Seitz et al.

Borderline Personality Disorder and Emotion Dysregulation

(2021) 8:2
https://doi.org/10.1186/s40479-020-00141-7

RESEARCH ARTICLE Open Access

An eye-tracking study of interpersonal

threat sensitivity and adverse childhood
experiences in borderline personality
disorder
Katja I. Seitz1*, Johanna Leitenstorfer1, Marlene Krauch1, Karen Hillmann1, Sabrina Boll1, Kai Ueltzhoeffer1,
Corinne Neukel1, Nikolaus Kleindienst2, Sabine C. Herpertz1 and Katja Bertsch1,3

Abstract
Background: Previous eye-tracking studies provide preliminary evidence for a hypersensitivity to negative,
potentially threatening interpersonal cues in borderline personality disorder (BPD). From an etiological point of
view, such interpersonal threat hypersensitivity might be explained by a biological vulnerability along with a history
of early life adversities. The objective of the current study was to investigate interpersonal threat hypersensitivity
and its association with adverse childhood experiences (ACE) in patients with BPD employing eye-tracking
technology.
Methods: We examined a sample of 46 unmedicated, adult female patients with BPD and 25 healthy female
volunteers, matched on age and intelligence, with a well-established emotion classification paradigm with angry,
fearful, happy, and neutral facial expressions. ACE were assessed retrospectively with the Childhood Trauma
Questionnaire.
Results: Patients as compared to healthy volunteers reflexively directed their gaze more quickly towards the eyes
of emotional and neutral faces and did not adapt their fixation patterns according to the facial expression
presented. Misclassifying emotional and neutral faces as angry correlated positively with the patients’ self-reported
ACE.
Conclusions: Building on and extending earlier findings, our results are likely to suggest a visual hypervigilance
towards the eyes of emotional and neutral facial expressions and a childhood trauma-related anger bias in patients
with BPD. Given the lack of a clinical control group, the question whether these findings are specific for BPD has to
remain open. Thus, further research is needed to elucidate the specificity of altered visual attention allocation and
the role of ACE in anger recognition in patients with BPD.
Keywords: Adverse childhood experiences, Borderline personality disorder, Childhood maltreatment, Eye-tracking,
Emotion dysregulation, Facial affect recognition, Interpersonal functioning, Threat hypersensitivity

* Correspondence: [email protected]
1
Department of General Psychiatry, Center for Psychosocial Medicine,
Medical Faculty, Heidelberg University, Heidelberg, Germany
Full list of author information is available at the end of the article

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons
licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the
data made available in this article, unless otherwise stated in a credit line to the data.
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 2 of 12

Background posttraumatic stress disorder (PTSD), suggesting a

Borderline personality disorder (BPD) is characterized by significant role of traumatic experiences. In two own
persistent and profound emotion dysregulation [1]. studies, we also revealed a biased visual attention in
Emotion dysregulation in BPD encompasses emotion female patients with BPD towards the eyes of emo-
sensitivity, heightened and labile negative affect, deficient tional facial expressions displayed in full intensity [27,
appropriate and excessive maladaptive emotion regula- 28]. In a functional neuroimaging study, patients with
tion strategies according to Linehan’s biosocial model BPD exhibited more and faster initial saccades to-
[2]. Emotion sensitivity, specifically the sensitivity to wards the eyes of angry faces compared to a healthy
negative or potentially threatening interpersonal signals, control group. This interpersonal threat hypersensitiv-
has been addressed in a number of recent review articles ity was associated with increased amygdala activation
[3–7]. Despite some inconsistencies (cf. [8, 9]), there is [28]. In a consecutive small behavioral study, patients
considerable empirical support for the hypothesized with BPD showed a tendency to misclassify emotional
hypersensitivity to interpersonal threat cues in BPD [3]. and neutral faces as angry, slower saccades away from
Experimental findings in mostly female patients with the eyes of fearful faces and faster saccades towards
BPD indicate a negatively biased perception of neutral, the eyes of neutral faces compared to healthy volun-
ambiguous, or positive facial expressions (“negativity teers [27]. So far, only one eye-tracking study in pa-
bias”; [10–12]), particularly a higher sensitivity in recog- tients with BPD has investigated the association
nizing anger in other emotions (“anger bias”; [13–17]) between interpersonal threat hypersensitivity and
[but, for inconsistent results, see [18, 19]. traumatization [29]. Since the authors did not analyze
Adverse childhood experiences (ACE) are seen as play- the association of ACE and indicators of interpersonal
ing a significant role in the etiology of interpersonal threat hypersensitivity, the question whether the anger
threat hypersensitivity in BPD. Patients with BPD report bias and hypervigilance towards the eyes of emotional
high rates of ACE such as emotional neglect, physical faces is indeed associated with early life adversity re-
maltreatment, or sexual abuse [20, 21]. Following Line- mains open.
han’s biosocial model [22], interpersonal threat hyper- The aim of the present study was to investigate the as-
sensitivity in BPD may emerge from an interplay sociation of ACE with the proposed interpersonal threat
between biological vulnerabilities and the exposure to hypersensitivity in patients with BPD employing eye-
such traumatic childhood experiences [1]. Studies sug- tracking technology. We hypothesized to find a biased
gest that the more patients with BPD report ACE, the perception of interpersonal threat cues in patients with
more they show a tendency to judge faces as less ap- BPD compared to healthy volunteers in terms of (1)
proachable [23], a heightened vigilance [24], and avoi- more misclassifications of facial expressions as angry, (2)
dant reactions towards angry faces [25], but the evidence more and faster initial saccades towards the eyes of
remains somewhat inconclusive (cf. [17, 26]). angry faces, and (3) longer fixation durations on the eye
To disentangle previous inconsistent findings, three region regardless of the emotion the faces are displaying.
recent studies have implemented eye-tracking technol- Moreover, we hypothesized that (4) the overall severity
ogy to capture ecologically more valid information of self-reported ACE would correlate positively with in-
about the attentional mechanisms relevant for per- dicators of threat hypersensitivity.
ceiving facial expressions in patients with BPD [27–
29]. Kaiser et al. [29] found female patients with BPD Methods
to be more likely to characterize ambivalent emo- Participants
tional blends as angry compared to a clinical control In total, 46 unmedicated female patients with a current
group (Cluster-C personality disorder) and to show DSM-IV diagnosis of BPD (BPD, Mage = 28.6, SD = 7.5,
prolonged fixations on the eye region of ambiguous range = 18–46 years) and 25 healthy women who had
facial stimuli compared to a non-patient control never received a psychiatric diagnosis or undergone any
group. The eye region is known to convey the most psychotherapeutic or psychiatric treatment (CON,
crucial information about threat-related facial expres- Mage = 26.4, SD = 5.5, range = 18–40 years) were included
sions [30, 31]. Thus, fixating extensively on the eyes in the current analyses. The sample size is large enough
of emotional faces might indicate a hypervigilance to- to detect small group by condition interactions (η2 = .05)
wards interpersonally relevant and potentially threat- as well as medium to large correlations between self-
ening cues, irrespective of the faces’ emotional report, behavioral, and eye-tracking data (r = .40) within
valence [32]. Interestingly, in the study of Kaiser et al. the BPD sample with a statistical power of 1-β ≥ .80 [33].
[29] the attentional bias towards the eyes of ambigu- Effect sizes of η2 ≥ .05 and correlations of r ≥ .40 may be
ous facial expressions was found to be most pro- expected according to previously published experimental
nounced in patients with BPD and comorbid behavioral and eye-tracking studies with female patients
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 3 of 12

