SEARCH

SEARCH BY CITATION

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. METHOD
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Mothers with borderline personality disorder (BPD) have disturbed relationships with their infants, possibly associated with poor nonverbal cue perception. Individuals with BPD are poor at recognizing emotion in adults and tend to misattribute neutral (i.e., no emotion) as sad. This study extends previous research by examining how mothers with BPD perceive known (own) and unknown (control) infant stimuli depicting happy, sad, and neutral emotions. The sample consisted of 13 women diagnosed with BPD and 13 healthy control mothers. All participants completed clinical and parenting questionnaires and an infant emotion recognition task. Compared to control mothers, mothers with BPD were significantly poorer at infant emotion recognition overall, but especially neutral expressions which were misattributed most often as sad. Performance was not related to disturbed parenting but rather mothers' age and illness duration. Neither the BPD nor control mothers showed enhanced accuracy for emotional displays of their own verses unknown infant-face images. Although the sample size was small, this study provides evidence that mothers with BPD negatively misinterpret neutral images, which may impact sensitive responding to infant emotional cues. These findings have implications for clinical practice and the development of remediation programs targeting emotion-perception disturbances in mothers with BPD.

Las madres con Trastorno de Personalidad Limítrofe (BPD) tienen relaciones preocupantes con sus infantes, posiblemente asociadas con la pobre percepción de la señal no verbal. Las personas con BPD tienen una débil facilidad de reconocer la emoción en los adultos y tienden a malinterpretar lo neutral (sin emoción) como triste. Este estudio extiende la investigación previa examinando cómo las madres con BPD perciben el estímulo del infante, conocido (propio) y desconocido (control) representando las emociones felices, tristes y neutras. El grupo muestra fue de 13 mujeres diagnóstico de BPD y 13 madres saludables en el grupo de control. Todas las participantes completaron cuestionarios clínicos y sobre la crianza y una tarea de reconocimiento de la emoción del infante. Comparadas con las madres del grupo de control, las madres con BPD se mostraron significativamente más débiles en cuanto al reconocimiento de la emoción del infante en general, pero especialmente las expresiones neutrales las cuales fueron mal atribuidas más a menudo como tristes. La ejecución no se relacionó con una crianza trastornada sino con la edad de las madres y la duración de la enfermedad. Ni las madres del grupo BPD ni las del grupo de control mostraron una precisión acrecentada para la demostración emocional de sí mismas frente a las imágenes faciales desconocidas del infante. Aunque el tamaño de la muestra fue pequeño, este estudio aporta evidencia de que las madres con BPD malinterpretan las imágenes neutrales negativamente, lo cual pudiera impactar la respuesta sensible a las señas emocionales del infante. Estos resultados tienen implicaciones en la práctica clínica y en el desarrollo de programas de remedio que se enfocan en los trastornos de percepción emocional en madres con BPD.

Les mères souffrant du trouble de personnalité limite (BPD) ont des relations perturbées avec leurs nourrissons, peut-être liées à une faible perception du repère non-verbal. Les individus souffrant de trouble de personnalité limite ne sont pas très bons pour reconnaître les émotions chez les adultes et ont tendance à mal attribuer le neutre (pas d'émotion) comme étant triste. Cette étude développe les recherches en examinant la manière dont les mères souffrant de trouble de personnalité limite perçoivent les stimuli connus (le sien) et inconnus (contrôle) du nourrisson en décrivant des émotions heureuses, tristes, et neutres. L'échantillon a consisté en 13 femmes diagnostiquées avec le trouble de personnalité limite et 13 mères en bonne santé pour le groupe de contrôle. Toutes les participantes ont rempli des questionnaires cliniques et des questionnaires de parentage ainsi qu'un exercice de reconnaissance de l'émotion du nourrisson. Comparées aux mères du groupe de contrôle, les mères avec le trouble de personnalité limite ont bien moins réussi à l'exercice de reconnaissance de l'émotion du nourrisson en général, mais surtout pour ce qui concerne les expressions neutres qui étaient souvent mal attribuées comme tristes. Le résultat n'était pas lié à un parentage perturbé mais plutôt à l'âge des mères et à la durée de la maladie. Ni les mères souffrant de trouble de personnalité limite ni les mères du groupe de contrôle n'ont fait preuve d'exactitude supérieure pour les étalages émotionnels de leur propre enfant par rapport à des images de visages de nourrissons inconnus. Bien que l'échantillon d'étude était petit cette étude montre que les mères ayant le trouble de personnalité limite interprétent les images neutres de façon négative, ce qui peut avoir un impact sur la réponse sensible aux signaux émotionnels du nourrisson. Ces résultats ont des implications pour la pratique clinique et le développement de programmes de remédiation ciblant les troubles de la perception de l'émotion chez les mères souffrant de trouble de personnalité limite.

