 <0.001
Mood swings 12.04 2.83 9.07 2.95 − 4.77 84 <0.001
Emptiness 4.22 0.97 3.02 1.13 − 5.28 84 <0.001
Impulsivity 14.36 4.09 10.78 3.06 − 4.55 84 <0.001
Parasuide 7.53 3.25 5.41 2.21 − 3.56 78.03 0.001
Dissociation and paranoia 19.22 6.61 14.46 4.81 − 3.84 80.22 <0.001
Anger outbursts 8.71 3.60 5.93 1.57 − 4.72 61.40 <0.001
Total 122.76 28.47 89.02 19.88 − 6.44 78.88 <0.001

Note: SCID-II = Dutch version of the structured clinical interview for DSM-IV axis II personality disorders. BPDC = Borderline Personality Disorder Checklist.
a
Degrees of freedom may vary because of missing data.

questionnaire (ADP-IV; Schotte and De Doncker, 2000)). Three partici­ In conclusion, this study examined three clearly distinguishable
pants who met the criteria for a personality disorder were excluded. The groups: The BPD group consisted of people with severe borderline per­
rest of the HC group did not meet the criteria of any personality disorder. sonality disorders. The OPD group consisted of people with personality
To investigate whether the groups differed in their history of trauma, disorders clearly distinguishable from the BPD group. The HC group did
participants were asked to complete a checklist for traumatic experi­ not meet criteria for a personality disorder. Furthermore, the three
ences (part of the Clinical Interview for PTSD, KIP), examining ‘self- groups differed on psychological symptoms in the previous week and
experienced trauma’, ‘witnessed trauma’ and ‘traumatic experiences of state anxiety and anger. The patient groups did not differ in number of
someone close’ (Hovens et al., 2005). They were also asked to complete traumatic (childhood) experiences.
the Dutch version of the STAI/STAS assessing trait proneness to anxiety
and anger (Lubke et al., 2015; Spielberger and Sydeman, 1994), and the
Symptom Checklist-90-R (SCL-90), assessing a broad spectrum of psy­ 2.3. Apparatus and analyses
chopathology in the previous week (Arrindell and Ettema, 2003; Dero­
gatis and Unger, 2010) (see Table 3). Participants completed an eye-tracked gaze-cueing task using real-
The two patient groups did not differ significantly on ‘total number time eye-tracking to record responses (see Fig. 1). Stimuli and design
of trauma’ (p = .77), but the differences between BPD and HC (p = .02), of this task were adapted from Terburg et al. (2012) and consisted of
and OPD and HC (p = .002) were significant. On subscale level, there video-clips of centrally presented faces changing rapidly (120ms) from
was a significant difference between the groups on ‘self-experienced neutral to either happy or fearful, while the eyes simultaneously moved
trauma’ (p < .001). Tukey post hoc test showed a significant difference from central to peripheral gaze (left and right). The final frame was
between the two patient groups and the HC group (p < .001), but the two maintained for an additional 80 ms, after which the face disappeared
patient groups did not differ significantly. Furthermore, there were no and in 2/3 of the trials, a target appeared either to the left or right (10◦
significant differences between the three groups on the subscales ‘wit­ visual angle) of the face. For the video-clips 8 different actors (4 female),
nessed trauma’ and ‘traumatic experiences of someone close’. with 2 emotions (happy and fearful) and 2 gaze-directions (left and
There was a significant difference between the three groups on trait right), were used, making 32 unique stimuli. These were presented 3
anxiety (p < .001), with the BPD group having a significantly higher times each; once with a target at the same location as the gaze-shift
score than both OPD and HC (p < .001) and OPD than the HC group (p (valid trial), once with a target at the opposite location as the
< .001). There also was a significant difference between the three gaze-shift (invalid trial) and once with no target to avoid habitual
groups on trait anger (p < .001), again with the BPD group having a saccade preparation (catch trial). These made a total of 96 trials,
significantly higher score than both OPD and HC (p < .001) and OPD counterbalanced for emotion and condition, and presented in random
than HC (p = .001). order. Preceding the task, 9 trials were presented for practicing, but
Furthermore, the BPD group had a significantly higher total score on without shift of emotion. Participants are instructed to shift their gaze
the SCL-90 than the OPD (p = .001) and the HC group (p < .001), and towards the target and were explicitly and correctly informed that
the total score of OPD was significantly higher than that of HC group (p gaze-direction of the presented face did not predict target appearance or
< .001). On a subscale level, the BPD group had significantly higher location. Responses were made with a shift of gaze to the target, which
scores on ‘Fear’ (p = .03), ‘Depression’ (p = .003), ‘Distrust’ (p = .001), disappeared when the eye-tracking data indicated that the target was
‘Hostility’ (p < .001) and ‘Other (not specified)’ (p < .001) than the OPD reached. Stimulus presentation commenced when the participant gazed
group. This higher degree of distrust and hostility are to be expected, at a fixation cross, positioned where the eyes of the subsequently pre­
given that they are characteristics of the borderline classification. The sented faces would appear, for a random time between 1000ms and
increased anxiety and depression could possibly be an expression of 1500ms to avoid timing habituation. During the catch trials, wherein no
these mood swings or could maybe signal comorbidity. target appeared, gaze had to be maintained at the fixation position until
start of the next trial.