with BPD or major depression and healthy volunteers study (patients: n = 1). The groups were matched with re-
[23, 27, 28, 34]. gard to age and intelligence (see Table 1 for detailed demo-
Exclusion criteria for all participants comprised neuro- graphic and psychometric information and group
logical disorders; current alcohol/drug abuse (urine toxicol- comparisons).
ogy screening) or alcohol/drug abuse in the last 2 months Participants were recruited by the central project of
(interview); severe medical illness; use of psychotropic the KFO-256, a Clinical Research Unit funded by the
medication for at least 2 weeks before participation in the German Research Foundation, investigating mechanisms
study; or impaired vision. Additional exclusion criteria for of disturbed emotion processing in BPD [36]. All pro-
the patient sample were lifetime diagnoses of schizophrenia, jects from this research unit include participants from a
schizoaffective or bipolar disorder and self-reported alco- joint database. Ethics approval was provided by the Eth-
hol/drug dependence in the last 12 months. Initially, 56 pa- ics Committee of the Medical Faculty of the University
tients and 28 healthy volunteers participated in the of Heidelberg. All participants gave their written in-
emotion classification task and filled out the Childhood formed consent and were reimbursed for their
Trauma Questionnaire (CTQ; [35]). Overall, 10 patients participation.
and 3 healthy volunteers had to be excluded due to equip-
ment malfunction (patients: n = 8, healthy volunteers: n = Measures
3), positive toxicology screenings (patients: n = 1) or neuro- Diagnoses of BPD and co-occurring axis I and II disor-
logical abnormalities, identified in a corresponding MRI ders were assessed by qualified diagnosticians using the

Table 1 Demographic and psychometric information (mean [M] ± one standard deviation [SD]) of patients with borderline
personality disorder (BPD) and healthy volunteers (CON)
BPD (n = 46) CON (n = 25) Group comparison
M ± SD M ± SD t or U p
Age (years) 28.6 ± 7.5 26.4 ± 5.5 1.27 .209
Intelligence (RAVEN) 112.0 ± 11.1 113.9 ± 9.7 −0.70 .486
Adverse Childhood Experiences (CTQ)
Sum Scorea 61.4 ± 19.4 32.0 ± 7.8 96.50 < .001
Emotional Abusea 17.2 ± 5.5 7.4 ± 3.3 95.00 < .001
Physical Abuse 8.3 ± 4.5 5.6 ± 1.4 266.50 < .001
Sexual Abuse 9.1 ± 6.5 5.1 ± 0.3 300.00 < .001
Emotional Neglect 16.7 ± 6.4 8.0 ± 3.1 162.50 < .001
Physical Neglect 9.6 ± 3.6 5.9 ± 1.7 199.50 < .001
b
Borderline Symptomatology (BSL-23) 1.6 ± 0.8 0.2 ± 0.3 37.00 < .001
Depressiveness (BDI-II)b, c 23.4 ± 10.6 3.0 ± 5.0 39.50 < .001
Trait Anxiety (STAI)b 62.1 ± 7.4 33.2 ± 8.2 15.03 < .001
Emotional Dysregulation (DERS)a 128.5 ± 20.0 62.8 ± 16.2 14.04 < .001
Current Comorbid Axis I Disorders (lifetime)
Affective Disorders 18 (36) 0 (0)
Posttraumatic Stress Disorder 8 (13) 0 (0)
Other Anxiety Disorders 25 (29) 0 (0)
Substance Use Disorders 0 (13) 0 (0)
Eating Disorders 12 (24) 0 (0)
Somatization Disorders (current only) 2 0
Adjustment Disorder 0 (0) 0 (0)
Current Comorbid Axis II Disorders (lifetime)
Avoidant Personality Disorder 13 (13) 0 (0)
Antisocial Personality Disorder 2 (2) 0 (0)
Groups were matched with respect to age (in years) and intelligence (IQ scores). RAVEN Raven’s Standard Progressive Matrices, CTQ Childhood Trauma
Questionnaire, BSL-23 Borderline Symptom List (Short Version), BDI-II Beck Depression Inventory-II, STAI State-Trait Anxiety Inventory, DERS Difficulties in Emotion
Regulation Scale
a
Data of one patient with BPD is missing . b Data of two patients with BPD are missing . c Data of one healthy volunteer is missing
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 4 of 12