Mütter mit Borderline-Persönlichkeitsstörung (BPS) haben gestörte Beziehungen zu ihren Säuglingen, die möglicherweise mit einer schlechten Wahrnehmung nonverbaler Signale verbunden sind. Personen mit BPS sind schlecht in der Gefühlserkennung Erwachsener und neigen dazu, neutral (keine Emotion) als traurig zu misattribuieren. Diese Studie erweitert frühere Forschungen, indem sie untersucht, wie Mütter mit BPS bekannte (eigene) und unbekannte (Kontrollkinder) Säuglinge, die glückliche, traurige und neutrale Gefühle zeigen, wahrnehmen. Die Stichprobe bestand aus 13 Frauen mit diagnostizierter BPS und 13 gesunden Kontrollmüttern. Alle Teilnehmer vervollständigten klinische Fragebögen sowie Fragebögen zur Elternschaft und nahmen an einer Aufgabe zur Emotionserkennung bei Säuglingen teil. Im Vergleich zu Müttern der Kontrollgruppe waren Mütter mit BPS insgesamt signifikant schlechter in der Emotionserkennung bei den Säuglingen, aber vor allem neutrale Ausdrücke wurden meist als traurige fehlinterpretiert. Die Leistung stand nicht mit gestörter Elternschaft im Zusammenhang, sondern eher mit dem Alter und der Krankheitsdauer der Mütter. Die BPS-Mütter und die Kontrollmütter zeigten keine verbesserte Genauigkeit bei der Deutung von den emotionalen Gesichtsbildern der eigenen Säuglinge im Vergleich zu denen unbekannter Säuglinge. Obwohl der Stichprobenumfang gering war, liefert diese Studie Hinweise darauf, dass Mütter mit BPS neutrale Bilder als negative missverstehen, was sich möglicherweise auf die sensitive Reaktion infolge emotionaler Signale des Säuglings auswirken kann. Diese Erkenntnisse haben Implikationen für die klinische Praxis und die Entwicklung von Förderprogrammen, die auf Emotionswahrnehmungsstörungen bei Müttern mit BPS zielen.

抄録:境界パーソナリティ障害Borderline Personality Disorder (BPD)の母親は、乳児との関係性が障害されていて、それはおそらく非言語的な合図の知覚の悪さと関連しているだろう。BPDの個人は大人の情緒の認識が悪く、中性の情緒(情緒なし)を悲しいと誤ってとらえる傾向がある。この研究では、うれしい、悲しい、そして中性の情緒を表現している、知っている(自分の)乳児と見知らぬ(対照の)乳児の刺激を、BPDの母親がどのように知覚するかを調査することによって、先行研究を拡張した。対象は、13人のBPDと診断された女性と、13人の健康対照群女性から構成された。すべての参加者は、臨床と育児の質問紙に記入し、乳児の情緒認識課題を行った。対照群の母親と比較すると、BPDの母親は、乳児の全体としての情緒認知が有意に悪かったが、特に中性の表情がしばしば悲しいと誤ってとらえられた。パフォーマンスは、障害のある育児とは関係しなかったが、むしろ母親の年齢と病気の持続期間に関連した。見知らぬ乳児の表情イメージと比べて、自分自身の子どもの表情イメージの情緒表出の正確さが高まることは、BPDの母親にも対照群の母親にも示されなかった。対象の数は少なかったが、この研究から、BPDの母親は中性のイメージを否定的に誤って解釈するという事実が示された。これは乳児の情緒的な合図に敏感に応答する上で、影響があるだろう。これらの所見は、臨床と、BPDの母親の情緒認知障害の改善を目的としたプログラムの開発に、意味を持つ。