3
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

Table 3
Clinical characteristics of the three study groups.
BPD (n OPD HC (n Statisticsa dfb p-
= 50) (n = = 46) value
51)

M±SD M±SD M±SD F

CTE
Self-experienced 4.1 ± 3.9 ± 1.7 ± 14.3 2, <.0011
trauma 2.3 3.0 1.6 131
Witnessed trauma 1.1 ± 1.5 ± 0.8 ± 2.7 2, 0.07
1.3 1.9 0.1 131
Trauma someone 2.6 ± 3.0 ± 2.8 ± 0.2 2, 0.81
close 2.0 2.6 2.1 131
Total trauma 7.7 ± 8.3 ± 5.3 ± 6.7 2, .0021
3.9 5.4 3.0 131

STAI 66.2 59.2 34.0 216.7 2, <.0012

± 8.3 ± 9.0 ± 7.0 143
STAS 25.3 18.4 14.5 51.4 2, <.0012
± 6.4 ± 5.8 ± 3.0 144
SCL-90
Anxiety 29.2 25.2 12.6 64.0 2, <.0012
± 8.7 ± 8.7 ± 3.4 138
Agorafobia 15.5 14.2 7.5 ± 28.4 2, <.0011
± 6.2 ± 6.7 1.9 138
Depression 53.8 45.7 21.6 96.6 2, <.0012
± 13.3 ± 13.4 ± 7.0 138
Fig. 1. Visual representation of the gaze-cueing task.
Somatic 30.0 27.3 16.1 40.2 2, <.0011
± 8.8 ± 9.0 ± 5.4 138
Insufficiency 27.7 24.7 13 ± 60.8 2, <.0011 Satterthwaite method for degrees of freedom and all continuous vari­
± 8.2 ± 8.0 2.9 138
2
ables were centred. All p-values are two-sided and if below 0.05 the
Distrust 49.2 40.5 23.2 65.3 2, <.001
± 14.5 ± 11.7 ± 5.2 138
results were considered statistically significant.
Hostility 13.9 9.3 ± 6.9 ± 48.0 2, <.0012
± 5.3 2.8 1.3 138 3. Results
Sleep 8.7 ± 8.8 ± 4.6 ± 27.9 2, <.0011
3.7 3.4 1.8 138
Other 23.4 18.9 11.2 59.8 2, <.0012
3.1. Gaze-cueing
± 7.3 ± 4.9 ± 3.4 138
Psychoneurotics 248.3 207.6 116.5 77.0 2, <.0012 We first tested the main task (VALIDITY and EMOTION) and GROUP
(Total) ± 62.3 ± 61.4 ± 23.2 142 effects and their interactions (SI-Model 1). We found the expected
Note: CTE = Checklist Trauma Experiences (part of the KIP). STAI = state VALIDITY effect (F(1,432) = 53.46, p < .001), which interacted
anxiety. STAS = state anger. SCL-90 = Symptom Checklist-90-R.12. marginally with GROUP (F(2,432) = 2.85, p = .059). Simple effects
1
BPD and OPD differed significantly from HC, p < .05. analysis showed reliable gaze-cueing effects throughout (all ps < 0.02,
2
All groups differed significantly from each other, p < .05. see Fig. 2) except for gaze-cueing of happy gaze-shifts in the HC group (p
a
BPD vs OPD vs HC. = .336). Unexpectedly, all EMOTION effects were non-significant (all ps
b
Degrees of freedom may vary because of missing data. > 0.20), which shows that the threat-bias reported in relation to BPD
does not extend to social orienting.
The gaze movements were recorded with a Tobii X2-60 eye tracker
(sampling at 60Hz with 0.4◦ precision, Tobii Technology, Danderyd,
3.2. Self-experienced trauma
Sweden). With this eye-tracking system head-fixation is not necessary.
Latency of the gaze-shifts was defined as the time between onset of, and
We next evaluated the influence of self-experienced trauma by
first gaze-point within 1◦ of the target. Trials with latencies with more
adding TRAUMA to SI-Model 1 as a continuous variable (SI-Model 2a).
than two standard deviations from the participants own mean score
TRAUMA interacted with EMOTION (F(2,384) = 4.28, p = .015), but
were rated as outliers and removed from analysis.
simple-slope analysis was inconclusive (Fear: t(134) = -0.43, p = .669,
All statistical analyses were performed using Jamovi (v1.8.1) (The