Structured Clinical Interview for DSM-IV (SCID-I for Experimental protocol and data acquisition
axis I diagnoses; [37]) and the International Personality All participants underwent a telephone screening for in-
Disorder Examination for DSM-IV (IPDE for BPD and clusion and exclusion criteria (approximately 45 min), and
comorbid avoidant and antisocial personality disorders; an onsite diagnostic session (approximately 3 h). After ex-
[38]). Prior to the study, diagnosticians completed a cluding acute substance abuse by a urine toxicology
standardized diagnostic training, leading to high levels of screening, participants performed the emotion classifica-
inter-rater reliability (ICC ≥ .911). tion task (approximately 1 h, including instruction, train-
ACE were measured with the Childhood Trauma ing, and short breaks between presentation time blocks).
Questionnaire (CTQ; [35]), assessing emotional, physical Eye-tracking data were recorded with a 60-Hz mon-
and sexual abuse, and emotional and physical neglect. ocular eye-tracking system (ViewPoint, Arrington Re-
BPD symptom severity (short version of the Borderline search, Scottsdale, AZ, USA). The head location was
Symptom List, BSL-23; [39]), depressiveness (revised fixed using a chin rest and a forehead bar with a viewing
version of Beck’s Depression Inventory, BDI-II; [40]), distance of 57 cm. Before each experimental block, the
trait anxiety (State-Trait Anxiety Inventory, STAI; [41]) eye-tracking camera was adjusted, eight training trials
and emotion regulation (Difficulties in Emotion Regula- were presented and nine-point calibration was applied.
tion Scale, DERS; [42]) were assessed with self-report Stimulus presentation and response recording were per-
questionnaires for further dimensional characterization formed using the software Presentation (Version 16.3,
of the present sample. Intelligence was assessed using Neurobehavioral Systems, Albany, CA, USA). Stimuli
Raven’s Standard Progressive Matrices [43] (see online were presented on an Eizo FlexScan S2202 display (47.5
supplement for psychometric properties of cm × 30.0 cm) with a resolution of 1680 × 1050 pixels
questionnaires). and a refresh rate of 60 Hz.
As shown in Table 1, patients with BPD revealed
higher scores in all state and trait questionnaires com- Data analysis
pared to healthy volunteers, with the exception of Ra- Details are presented in the online supplement. Follow-
ven’s Standard Progressive Matrices (matching variable). ing the procedure of previous studies (e.g., [27, 32]), we
Patients with BPD reported more severe ACE, more extracted behavioral data (i.e., proportion of correct re-
BPD symptoms, and higher levels of emotion dysregula- sponses, types of errors, response latencies in trials with
tion, depressiveness, and trait anxiety. correct emotion classifications) and eye-tracking data
(i.e., proportion and latency of initial saccades, fixation
Emotion classification paradigm durations). Initial saccades were classified according to
The emotion classification task was based on a 2 × 2 × 4- whether they were directed towards the other major fa-
design, with the within-subject factors presentation time cial feature (i.e., eyes, mouth). When the eyes were pre-
(150 ms, 5000 ms), initial fixation (eyes, mouth) and sented at the former position of the fixation cross, the
emotional expression (angry, fearful, happy, neutral) (for proportion and latency of the downward fixation
similar designs, see [27, 32, 44, 45]). Employing these changes towards the mouth were scored, whereas when
two presentation time conditions allows for investigating the mouth was shown at fixation, the proportion and la-
reflexive attentional shifts (150 ms) as well as sustained tency of the upward fixation changes towards the eyes
attention (5000 ms) towards diagnostically relevant facial were scored. Fixation durations were defined as the cu-
features. In each presentation time condition, partici- mulative amount of time participants spent looking at
pants were required to classify 80 emotional faces, un- either the eye or the mouth region in the long condition.
ambiguously displaying angry, fearful, happy, and neutral Proportion of correct responses, response latencies as
expressions (see also online supplement), as quickly and well as proportion and latency of initial saccades were
as accurately as possible by pressing one of four corre- submitted to 2 × 2 × 2 × 4 mixed-design analyses of vari-
sponding keys. To control for initial fixation, half of the ance (ANOVAs) with the between-subject factor group
faces within each emotional category were unpredictably (BPD, CON) and the within-subject factors presentation
shifted either downward or upward in each trial. Each time (brief, long), initial fixation (eyes, mouth), and emo-
trial started with the presentation of a fixation cross tional expression (angry, fearful, happy, neutral). Types
(2000 ms), followed by a facial stimulus with the eyes or of errors were analyzed using a 2 × 2 × 4 mixed-design
the mouth appearing at the former position of the fix- ANOVA with the between-subject factor group (BPD,
ation cross (brief condition: 150 ms; long condition: CON) and the within-subject factors presentation time
5000 ms), and a blank screen in the brief condition only (brief, long) and error type (misclassifications of faces as
(1850 ms). Each trial ended with another fixation cross angry, fearful, happy, neutral). Finally, fixation duration
with a variable duration (1000–3000 ms) to reduce an- data (only available in the long presentation time condi-
ticipation effects. tion) were subjected to a 2 × 2 × 4 mixed-design
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 5 of 12

ANOVA with the between-subject factor group (BPD, proportion of correct responses and types of errors.
CON) and the within-subject factors facial feature (eyes, Emotion recognition accuracy was high for both patients
mouth) and emotional expression (angry, fearful, happy, with BPD (M = 93.6%, SD = 5.0%) and healthy volunteers
neutral). In accordance with previous studies with the (M = 93.5%, SD = 4.9%) across experimental conditions.
same experimental paradigm (e.g., [27, 43]), ANOVAs Participants were significantly better at recognizing facial
were also conducted separately for both presentation expressions in the long than in the brief condition (main
time conditions to identify additional group effects. We effect presentation time, F [1, 69] = 42.27, p < .001, η2 =
applied the Huynh-Feldt procedure [46] to correct for .38). While no significant effect including the factor
potential violations of the sphericity assumption where group emerged across presentation time conditions (all
indicated. Partial eta squared (η2) is provided as an effect F ≤ 2.29, p ≥ .103, η2 ≤ .03), separate analyses of the two
size index. In cases of significant effects, Dunn’s Multiple presentation time conditions revealed a significant group
Comparisons are reported as post-hoc tests. by emotional expression interaction in the brief condi-
Furthermore, we performed bivariate correlational tion (F [3, 207] = 3.07, p = .041, η2 = .04), but not in the
analyses between self-reported ACE of patients with BPD long condition (F [3, 207] = 0.75, p = .467, η2 = .01). In
(CTQ sum score), and behavioral (i.e., misclassifications the brief condition, healthy volunteers were significantly
as angry) and eye-tracking measures of interpersonal more accurate in classifying happy compared to fearful
threat hypersensitivity (i.e., proportion and latency of ini- faces (p < .01) as well as neutral compared to angry
tial saccades towards the eyes or the mouth of angry faces, (p < .01) and fearful faces (p < .01), while this was not
fixation durations on the eyes or the mouth of angry shown for patients with BPD. However, pairwise com-
faces). Correlations are reported as Pearson product- parisons between patients with BPD and healthy volun-
moment correlation coefficients (r) with Bonferroni-Holm teers did not reach statistical significance. Furthermore,
correction for multiple comparisons [47]. patients with BPD and healthy volunteers misclassified
Statistical analyses were conducted using the statistical facial expressions more often as angry than as happy or
programming language R (version 3.6.2) and SPSS for as neutral and more often as fearful than as happy (all p-
Windows (version 26). A significance threshold of values < .01) across presentation time conditions (main
p < .05, two-tailed, was set for all statistical analyses. effect error type, F [3, 207] = 10.75, p < .001, η2 = .14; see
Fig. 1a). No significant group difference was found in
Results terms of types of errors when analyzing the two presen-
Hypothesis 1 tation time conditions together or separately (no signifi-
To test our first hypothesis, proposing that patients with cant effect including the factor group, all F ≤ .48,
BPD would misclassify facial expressions more often as p ≥ .581, η2 ≤ .01). Thus, these results did not confirm
angry compared to healthy volunteers, we analyzed our a priori hypothesis.

Fig. 1 Anger bias in patients with BPD and healthy volunteers and its association with patients’ self-reported ACE. a. Types of errors (mean
amount of errors ± one standard error) of patients with borderline personality disorder (BPD) and healthy volunteers (CON) across presentation
time conditions. The displayed graph represents the significant main effect of error type, with significantly more misclassifications as angry than
as happy or as neutral and significantly more misclassifications as fearful than as happy in both groups. Groups did not differ in their types of
errors. ** p < .01. b. Association of patients’ self-reported ACE (CTQ sum score) and total amount of faces misclassified as angry in both
presentation time conditions. The displayed graph represents the positive correlation between misclassifications as angry and ACE in patients
with BPD (r = .43, p = .023, Bonferroni-Holm-corrected)
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 6 of 12