Borderline personality disorder (BPD) is characterized by emotion dysregulation, impulsiveness, intense and unstable relationships, substance abuse, self-harming behaviors, experiences of emotional invalidation, and often childhood trauma (Eaton et al., 2011; Linehan, 1993). While BPD is equally prevalent in men and women (with approximately 0.7–5.9% of the population affected), it is predominantly diagnosed in women, most of whom are in their childbearing years (Coid, Yang, Tyrer, Roberts, & Ullrich, 2006; Grant et al., 2008). Central to the pathology of the disorder are difficulties with the recognition and labeling of emotion in oneself and in others, and problems regulating the intensity with which emotions are perceived and experienced (Levine, Marziali, & Hood, 1997; Linehan, 1993). These disturbances significantly affect the interpersonal interactions of women with BPD, including the relationship with their infant(s).

A mother's ability to accurately identify and respond sensitively to her infant's different emotional cues allows her to soothe, or regulate, an infant's emotional experiences (Gunnar & Donzella, 2002; Gunnar & Quevedo, 2007; Schore, 2001a; Spinrad, Stifter, Donelan-McCall, & Turner, 2004). This capacity is recognized as the basis for establishing secure mother–infant attachment (Bowlby, 1969; Grienenberger, Kelly, & Slade, 2005) and is considered fundamental to the healthy cognitive, social, emotional, and physical development of an infant (e.g., Perry, 2002; Schore, 2001a, 2001b; Siegel, 2001). Mothers with BPD are observed to have difficulties with responding sensitively to their children's emotional needs (Newman & Stevenson, 2005), with labeling and responding to infant emotional communication, and parental reflective capacity; that is, understanding the inner emotional world of the infant (Slade, 2005; Slade, Grienenberger, Bernbach, Levy, & Locker, 2005).

Mother–infant interactions in this population are characterized by high levels of stress, low parental self-efficacy, and increased parental intrusiveness, control, overprotectiveness, hostility, inconsistency, avoidance, and in some cases, abuse (Crandell, Patrick, & Hobson, 2003; Hobson, Patrick, Crandell, García-Pérez, & Lee, 2005; Hobson et al., 2009; Newman & Stevenson, 2005; Newman, Stevenson, Bergman, & Boyce, 2007; White, Flanagan, Martin, & Silvermann, 2011). The extent to which these disturbances are due to problems with the perception of the infant's nonverbal emotional cues is unclear. Understanding the deficits that underlie disturbed mother–infant interactions in women with BPD is not only important for informing clinical practice but also for targeting treatment strategies toward specific deficits and developing remediation programs to improve the outcomes of at-risk mother–infant dyads.

Women with BPD also are known to have disturbances in social cognition, particularly associated with emotion perception (Preißler, Dziobek, Ritter, Heekeren, & Roepke, 2010). Perception appears particularly impaired for negative emotions such as sadness, anger, disgust, and fear (e.g., Bland, Williams, Scharer, & Manning, 2004; Levine et al., 1997) while the identification of emotions such as happy (Fertuck et al., 2009; Lynch et al., 2006; Unoka, Fogd, Fuzy, & Csukly, 2011) and fearful (Wagner & Linehan, 1999) remain intact. A consistent finding throughout the literature is that of a negative misattribution bias, with BPD participants rating adult-face images as being less trustworthy, less friendly, or rejecting than do healthy controls (Meyer, Pilkonis, & Beevers, 2004; Murphy, 2006). This negative misattribution bias is particularly apparent for neutral (no emotion) and ambiguous face images (i.e., faces showing artificial blends of more than one emotion) (e.g., Domes et al., 2008; Dyck et al., 2009; Meyer et al., 2004; Murphy, 2006; Wagner & Linehan, 1999). A recent meta-analysis by Daros, Zakzanis, and Ruocco (2012) using unknown adult-face stimuli has confirmed these findings, showing patients with BPD to be significantly less accurate than controls at recognizing anger and disgust, and to have a pronounced deficit for the recognition of neutral face images.