Happy: t(134) = 0.23, p = .821). Interestingly, we found a three-way
Jamovi Project, 2020) using the GAMLj toolbox (v2.4.8) (Gallucci,
interaction of GROUP by VALIDITY by TRAUMA interaction (F
2019). Mean reaction time latencies were computed for all 12 condi­
(2,384) = 5.28, p = .005). To estimate the simple-slopes of this inter­
tions; GROUP (BPD/OPD/HC), EMOTION (fear/happy) and VALIDITY
action we computed gaze-cueing scores by subtracting reaction times on
(valid/invalid) and log-transformed due to a right-skewed distribution.
congruent trials from incongruent trials and used these in a new LMM
These data were analyzed using linear mixed models (LMM). First, a
with GROUP and EMOTION as factors and TRAUMA as a continuous
model was constructed to test for the basic task effects using GROUP as
variable (see SI-Model 2b). Simple slopes indicated no significant slope
between factor, EMOTION and VALIDITY as within-factors and a
for HC (p = .665), but TRAUMA decreased gaze-cueing in OPD (t(128) =
random effect for the intercept across participants was included to ac­
-2.23, p = .028) and increased gaze-cueing in BPD (t(128) = 2.14, p =
count for general between-subjects variance (see SI-Model 1). Next, the
.035), see Fig. 3A. Fixed-effects estimates indicate that these slopes are
three continuous variables (TRAUMA, STAI and STAS) were added to
indeed significantly different (t(128) = -3.06, p = .003).
this model one by one (including all interaction terms) to create three
additional models (see SI-Models 2–4). Finally, a model was estimated
using all three continuous variables (including their interaction with the 3.3. Trait anxiety
within and between factors) to test if their effects are independent from
each other (see SI-Model 5). All LMMs were estimated using the We next evaluated the influence of trait anxiety by adding STAI to SI-
Model 1 as a continuous variable (SI-Model 3a) and find a four-way

4
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

Fig. 2. Reaction times (gaze-shift latencies) across all conditions and groups showing reliable gaze-cueing effects, but no emotion effects. Error-bars represent SEM
and *p < .05, **p < .01, ***p < .001.

Fig. 3. A) Relation (simple-effects slopes) between gaze-cueing and self-experienced trauma. Trauma predicts increased gaze-cueing in BPDs but decreased gaze-
cueing in OPDs. B) Relation (simple-effects slopes) between gaze-cueing and STAI. STAI predicts increased gaze-cueing of fearful gazes in HCs. Error-ranges
represent SEM and *p < .05, ***p < .001, n.s. = non-significant.

interaction of GROUP by VALIDITY by EMOTION by STAI (F(2,420) = significant (all ps > 0.31). See Fig. 3B.
3.72, p = .025). As before we used gaze-cueing scores to estimate the
simple slopes for this interaction in a new LMM with GROUP and 3.4. Trait anger
EMOTION as factors and STAI as a continuous variable (see SI-Model
3b). Simple slopes indicate that STAI only increases gaze-cueing of We next evaluated the influence of trait anger by adding STAS to SI-
fearful gaze in HC (t(278) = 3.47, p < .001). All other slopes are non- Model 1 as a continuous variable (SI-Model 4) and find a GROUP by