Hypothesis 2 With regard to latencies of initial saccades, we could

To investigate our second hypothesis, stating that patients analyze only a subsample of n = 37 patients with BPD
with BPD would show a visual hypervigilance towards the and n = 19 healthy volunteers in the long condition and
eyes of angry faces compared to healthy volunteers, we an- n = 22 patients with BPD and n = 8 healthy volunteers in
alyzed proportion and latency of initial saccades. the brief condition since the other participants did not
Participants generally made more initial saccades in show any initial saccade (eye movements exceeding 1°
the long than in the brief condition (main effect presen- within the time interval of 150 to 1000 ms after facial
tation time, F [1, 69] = 122.01, p < .001, η2 = .64). In stimulus onset) towards the other major facial feature in
addition, a significant group by presentation time inter- at least one of the eight experimental conditions (initial
action emerged (F [1, 69] = 6.62, p = .012, η2 = .09), with fixation × emotional expression). Since fixations may
patients with BPD showing more initial saccades com- rest on the initially fixated major facial feature without
pared to healthy volunteers in the brief condition (p < any shifts of orientation, and initial saccades may not be
.05), however, irrespective of emotional expression or required for accurate emotion recognition, these trials
initial fixation (see Figure 2a). Separate analyses of the may not be classified as invalid [27]. According to a
two presentation time conditions did not yield any add- post-hoc sensitivity analysis, the sample size of 30 partic-
itional significant group effects. ipants is still large enough to detect medium group by

Fig. 2 Proportion and latency of initial saccades in patients with BPD and healthy volunteers in the brief condition. a. Mean proportion of initial
saccades (in %, ± one standard error) to the eyes (left graph) and to the mouth (right graph) in patients with borderline personality disorder
(BPD) and healthy volunteers (CON) in the brief condition (150 ms). The displayed graphs represent the significant group by presentation time
interaction, with patients with BPD showing more initial saccades compared to healthy volunteers across all experimental conditions in the brief
condition. Please note that only significant group differences are highlighted; refer to text and online supplement for further significant effects. *
p < .05. b. Mean latency of initial saccades (in ms, ± one standard error) to the eyes (left graph) and the mouth (right graph) of patients with
borderline personality disorder (BPD) and healthy volunteers (CON) in the brief condition (150 ms). The displayed graphs represent the significant
group by initial fixation interaction, with patients with BPD showing faster initial saccades towards the eyes compared to healthy volunteers.
Please note that only significant group differences are highlighted; refer to text and online supplement for further significant effects. * p < .05
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 7 of 12

condition interactions (η2 = .07) with a statistical power feature, F[1, 69] = 100.26, p < .001, η2 = .59) and happy
of 1-β ≥ .80 [33]. Overall, participants made faster initial faces longer than angry and fearful faces (main effect
saccades in the long than in the brief condition (main ef- emotional expression, F[3, 207] = 6.29, p < .001, η2 = .08).
fect presentation time, F [1, 26] = 20.99, p < .001, η2 = With regard to our a priori hypothesis, a significant fa-
.45). Furthermore, a significant group by initial fixation cial feature by emotional expression interaction (F[3,
interaction emerged (F [1, 26] = 8.16, p = .008, η2 = .24), 207] = 23.13, p < .001, η2 = .25) and a significant group by
with patients with BPD showing faster initial saccades facial feature by emotional expression interaction (F[3,
towards the eyes compared to healthy volunteers. This 207] = 5.65, p = .001, η2 = .08) were revealed. While
effect was qualified by a significant group by presenta- healthy volunteers fixated the eyes of angry, fearful, and
tion time by initial fixation interaction (F [1, 26] = 4.53, neutral faces significantly longer than the eyes of happy
p = .043, η2 = .15; see Fig. 2b). Post-hoc pairwise compar- faces and the mouth of happy faces significantly longer
isons within each presentation time condition revealed than the mouth of angry, fearful, and neutral faces (all p-
that patients with BPD exhibited faster initial saccades values < .01), this could not be shown for patients with
towards the eyes than towards the mouth in the long BPD (see Fig. 3). Contrary to our a priori hypothesis,
condition (p < .05) and faster initial saccades towards the however, patients with BPD did not fixate the eyes of
eyes compared to healthy volunteers in the brief condi- emotional and neutral faces longer than healthy
tion (p < .05). In the brief condition, healthy volunteers volunteers.
showed the exact opposite gazing behavior with faster
initial saccades towards the mouth than towards the eyes Hypothesis 4
(p < .01). Thus, and partly in line with our a priori hy- To test our fourth hypothesis concerning the positive as-
pothesis, patients with BPD showed a visual hypervigi- sociation between interpersonal threat hypersensitivity
lance in terms of faster initial saccades towards the eyes and ACE in patients with BPD, we performed correl-
of briefly presented emotional and neutral faces in gen- ational analyses with the CTQ sum score in the patient
eral rather than towards the eyes of angry faces in sample only. In patients with BPD, self-reported ACE
particular. were positively correlated with misclassifying fearful,
happy, and neutral faces as angry (r = .43, p = .023,
Hypothesis 3 Bonferroni-Holm corrected; see Fig. 1b). Due to outliers,
To examine our third hypothesis, expecting longer fix- we winsorized our data (i.e., replaced the total numbers
ation durations on the eye region regardless of emo- of misclassifications as angry of five patients with BPD
tional valence in patients with BPD compared to healthy with the next highest number that was not an outlier
volunteers, we analyzed the fixation durations in the [48]). By doing this, the association was reduced to a
long condition. Participants generally fixated the eye re- non-significant statistical trend (r = .28, p = .062). Correl-
gion longer than the mouth region (main effect facial ational analyses between ACE and eye-tracking

Fig. 3 Fixation patterns of patients with BPD and healthy volunteers. Mean fixation duration on the eye (a) or the mouth region (b) relative to
the whole face (in %, ± one standard error) of patients with borderline personality disorder (BPD) and healthy volunteers (CON) in the long
condition (5000 ms). The displayed graphs represent the significant group by facial feature by emotional expression interaction, with patients with
BPD not modulating their fixation patterns according to the emotion presented like healthy volunteers. Please note that only significant group
differences are highlighted; refer to text for further significant effects. ** p < .01
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 8 of 12

performance in reaction to angry faces did not yield any of more readily identifying signals of interpersonal threat
significant associations (r ≤ .38, p ≥ .185, Bonferroni- may have been increased.
Holm corrected). Please note that the correlational ana- Our results are only partly in line with our second hy-
lyses with latencies of initial saccades are only based on pothesis. Patients with BPD showed more initial sac-
a smaller subsample (see Hypothesis 2). Taken together, cades compared to healthy volunteers in the brief
these results partly confirmed our a priori hypothesis. condition, regardless of emotional expression and initial
fixation. This enhanced tendency to reflexively direct at-
tention towards diagnostically relevant features of facial
Additional behavioral results
expressions could be interpreted as a manifestation of a
In addition to our hypothesis-driven results and in line
hypervigilant mindset constantly searching for potential
with previous studies (e.g., [27, 28, 32]), we were able to
interpersonal threat cues in the environment [51]. More-
replicate a number of more general behavioral and eye-
over, patients with BPD showed faster initial saccades
tracking effects of the emotion classification paradigm in
towards the eyes of briefly presented emotional and neu-
patients with BPD and healthy volunteers, which are
tral expressions. Recognizing threatening facial expres-
presented in an online supplement.
sions (i.e., anger) relies mostly on allocating visual
attention to the face’s eye region [30, 31]. Hence, direct-
Discussion ing one’s gaze as quickly as possible towards the facial
To the best of our knowledge, this is the first study to feature that provides the most salient information about
investigate the relationship between behavioral and eye- interpersonal threat cues might mirror a highly reflexive
tracking measures of interpersonal threat sensitivity and hypersensitivity for potential signs of social threat in pa-
ACE in patients with BPD. Our results, although partly tients with BPD. Contrary to our second hypothesis,
inconsistent, point towards a visual hypervigilance for however, the enhanced proportion and latency of initial
the eye region and a childhood trauma-related anger saccades were not specific for the eyes of angry faces (cf.
bias in patients with BPD, potentially supporting the [28]), but could be found for any facial expression pre-
prevalent assumption that interpersonal threat hypersen- sented. Thus, our findings might also point towards a
sitivity is indeed associated with early life adversity in visual hypervigilance for social cues in general rather
BPD. than for interpersonal threat cues in particular. In line
In line with previous studies (e.g., [27, 28]), both pa- with Kaiser et al. [29], factors other than interpersonal
tients with BPD and healthy volunteers were highly ac- threat hypersensitivity should be considered to elucidate
curate in classifying facial expressions across this visual attention bias for the eye region of emotional
presentation time conditions. Consistent with the notion faces in general rather than of angry faces in particular.
that patients with BPD do not experience general im- Following Lazarus et al. [52], inconsistent findings con-
pairments in emotion recognition but specific subtle def- cerning facial affect recognition in BPD might be ex-
icits in the processing of interpersonal threat cues [3], plained by differences in the experimental setup (i.e.,
patients with BPD did not differ from healthy volunteers fMRI vs. laboratory, set of facial stimuli), influence of
with regard to the proportion of correctly identified fa- psychotropic medications, and co-occurring symptom
cial expressions. Contrary to our first hypothesis, how- disorders. With regard to experimental methodology
ever, both patients and healthy volunteers misclassified and medication status, a hypervigilance for the eye re-
facial expressions more often as angry than as happy or gion was found in medicated and unmedicated patients
as neutral across presentation time conditions, thus re- with BPD presented with ambiguous or full intensity
vealing an anger bias which was not specific for patients emotional expressions in a previous study by Kaiser
with BPD. This result was unexpected since several stud- et al. [29] and the present study. However, high rates of
ies document an enhanced tendency to perceive anger in comorbid axis I disorders, especially a 30% rate of BPD
emotional faces in patients with BPD compared to patients with a current diagnosis of social phobia, might
healthy or clinical control groups (e.g., [13–17, 27, 29]), account for this generalized hypersensitivity for social
which seems to persist even after symptomatic remission stimuli in the present sample. This would be in line with
[49] [but see [50], for inconsistent findings]. Differences a previous eye-tracking study [32], reporting an en-
in the type of stimuli used may be considered as one hanced attentional orienting towards the eye region in
possible explanation for this divergent finding. With the patients with social phobia as compared to healthy con-
exception of Bertsch et al. [27, 28], in all of the above- trols, regardless of the type of emotional expression
mentioned studies, pictures of morphed instead of full displayed.
intensity emotional facial expressions were presented. Contrary to our third hypothesis, patients with BPD
Thereby, the difficulty of the experimental task, the did not fixate the eye region longer than healthy vol-
probability of misclassifications and thus, the tendency unteers, irrespective of emotional valence. Hence, our
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 9 of 12

results are inconsistent with a recent eye-tracking representativeness of our sample, it also questions
study [29], reporting longer fixations on the eye re- the specificity of our results for BPD, especially since
gion of ambiguous faces in patients with BPD com- our study lacks a clinical control group. The absence
pared to non-patients. Our findings are, however, in of a clinical control group therefore precludes any
line with a previous eye-tracking study of our group inferences regarding the BPD-specific nature of our
[27], which also failed to detect longer fixation dura- findings. Third, since our healthy control group re-
tions on the eye region of emotional expressions dis- ported only limited exposure to ACE (CTQ sum
played in full intensity in a smaller, independent score), we focused the correlational analyses with
sample of female patients with BPD compared to early life adversity on the patient group only. Of
healthy volunteers. Employing full intensity rather note, exploratory correlational analyses between ACE
than ambiguous blends of emotional facial expressions and behavioral and eye-tracking measures of inter-
might, again, have played a role in producing these personal threat sensitivity in the healthy control
heterogeneous findings across studies. Nevertheless, group did not yield any significant associations
patients with BPD did exhibit aberrant fixation pat- (r ≤ .41, p ≥ .691, Bonferroni-Holm corrected). Due to
terns compared to healthy volunteers. While healthy insufficient power and lack of variance in CTQ
volunteers fixated longer on the facial feature more scores, however, we cannot draw any conclusions
relevant for the recognition of the distinct emotional from these non-significant findings. It therefore re-
expression presented, patients with BPD did not adapt mains unclear whether the association between inter-
their fixation patterns accordingly. personal threat hypersensitivity and ACE is due to
In line with our fourth hypothesis, the results of the diagnosis of BPD or due to early life maltreat-
the correlational analyses revealed more misclassifica- ment in general. Future studies should include BPD
tions of facial expressions as angry in patients with patient, healthy and clinical control groups, each
more self-reported ACE. This result should, however, with male and female participants with a broad
be interpreted cautiously. Due to high emotion recog- range of ACE, to get a better insight in the specifi-
nition accuracy in patients with BPD, the total num- city of this association. Fourth, although our sample
ber of anger misclassifications lacks sufficient variance size was large enough to detect small to medium
and our correlational analysis might be biased by out- group by condition interactions for all behavioral
liers. Nonetheless, calculating the correlation with and eye-tracking measures of interpersonal threat
winsorized misclassifications as angry still revealed a hypersensitivity as well as medium to large correla-
positive, medium-sized association which can be char- tions between these measures of interpersonal threat
acterized as a non-significant statistical trend (p < .10). hypersensitivity and ACE, it was not adequate to de-
Together with previous studies [23–25, 53], the tect small to medium correlations. This applies espe-
current findings may provide evidence supporting the cially to the analyses of saccadic latencies which had
assumption that interpersonal threat hypersensitivity to be conducted with a smaller subsample of pa-
is indeed associated with early life adversity in pa- tients. Given the lack of power, expected associations
tients with BPD [but see [17], for inconsistent re- between ACE and eye-tracking measures of interper-
sults]. Contrary to our expectations, however, we did sonal threat hypersensitivity in general, and latencies
not find significant associations between eye-tracking of initial saccades in particular, might have not been
measures of interpersonal threat hypersensitivity and detected and further replication studies with larger
ACE. groups of patients with BPD are needed to allow for
Our study has some strengths, investigating behav- strong conclusions about non-significant effects.
ioral and eye-tracking measures of interpersonal Fifth, we presented full intensity static facial stimuli
threat sensitivity and their association with ACE in that do not adequately mirror the dynamics and
well-matched, unmedicated samples of patients with subtlety of emotional expressions in our everyday so-
BPD and healthy volunteers. Several limitations cial interactions. Employing recently developed dy-
should, however, be acknowledged. First, we included namic sets of non-stereotypical facial expressions
only female participants. Since sex differences in vis- [56, 57]) in future studies may extend our findings
ual attention allocation during facial emotion recog- to more naturalistic settings. Moreover, it might ren-
nition have been documented [54], we cannot draw der the experimental task more difficult and thus
conclusions concerning patients with BPD of other lead to a higher variance in emotion recognition ac-
sex than female. Second, our patient sample reported curacy and types of misclassifications. Sixth, we ap-
high rates of psychiatric comorbidities. While high plied a matching approach to “control for” potential
rates of axis I comorbidities are a common finding differences in sex, age, and intelligence between the
in the literature [55] and thus emphasize the patient and healthy control group. According to
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 10 of 12