5
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

STAS interaction (F(2,141) = 3.11, p = .048). Simple slopes indicate that data on the content of the traumatic experiences from our participants,
STAS slows down performance in OPD (t(278) = 2.56, p = .011). Other but a recent structured review showed that BPDs have experienced more
slopes are insignificant (both ps > 0.26), and fixed effects estimates childhood abuse compared to patients with other personality disorders
indicate that the STAS-performance slope in OPD is significantly (Bozzatello et al., 2021). Arguably, these early experiences of interper­
different from the HC group (t(141) = -2.14, p = .034) and marginally sonal trauma promote the amplified sensitivity for social cues in general
different from the BPD (t(141) = -1.90, p = .059) group. that we observe in the present study. Indeed, we also observe higher
scores of abandonment and relationship issues in the BPD compared to
3.5. Independency test OPD group, suggesting heightened sensitivity for social interaction.
Future research should therefore address which BPD symptoms and
Lastly, we tested for independence of the TRAUMA, STAI and STAS what types of traumatic experience contribute to the increased sensi­
effects reported earlier by adding all three continuous predictors tivity for social interaction and social orienting, which might also shed
simultaneously to SI-Model 1 (SI-Model 5). All three earlier reported light on why OPDs with high traumatic experience show the opposite
interaction effects remain significant (GROUP by VALIDITY by effect.
TRAUMA interaction (F(2,366) = 5.07, p = .007); GROUP by VALIDITY Interesting in this respect is Witthöft and colleagues’ (2015) obser­
by EMOTION by STAI interaction (F(2,366) = 3.39, p = .035); GROUP vation that attention for trauma-related words in an emotional Stroop
by STAS interaction (F(2,122) = 5.24, p = .007)), thus indicating that design is particularly promoted in individuals with combined PTSD and
they are independent predictors of gaze-cueing behavior. BPD. In contrast to our observations, this study shows an attentional
effect selective for threat-stimuli, which again might be explained by the
4. Discussion difference in the used paradigms, i.e. Stroop designs measure how the
stimuli themselves attract attention while gaze-cueing measures how the
We tested whether the threat-bias typically observed in highly stimuli affect attention elsewhere. When taking this in consideration,
anxious individuals and in BPDs with PTSD extends to social orienting. our results seem to be in line with Witthöft et al. (2015) observation that
We particularly tested the attention-directing effect of a fearful it is the combination of BPD and trauma that most strongly affects
(compared to happy) face gazing towards the periphery, signaling a attentional processing.
threat in the nearby environment. We found reliable gaze-cueing Another issue in this respect is that exposure to trauma, compared to
(congruent targets were detected faster across emotion), which in HCs PTSD, can lead to different patterns of attentional bias. Lakshman et al.
was indeed mediated by trait anxiety such that highly anxious HCs show (2020) found that children with a history of trauma had a greater
a threat bias in gaze-cueing. This threat bias was however not present in attention bias towards angry faces compared to happy faces, which was
BPDs and OPDs, which refutes the hypothesis that the threat bias in associated with childhood trauma exposure, but not with PTSD. PTSD
these groups extends to social orienting. In the two groups with per­ symptoms in these children were significantly correlated with greater
sonality problems, however, gaze-cueing interacted with experienced attentional bias toward happy faces. In our study we did not distinguish
trauma such that gaze-cueing, irrespective of emotion, is amplified in between BPD with or without PTSD, which might explain why trauma
BPDs with high trauma experience while OPDs show the opposite effect. experience has effect on social orienting to threat as well as reward.
Furthermore, this effect was independent from trait anxiety and anger Further research should therefore investigate whether interpersonal
levels. Taken together we can confirm that the threat bias extends to trauma experiences compared to PTSD have different effects on