classic papers by Meehl [58], and Miller and Chap- Authors’ contributions
man [59], by doing so, we might have generated a SCH and KB designed the study. JL, MK, KH and CN acquired the data, which
KIS, JL, SB, KU, NK and KB analyzed. KIS and KB wrote the article, which all
systematic mismatch in other variables and thus in- authors reviewed and approved for publication.
troduced artifacts in our data. Finally, due to the
cross-sectional design of our study, causal inferences Funding
This work was supported by grants from the German Research Foundation
cannot be drawn from our findings. This is of par- (DFG), awarded to K. Bertsch and S. C. Herpertz (grant numbers: BE 5292/2–1,
ticular importance with regard to the correlational BE 5292/3–2, HE 2660/14–1, HE 2660/12–2). The funding source was not
analyses, given that retrospective, subjective self- involved in the collection, analysis and interpretation of data; in the writing
of the report; and in the decision to submit the article for publication. Open
reports of ACE are prone to various memory biases Access funding enabled and organized by Projekt DEAL.
[60] and cannot be interpreted interchangeably with
prospective measures of ACE [61]. Availability of data and materials
The datasets used and analyzed during the current study are available from
the corresponding author on reasonable request.
Conclusions
Ethics approval and consent to participate
In summary, our current study suggests an association The study was conducted in accordance with the ethical principles of the
between interpersonal threat hypersensitivity in terms of Declaration of Helsinki and approved by the Ethics Committee of the
an anger bias in BPD and ACE. Preliminary models pos- Medical Faculty of the University of Heidelberg. All participants gave written
informed consent to participate in the study.
tulate that the hypervigilance towards potentially threat-
ening interpersonal cues might lead to severe Consent for publication
impairments in psychosocial functioning of patients with Not applicable.
BPD [62]. Effective and efficient treatment strategies to
Competing interests
reduce the attentional bias for potential signs of social The authors declare that they have no competing interests.
threat are therefore needed. In BPD, increased amygdala
activation to (negative) emotional stimuli has been sug- Author details
1
Department of General Psychiatry, Center for Psychosocial Medicine,
gested as a neural underpinning of interpersonal threat Medical Faculty, Heidelberg University, Heidelberg, Germany. 2Institute for
hypersensitivity [3] and deficient amygdala habituation Psychiatric and Psychosomatic Psychotherapy, Central Institute of Mental
to threatening cues has been shown to be associated Health, Medical Faculty Mannheim, Heidelberg University, Mannheim,
Germany. 3Department of Psychology, Ludwig-Maximilians-University
with ACE [63]. In a recent single-arm trial, patients with Munich, Munich, Germany.
BPD succeeded in downregulating their amygdala activa-
tion and reported less BPD symptoms after receiving Received: 29 July 2020 Accepted: 25 November 2020

four sessions of amygdala neurofeedback [64]. Together

with psychotherapeutic interventions to reduce threat- References
related attentional biases (cf. [65]), these efforts might 1. Crowell SE, Beauchaine TP, Linehan MM. A biosocial developmental model
of borderline personality: elaborating and extending Linehan's theory.
lead to novel treatment approaches to reduce threat Psychol Bull. 2009. https://doi.org/10.1037/a0015616.
hypersensitivity and interpersonal dysfunction in pa- 2. Carpenter RW, Trull TJ. Components of emotion dysregulation in borderline
tients with BPD and ACE. personality disorder: a review. Curr Psychiatry Rep. 2013. https://doi.org/10.
1007/s11920-012-0335-2.
3. Bertsch K, Hillmann K, Herpertz SC. Behavioral and neurobiological
correlates of disturbed emotion processing in borderline personality
Supplementary Information disorder. Psychopathology. 2018. https://doi.org/10.1159/000487363.
The online version contains supplementary material available at https://doi. 4. Mancke F, Herpertz SC, Bertsch K. Aggression in borderline personality
org/10.1186/s40479-020-00141-7. disorder: a multidimensional model. Personal Disord. 2015. https://doi.org/
10.1037/per0000098.
Additional file 1. Supplemental_Material The supplemental material 5. Daros AR, Zakzanis KK, Ruocco AC. Facial emotion recognition in borderline
contains more extensive detail on our methods (psychometric properties personality disorder. Psychol Med. 2013. https://doi.org/10.1017/
of the questionnaires used in our study, stimulus material, and analysis of s0033291712002607.
eye-tracking data) and our results (replication of general behavioral and 6. Domes G, Schulze L, Herpertz SC. Emotion recognition in borderline
eye-tracking effects of the emotion classification paradigm). personality disorder - a review of the literature. J Personal Disord. 2009.
https://doi.org/10.1521/pedi.2009.23.1.6.
7. Mitchell AE, Dickens GL, Picchioni MM. Facial emotion processing in
Abbreviations borderline personality disorder: a systematic review and meta-analysis.
ACE: Adverse childhood experiences; BPD: Borderline personality disorder Neuropsychol Rev. 2014. https://doi.org/10.1007/s11065-014-9254-9.
8. Matzke B, Herpertz SC, Berger C, Fleischer M, Domes G. Facial reactions
during emotion recognition in borderline personality disorder: a facial
Acknowledgements electromyography study. Psychopathology. 2014. https://doi.org/10.1159/
The authors thank K. Herwig for supporting the eye-tracking measurements, 000351122.
C. Bermeo, M. Elsaesser, J. Froehlich, P. Ikier, J. Nusskern, L. Rang, C. Roth, M. 9. Mier D, Lis S, Esslinger C, Sauer C, Hagenhoff M, Ulferts J, et al. Neuronal
Schmitz and A. Steinbrenner for their help with data collection, and the team correlates of social cognition in borderline personality disorder. Soc Cogn
of the Clinical Research Group on Mechanisms of Disturbed Emotion Pro- Affect Neurosci. 2013. https://doi.org/10.1093/scan/nss028.
cessing in Borderline Personality Disorder (KFO 256) for participant recruit- 10. Catalan A, Gonzalez de Artaza M, Bustamante S, Orgaz P, Osa L, Angosto V,
ment and organization. et al. Differences in facial emotion recognition between first episode
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 11 of 12