social orienting in highly anxious individuals, but this is not the case for gaze-cueing in BPD.
the typical threat bias observed in patients with personality problems. It should furthermore be noted that the level of education was
Instead, self-experienced trauma influences social orienting in BPD and significantly higher in the HC group compared to the two clinical
OPD in opposite ways, which emphasizes the importance of taking ac­ groups. Although we do not expect level of education to affect perfor­
count of trauma exposure when dealing with personality disorders. mance in the relatively simple detection task we used here, we cannot
Our results thus suggest that, although BPDs are biased in their rule out that this could be related to the observation that the HC group
attention to, and interpretation of, facial expressions of threat directed showed overall faster performance. Importantly, all groups showed
at themselves, this is not the case when the gaze of these expressions reliable within-subject gaze-cueing effects and all main results tap into
signal threat elsewhere. To our knowledge no other research has this within-subject comparison rendering potential differences in rela­
addressed such social orienting in BPD. There is however some evidence tion to education level less influential.
that BPDs are indeed sensitive to gaze cues as they recognize neutral Strengths of this study lie in the fact that this is, to our knowledge,
faces with averted gaze more quickly (Berchio et al., 2017) and attend the first study into gaze-cueing in relation to BPD, and the inclusion of
more quickly to the eyes of faces in general, during emotion recognition healthy as well as personality-disorder control groups. The use of dy­
(Seitz et al., 2021). This is in line with our observation that BPDs, namic gaze-cues is also a strength of this study, but also imposes the
particularly those with high trauma experience, are sensitive for social limitation that neutral gaze-cues could not be included in a well-
gaze-cues generalized across emotions (in our case happy and fearful). It controlled manner. Our study is also limited in the fact that we did
might thus be the case that BPD, particularly in combination with not assess the content of the experienced trauma. Lastly, we cannot rule
trauma, increases sensitivity for social signaling in general. This not only out effects of medication on performance in the gaze-cueing task,
provides an explanation why we do not find the hypothesized threat-bias because we allowed the participants to take their medication in the days
in gaze-cueing, but also suggests that in terms of attention any social cue leading up to the experiment in order not to disrupt their treatment.
might be of high relevance to traumatized individuals with BPD. Of note, However, this also ensures that we have a participant sample repre­
most previous research on threat-bias in BPD is focused on interpreta­ sentative of the BPD (and OPD) population commonly seen in patient
tion of, and attention towards, social cues, whereas gaze-cueing para­ facilities, which provides for increased clinical relevance.
digms measure how social cues influence attention elsewhere. Thus, In terms of clinical relevance, the observation that experienced
while attention towards social threat might be amplified in BPD, trauma affects sensitivity for social orienting in BPD and other person­
attention driven elsewhere due to social signaling might be amplified in ality disorders in opposite ways emphasizes the importance of taking
general. Future research using other emotions as well as emotionally account of experienced trauma when dealing with personality disorders.
neutral gaze-cues is needed to give more insight into this matter. Particularly the oftentimes severe interpersonal difficulties in BPD might
The intriguing difference in how interpersonal trauma experience be fed by how experienced trauma affects sensitivity for social signaling
predicts gaze-cueing in BPDs and OPDs could reflect that the type of in this group, which could provide handles for more focused treatment.
trauma is different for these patient groups. Unfortunately, we have no In other personality disorders experienced trauma might instead blunt

6
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

the sensitivity for social orienting, thereby reducing the effectiveness of movements. Proceedings. Biological Sciences 276 (1662), 1635–1641. https://doi.
org/10.1098/RSPB.2008.1744.
social support. However, given the diversity of this group, in general as
Berchio, C., Piguet, C., Gentsch, K., Küng, A.L., Rihs, T.A., Hasler, R., Aubry, J.M.,
well as within the context of our study, strong conclusions cannot be Dayer, A., Michel, C.M., Perroud, N., 2017. Face and gaze perception in borderline
drawn. Lastly, more knowledge about patients’ specific sensitivity to personality disorder: an electrical neuroimaging study. Psychiatr. Res. Neuroimaging
social cues, such as gaze direction, could not only provide greater insight 269, 62–72. https://doi.org/10.1016/J.PSCYCHRESNS.2017.08.011.
Bertsch, K., Krauch, M., Stopfer, K., Haeussler, K., Herpertz, S.C., Gamer, M., 2017.
and adaptation to these sensitivities but also, together with ostensive Interpersonal Threat Sensitivity in Borderline Personality Disorder: An Eye-Tracking
cueing, potentially improve communication and the building of Study 31 (5), 647–670. https://doi.org/10.1521/PEDI_2017_31_273, 10.1521/Pedi_
epistemic trust (Fonagy et al., 2019). 2017_31_273.
Bloo, J., Arntz, A., Schouten, E., 2017. The borderline personality disorder checklist:
psychometric evaluation and factorial structure in clinical and nonclinical samples.
4.1. Conclusion Annals of Psychology 20 (2), 281–336. https://doi.org/10.18290/
RPSYCH.2017.20.2-3EN.
Bozzatello, P., Rocca, P., Baldassarri, L., Bosia, M., Bellino, S., 2021. The role of trauma
We show that the commonly observed increased attention for (so­ in early onset borderline personality disorder: a biopsychosocial perspective. Front.
cial) threat in BPD does not extend to social orienting. This threat bias in Psychiatr. 12, 721361 https://doi.org/10.3389/FPSYT.2021.721361.
social orienting was however confirmed in relation to trait anxiety in Carpenter, R.W., Trull, T.J., 2012. Components of emotion dysregulation in borderline
personality disorder: a review. https://doi.org/10.1007/s11920-012-0335-2.
healthy individuals. Interpersonal traumatic experiences amplified so­ Daros, A.R., Zakzanis, K.K., Ruocco, A.C., 2013. Facial emotion recognition in borderline
cial orienting in general in the BPD group but decreased social orienting personality disorder. Psychol. Med. 43 (9), 1953–1963. https://doi.org/10.1017/
in the OPD group. Thus, borderline patients with high experience of S0033291712002607.
Derogatis, L.R., Unger, R., 2010. Symptom checklist-90-revised. The Corsini
interpersonal trauma seem to be highly sensitive to the overt attention of Encyclopedia of Psychology 1–2. https://doi.org/10.1002/9780470479216.
others irrespective of the expressed emotion. The opposite influence of CORPSY0970.
self-experienced trauma in BPD and OPD emphasizes the importance of Domes, G., Schulze, L., Herpertz, S.C., 2009. Emotion Recognition in Borderline
Personality Disorder—A Review of the Literature 23 (1), 6–19. https://doi.org/
taking trauma into account when dealing with personality disorders. 10.1521/PEDI.2009.23.1.6, 10.1521/Pedi.2009.23.1.6.
Finally, the childhood adversity typically associated with the develop­ Fodor, L.A., Georgescu, R., Cuijpers, P., Szamoskozi, Ş., David, D., Furukawa, T.A.,
ment of BPD might promote the here observed heightened social Cristea, I.A., 2020. Efficacy of cognitive bias modification interventions in anxiety
and depressive disorders: a systematic review and network meta-analysis. Lancet
orienting.
Psychiatr. 7 (6), 506–514. https://doi.org/10.1016/S2215-0366(20)30130-9.
Fonagy, P., Luyten, P., Allison, E., Campbell, C., 2019. Mentalizing, epistemic trust and
Funding the phenomenology of psychotherapy. Psychopathology 52 (2), 94–103. https://doi.
org/10.1159/000501526.
Gallucci, M., 2019. GAMLj: general analyses for linear models ([jamovi module]). https
This research received no specific grant from any funding agency, ://gamlj.github.io/.
commercial or non-profit. Glenn, C.R., Klonsky, E.D., 2009. Emotion dysregulation as a core feature of borderline
personality disorder. J. Pers. Disord. 23 (1), 20–28. https://doi.org/10.1521/
PEDI.2009.23.1.20.
CRediT authorship contribution statement Hovens, J., Luinge, B., van Minnen, A., 2005. Het Klinisch Interview Voor PTSS (KIP).
Cure & Care Publishers.
Corine van Heusden: Writing – original draft, Resources, Investi­ Izurieta Hidalgo, N.A., Oelkers-Ax, R., Nagy, K., Mancke, F., Bohus, M., Herpertz, S.C.,