psychosis, borderline personality disorder and healthy controls. PLoS One. 30. Schurgin MW, Nelson J, Iida S, Ohira H, Chiao JY, Franconeri SL. Eye
2016. https://doi.org/10.1371/journal.pone.0160056. movements during emotion recognition in faces. J Vis. 2014. https://doi.org/
11. Daros AR, Uliaszek AA, Ruocco AC. Perceptual biases in facial emotion 10.1167/14.13.14.
recognition in borderline personality disorder. Personal Disord. 2014. https:// 31. Calvo MG, Fernandez-Martin A, Gutierrez-Garcia A, Lundqvist D. Selective
doi.org/10.1037/per0000056. eye fixations on diagnostic face regions of dynamic emotional expressions:
12. Fenske S, Lis S, Liebke L, Niedtfeld I, Kirsch P, Mier D. Emotion recognition in KDEF-dyn database. Sci Rep. 2018. https://doi.org/10.1038/s41598-018-
borderline personality disorder: effects of emotional information on 35259-w.
negative bias. Borderline Personal Disord Emot Dysregulation. 2015. https:// 32. Boll S, Bartholomaeus M, Peter U, Lupke U, Gamer M. Attentional
doi.org/10.1186/s40479-015-0031-z. mechanisms of social perception are biased in social phobia. J Anxiety
13. Berenson KR, Dochat C, Martin CG, Yang X, Rafaeli E, Downey G. Disord. 2016. https://doi.org/10.1016/j.janxdis.2016.04.004.
Identification of mental states and interpersonal functioning in borderline 33. Faul F, Erdfelder E, Lang AG, Buchner A. G*power 3: a flexible statistical
personality disorder. Personal Disord. 2018. https://doi.org/10.1037/ power analysis program for the social, behavioral, and biomedical sciences.
per0000228. Behav Res Methods. 2007. https://doi.org/10.3758/bf03193146.
14. Ferreira G, Sanches R, Crippa J, Mello M, Osório F. Borderline personality 34. Bodenschatz CM, Skopinceva M, Russ T, Suslow T. Attentional bias and
disorder and bias in the recognition of facial expressions of emotion: a childhood maltreatment in clinical depression - an eye-tracking study. J
pathway to understand the psychopathology. Arch Clin Psychiatry. 2018. Psychiatr Res. 2019. https://doi.org/10.1016/j.jpsychires.2019.02.025.
https://doi.org/10.1590/0101-60830000000146. 35. Bernstein DP, Fink L. Childhood Trauma Questionnaire: a retrospective self-
15. Domes G, Czieschnek D, Weidler F, Berger C, Fast K, Herpertz SC. report. Manual. San Antonio: The Psychological Corporation; 1998.
Recognition of facial affect in borderline personality disorder. J Personal 36. Schmahl C, Herpertz SC, Bertsch K, Ende G, Flor F, Kirsch P, et al.
Disord. 2008. https://doi.org/10.1521/pedi.2008.22.2.135. Mechanisms of disturbed emotion processing and social interaction in
16. Izurieta Hidalgo NA, Oelkers-Ax R, Nagy K, Mancke F, Bohus M, Herpertz SC, borderline personality disorder: state of knowledge and research agenda of
et al. Time course of facial emotion processing in women with borderline the German clinical research unit. Borderline Personal Disord Emot
personality disorder: an ERP study. J Psychiatry Neurosci. 2016. https://doi. Dysregulation. 2014. https://doi.org/10.1186/2051-6673-1-12.
org/10.1503/jpn.140215. 37. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured clinical interview
17. Veague HB, Hooley JM. Enhanced sensitivity and response bias for male for DSM-IV (SCID-I). New York State Psychiatric Institute, Biometrics Research
anger in women with borderline personality disorder. Psychiatry Res. 2014. Department: New York; 1995.
https://doi.org/10.1016/j.psychres.2013.12.045. 38. Loranger AW, Sartorius N, Andreoli A, Berger P, Buchheim P,
18. Thome J, Liebke L, Bungert M, Schmahl C, Domes G, Bohus M, et al. Channabasavanna SM, et al. The international personality disorder
Confidence in facial emotion recognition in borderline personality disorder. examination (IPDE): the World Health Organization/alcohol, drug abuse, and
Personal Disord. 2016. https://doi.org/10.1037/per0000142. mental health administration international pilot study of personality
19. van Dijke A, van’ t Wout M, Ford JD, Aleman A. Deficits in degraded facial disorders. Arch Gen Psychiatry. 1994. https://doi.org/10.1001/archpsyc.1994.
affect labeling in schizophrenia and borderline personality disorder. PLoS 03950030051005.
One. 2016. https://doi.org/10.1371/journal.pone.0154145. 39. Bohus M, Kleindienst N, Limberger MF, Stieglitz RD, Domsalla M, Chapman
20. de Aquino Ferreira LF, Queiroz Pereira FH, Neri Benevides AML, Aguiar Melo AL, et al. The short version of the borderline symptom list (BSL-23):
MC. Borderline personality disorder and sexual abuse: a systematic review. development and initial data on psychometric properties. Psychopathology.
Psychiatry Res. 2018. https://doi.org/10.1016/j.psychres.2018.01.043. 2009. https://doi.org/10.1159/000173701.
21. Zanarini MC, Williams AA, Lewis RE, Reich RB, Vera SC, Marino MF, et al. 40. Beck AT, Steer RA, Brown GK. Manual for the Beck depression inventory-II.
Reported pathological childhood experiences associated with the San Antonio: The Psychological Corporation; 1996.
development of borderline personality disorder. Am J Psychiatry. 1997. 41. Laux L, Glanzmann P, Schaffner P, Spielberger CD. Das state-trait-
https://doi.org/10.1176/ajp.154.8.1101. Angstinventar (STAI). Beltz: Göttingen; 1981.
22. Linehan M. Cognitive–behavioral treatment of borderline personality 42. Gratz K, Roemer L. Multidimensional assessment of emotion regulation and
disorder. New York: Guilford Press; 1993. dysregulation: development, factor structure, and initial validation of the
23. Nicol K, Pope M, Sprengelmeyer R, Young AW, Hall J. Social judgement in difficulties in emotion regulation scale. J Psychopathol Behav Assess. 2004.
borderline personality disorder. PLoS One. 2013. https://doi.org/10.1371/ https://doi.org/10.1007/s10862-008-9102-4.
journal.pone.0073440. 43. Heller KA, Kratzmeier H, Lengfelder A. Matrizen-test-manual, band 2. Ein
24. Kaiser D, Jacob GA, van Zutphen L, Siep N, Sprenger A, Tuschen-Caffier Handbuch mit deutschen Normen zu den advanced progressive matrices
B, et al. Patients with borderline personality disorder and comorbid von J. C. Raven [matrix test manual. A handbook for Raven’s advanced
PTSD show biased attention for threat in the facial dot-probe task. J progressive matrices]. Göttingen: Beltz; 1998.
Behav Ther Exp Psychiatry. 2020. https://doi.org/10.1016/j.jbtep.2018.11. 44. Scheller E, Büchel C, Gamer M. Diagnostic features of emotional expressions
005. are processed preferentially. PLoS One. 2012. https://doi.org/10.1371/journal.
25. Bruene M, Ebert A, Kolb M, Tas C, Edel MA, Roser P. Oxytocin pone.0041792.
influences avoidant reactions to social threat in adults with borderline 45. Boll S, Gamer M. 5-HTTLPR modulates the recognition accuracy and
personality disorder. Hum Psychopharmacol. 2013. https://doi.org/10. exploration of emotional facial expressions. Front Behav Neurosci. 2014.
1002/hup.2343. https://doi.org/10.3389/fnbeh.2014.00255.
26. Lowyck B, Luyten P, Vanwalleghem D, Vermote R, Mayes LC, Crowley MJ. 46. Huynh H, Feldt LS. Estimation of the box correction for degrees of freedom
What's in a face? Mentalizing in borderline personality disorder based on from sample data in randomized block and split-plot designs. J Educ Behav
dynamically changing facial expressions. Personal Disord. 2016. https://doi. Stat. 1976. https://doi.org/10.3102/10769986001001069.
org/10.1037/per0000144. 47. Holm S. A simple sequentially rejective multiple test procedure. Scand Stat
27. Bertsch K, Krauch M, Stopfer K, Haeussler K, Herpertz SC, Gamer M. Theory Appl. 1979;6:65–70.
Interpersonal threat sensitivity in borderline personality disorder: an eye- 48. Dixon WJ. Simplified estimation from censored normal samples. Ann Math
tracking study. J Personal Disord. 2017. https://doi.org/10.1521/pedi_2017_ Stat. 1960;31:385–91.
31_273. 49. Kleindienst N, Hauschild S, Liebke L, Thome J, Bertsch K, Hensel S, et al. A
28. Bertsch K, Gamer M, Schmidt B, Schmidinger I, Walther S, Kästel T, et al. negative bias in decoding positive social cues characterizes emotion
Oxytocin and reduction of social threat hypersensitivity in women with processing in patients with symptom-remitted borderline personality
borderline personality disorder. Am J Psychiatry. 2013. https://doi.org/10. disorder. Borderline Personal Disord Emot Dysregulation. 2019. https://doi.
1176/appi.ajp.2013.13020263. org/10.1186/s40479-019-0114-3.
29. Kaiser D, Jacob GA, van Zutphen L, Siep N, Sprenger A, Tuschen-Caffier B, 50. Schneider I, Bertsch K, Izurieta Hidalgo NA, Muller LE, Schmahl C, Herpertz
et al. Biased attention to facial expressions of ambiguous emotions in SC. Remnants and changes in facial emotion processing in women with
borderline personality disorder: an eye-tracking study. J Personal Disord. remitted borderline personality disorder: an EEG study. Eur Arch Psychiatry
2019. https://doi.org/10.1521/pedi_2019_33_363. Clin Neurosci. 2018. https://doi.org/10.1007/s00406-017-0841-7.
Seitz et al. Borderline Personality Disorder and Emotion Dysregulation (2021) 8:2 Page 12 of 12

51. Kimble M, Boxwala M, Bean W, Maletsky K, Halper J, Spollen K, et al. The

impact of hypervigilance: evidence for a forward feedback loop. J Anxiety
Disord. 2014. https://doi.org/10.1016/j.janxdis.2013.12.006.
52. Lazarus SA, Cheavens JS, Festa F, Zachary RM. Interpersonal functioning in
borderline personality disorder: a systematic review of behavioral and
laboratory-based assessments. Clin Psychol Rev. 2014. https://doi.org/10.
1016/j.cpr.2014.01.007.
53. Wagner AW, Linehan MM. Facial expression recognition ability among
women with borderline personality disorder: implications for emotion
regulation? J Personal Disord. 1999. https://doi.org/10.1521/pedi.1999.13.4.
329.
54. Hall JK, Hutton SB, Morgan MJ. Sex differences in scanning faces: does
attention to the eyes explain female superiority in facial expression
recognition? Cognit Emot. 2010. https://doi.org/10.1080/
02699930902906882.
55. Shah R, Zanarini MC. Comorbidity of borderline personality disorder: current
status and future directions. Psychiatr Clin North Am. 2018. https://doi.org/
10.1016/j.psc.2018.07.009.
56. Calvo MG, Fernandez-Martin A, Recio G, Lundqvist D. Human observers and
automated assessment of dynamic emotional facial expressions: KDEF-dyn
database validation. Front Psychol. 2018. https://doi.org/10.3389/fpsyg.2018.
02052.
57. Yitzhak N, Giladi N, Gurevich T, Messinger DS, Prince EB, Martin K, et al.
Gently does it: humans outperform a software classifier in recognizing
subtle, nonstereotypical facial expressions. Emotion. 2017. https://doi.org/10.
1037/emo0000287.
58. Meehl PE. High school yearbooks: a reply to Schwarz. J Abnorm Psychol.
1971. https://doi.org/10.1037/h0030750.
59. Miller GA, Chapman JP. Misunderstanding analysis of covariance. J Abnorm
Psychol. 2001. https://doi.org/10.1037/0021-843X.110.1.40.
60. Baldwin JR, Reuben A, Newbury JB, Danese A. Agreement between
prospective and retrospective measures of childhood maltreatment: a
systematic review and meta-analysis. JAMA Psychiatry. 2019. https://doi.org/
10.1001/jamapsychiatry.2019.0097.
61. Danese A. Annual research review: rethinking childhood trauma-new
research directions for measurement, study design and analytical strategies.
J Child Psychol Psychiatry. 2020. https://doi.org/10.1111/jcpp.13160.
62. Herpertz SC, Jeung H, Mancke F, Bertsch K. Social dysfunctioning and brain
in borderline personality disorder. Psychopathology. 2014. https://doi.org/10.
1159/000365106.
63. Bilek E, Itz ML, Stößel G, Ma R, Berhe O, Clement L, et al. Deficient amygdala
habituation to threatening stimuli in borderline personality disorder relates
to adverse childhood experiences. Biol Psychiatry. 2019. https://doi.org/10.
1016/j.biopsych.2019.06.008.
64. Zähringer J, Ende G, Santangelo P, Kleindienst N, Ruf M, Bertsch K, et al.
Improved emotion regulation after neurofeedback: a single-arm trial in
patients with borderline personality disorder. NeuroImage Clin. 2019.
https://doi.org/10.1016/j.nicl.2019.102032.
65. Vazquez C, Duque A, Blanco I, Pascual T, Poyato N, Lopez-Gomez I, et al.
CBT and positive psychology interventions for clinical depression promote
healthy attentional biases: an eye-tracking study. Depress Anxiety. 2018.
https://doi.org/10.1002/da.22786.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.