Bertsch, K., 2016. Time course of facial emotion processing in women with
gation, Formal analysis. Barbara Montagne: Writing – review & edit­ borderline personality disorder: an ERP study. J. Psychiatr. Neurosci. 41 (1), 16–26.
ing, Resources, Methodology, Conceptualization. Jack van Honk: https://doi.org/10.1503/JPN.140215.
Writing – review & editing, Conceptualization. David Terburg: Writing Jovev, M., Green, M., Chanen, A., Cotton, S., Coltheart, M., Jackson, H., 2012.
Attentional processes and responding to affective faces in youth with borderline
– original draft, Visualization, Software, Methodology, Formal analysis, personality features. Psychiatr. Res. 199 (1), 44–50. https://doi.org/10.1016/j.
Conceptualization. psychres.2012.03.027.
Kaiser, D., Jacob, G.A., Domes, G., Arntz, A., 2017. Attentional bias for emotional stimuli
in borderline personality disorder: a meta-analysis. Psychopathology 49 (6),
Declaration of competing interest 383–396. https://doi.org/10.1159/000448624.
Kaiser, D., Jacob, G.A., Van Zutphen, L., Siep, N., Sprenger, A., Tuschen-Caffier, B.,
The authors declare that they have no competing interests. Senft, A., Arntz, A., Domes, G., 2018. Patients with borderline personality disorder
and comorbid PTSD show biased attention for threat in the facial dot-probe task. http
s://doi.org/10.1016/j.jbtep.2018.11.005.
Acknowledgements Klanecky Earl, A.K., Robinson, A.M., Mills, M.S., Khanna, M.M., Bar-Haim, Y., Badura-
Brack, A.S., 2020. Attention bias variability and posttraumatic stress symptoms: the
mediating role of emotion regulation difficulties. Cognit. Emot. 34 (6), 1300–1307.
We would like to thank Lisa van Hal, Annemiek de Kock, Annegeertje https://doi.org/10.1080/02699931.2020.1743235.
Phaff and Lena Verhoeff for their assistance with data-collection. Lakshman, M., Murphy, L., Mekawi, Y., Carter, S., Briscione, M., Bradley, B., Tone, E.B.,
Norrholm, S.D., Jovanovic, T., Powers, A., 2020. Attention bias towards threat in
African American children exposed to early life trauma. Behav. Brain Res. 383
Appendix A. Supplementary data https://doi.org/10.1016/J.BBR.2020.112513.
Lazarov, A., Suarez-Jimenez, B., Tamman, A., Falzon, L., Zhu, X., Edmondson, D.E.,
Supplementary data to this article can be found online at https://doi. Neria, Y., 2019. Attention to threat in posttraumatic stress disorder as indexed by
eye-tracking indices: a systematic review. Psychol. Med. 49 (5), 705–726. https://
org/10.1016/j.psycom.2024.100189. doi.org/10.1017/S0033291718002313.
Lubke, G.H., Ouwens, K.G., de Moor, M.H.M., Trull, T.J., Boomsma, D.I., 2015.
References Population heterogeneity of trait anger and differential associations of trait anger
facets with borderline personality features, neuroticism, depression, Attention
Deficit Hyperactivity Disorder (ADHD), and alcohol problems. Psychiatr. Res. 230
American Psychiatric Association, 2013. Diagnostic and statistical manual of mental
(2), 553–560. https://doi.org/10.1016/J.PSYCHRES.2015.10.003.
disorders. Diagnostic and Statistical Manual of Mental Disorders. https://doi.org/
McCrackin, S.D., Itier, R.J., 2019. Individual differences in the emotional modulation of
10.1176/APPI.BOOKS.9780890425596.
gaze-cuing. Cognit. Emot. 33 (4), 768–800. https://doi.org/10.1080/
Armstrong, T., Olatunji, B.O., 2012. Eye tracking of attention in the affective disorders: a
02699931.2018.1495618.
meta-analytic review and synthesis. https://doi.org/10.1016/j.cpr.2012.09.004.
Mitchell, A.E., Dickens, G.L., Picchioni, M.M., 2014. Facial emotion processing in
Arrindell, W.A., Ettema, J.H.M., 2003. SCL-90. Handleiding Bij Een Multidimensionele
borderline personality disorder: a systematic review and meta-analysis.
Psychopathologie-Indicator. Swets & Zeitlinger. https://research.tilburguniversity.
Neuropsychol. Rev. 24 (2), 166–184. https://doi.org/10.1007/S11065-014-9254-9.
edu/en/publications/scl-90-handleiding-bij-een-multidimensionele-psychopatholo
Penton-Voak, I.S., Thomas, J., Gage, S.H., McMurran, M., McDonald, S., Munafò, M.R.,
gie-ind.
2013. Increasing recognition of happiness in ambiguous facial expressions reduces
Baer, R.A., Peters, J.R., Eisenlohr-Moul, T.A., Geiger, P.J., Sauer, S.E., 2012. Emotion-
anger and aggressive behavior. Psychol. Sci. 24 (5), 688–697. https://doi.org/
related cognitive processes in borderline personality disorder: a review of the
10.1177/0956797612459657/FORMAT/EPUB.
empirical literature. https://doi.org/10.1016/j.cpr.2012.03.002.
Bannerman, R.L., Milders, M., De Gelder, B., Sahraie, A., 2009. Orienting to threat: faster
localization of fearful facial expressions and body postures revealed by saccadic eye

7
C. van Heusden et al. Psychiatry Research Communications 4 (2024) 100189

Porter, C., Palmier-Claus, J., Branitsky, A., Mansell, W., Warwick, H., Varese, F., 2020. Spielberger, C.D., Sydeman, S.J., 1994. State-trait anxiety inventory and state-trait anger
Childhood adversity and borderline personality disorder: a meta-analysis. Acta expression inventory. In: Maruish, M.E. (Ed.), The Use of Psychological Testing for
Psychiatr. Scand. 141 (1), 6–20. https://doi.org/10.1111/ACPS.13118. Treatment Planning and Outcome Assessment. L. Erlbaum Associates, pp. 292–321.
Putman, P., Hermans, E., Van Honk, J., 2006. Anxiety meets fear in perception of Terburg, D., Aarts, H., Putman, P., van Honk, J., 2012. In the eye of the beholder:
dynamic expressive gaze. Emotion 6 (1), 94–102. https://doi.org/10.1037/1528- reduced threat-bias and increased gaze-imitation towards reward in relation to trait
3542.6.1.94. anger. PLoS One 7 (2). https://doi.org/10.1371/JOURNAL.PONE.0031373.
Ricciardelli, P., Bricolo, E., Aglioti, S.M., Chelazzi, L., 2002. My eyes want to look where The jamovi project, 2020. jamovi [Computer Software], Version 1.2. https://www.
your eyes are looking: exploring the tendency to imitate another individual’s gaze. jamovi.org.
Neuroreport 13 (17), 2259–2264. https://doi.org/10.1097/00001756-200212030- Von Ceumern-Lindenstjerna, I.A., Brunner, R., Parzer, P., Mundt, C., Fiedler, P.,
00018. Resch, F., 2009. Attentional bias in later stages of emotional information processing
Schotte, C.K.W., De Doncker, D., 2000. De adp-iv: een vragenlijst voor een therapeutisch in female adolescents with borderline personality disorder. Psychopathology 43 (1),
georiënteerde diagnostiek van persoonlijkheidsstoornissen. Psychopraxis 2 (6), 25–32. https://doi.org/10.1159/000255960.
124–130. https://doi.org/10.1007/BF03071865. Weertman, A., Arntz, A., Kerkhofs, M.L.M., 2000. SCID-II; Gestructureerd Diagnostisch
Seitz, K.I., Leitenstorfer, J., Krauch, M., Hillmann, K., Boll, S., Ueltzhoeffer, K., Interview Voor DSM-IV Persoonlijkheidsstoornissen. Harcourt Test Publishers.
Neukel, C., Kleindienst, N., Herpertz, S.C., Bertsch, K., 2021. An eye-tracking study Witthöft, M., Borgmann, E., White, A., Dyer, A., 2015. Body-related attentional biases in
of interpersonal threat sensitivity and adverse childhood experiences in borderline patients with posttraumatic stress disorder resulting from childhood sexual abuse
personality disorder. Borderline Personality Disorder and Emotion Dysregulation 8 with and without co-occurring borderline personality disorder. J. Behav. Ther. Exp.
(1). https://doi.org/10.1186/S40479-020-00141-7. Psychiatr. 46, 72–77. https://doi.org/10.1016/J.JBTEP.2014.07.010.