The impact of emotion perception deficits in women with BPD and young infants has not yet been explored. Research in nonclinical populations has shown that parents differentially process images of their own (i.e., familiar) as compared to unknown (i.e., novel) infant-face images. In an event-related potential (ERP) study, birth and foster/adoptive parents produced ERP patterns suggestive of increased attention allocation to their own infant's images as compared with images of unknown infants or adults (Grasso, Moser, Dozier, & Simons, 2009). A functional magnetic resonance imaging study that grouped mothers on the basis of attachment style (secure or insecure-dismissing) found that mothers with a secure attachment style displayed significantly greater brain activation in the lateral prefrontal cortex bilaterally and the left medial prefrontal cortex than did mothers with an insecure attachment style (Strathearn, Fonagy, Amico, & Montague, 2009). Securely attached mothers continued to show greater activation in the hypothalamic/pituitary region (where oxytocin is produced and released peripherally) when viewing happy and sad images of their own infant as compared to unknown infant images. By contrast, insecure mothers showed increased activation of the anterior insula, a region associated with feelings of unfairness, sadness, anger, and disgust. This study did not assess emotion-recognition accuracy concurrently, limiting understanding of how these brain-activation differences might be associated with the quality of maternal recognition of infants' emotional displays. However, the study findings indicate that mothers with attachment disturbance (also a common feature of BPD; see Choi-Kain, Fitzmaurice, Zanarini, Laverdière, & Gunderson, 2009; Fonagy, Target, & Gergely, 2000; Levy, Meehan, Weber, Reynoso, & Clarkin, 2005; Meyer et al., 2004) do not display an observable attentional preference for the emotional displays of their own infant. This lack of attentional preference may, in turn, reduce the ability of mothers with attachment disturbance to accurately interpret the emotional displays of their infant thereby increasing the likelihood of inappropriate responses to infant distress (Leerkes & Siepak, 2006), which over time may have adverse consequences for the infant's socioemotional, cognitive, and physical well-being and development (e.g., Perry, 2002; Schore, 2001a, 2001b; Siegel, 2001).

There is a strong need, given the paucity of research and the potential serious long-term consequences for infants, to examine mother–infant interactions in women with BPD, particularly those related to emotion perception and engagement, to help further scientific understanding and to inform clinical practice (Newman et al., 2007). The influence of attachment-relevant stimuli on emotion-perception ability has not been examined previously. To our knowledge, the present study is the first to investigate infant-face emotion perception in mothers with BPD and in healthy control mothers. To address some of the limitations observed in previous studies, this research utilized images of both the mothers' own infant (attachment-relevant) and those of unknown (control) infants displaying three emotional valences (happy, sad, and neutral). It was expected that, compared with the healthy control group, mothers with BPD would be significantly poorer at accurately recognizing the emotional displays of infants, and more specifically, would show a misattribution pattern with a tendency to more negatively interpret neutral facial expressions of infants. It was further expected that control mothers would show preferential attention for images of their own infant, measured by greater emotion-recognition accuracy, as compared with images of unknown infants, whereas mothers with BPD, who have been observed previously to have attachment-related disturbances (e.g., Choi-Kain et al., 2009; Fonagy et al., 2000; Levy et al., 2005; Meyer et al., 2004), would not show this differential pattern. The clinical symptom profile and parental behaviors of mothers with BPD and controls also was assessed. It was predicted that emotion-recognition deficits in mothers with BPD (particularly the negative misattribution of neutral infant expressions) would be significantly correlated with higher scores on parenting stress, perceptions of the infant and mother–infant relationship as difficult, and hostile/reactive behavior and overprotectiveness toward the infant, and correlated with lower scores on perceived parental self-efficacy and perceived parental impact on infant development. It also was expected that parenting variables would correlate significantly with key BPD clinical symptoms (i.e., BPD symptom severity, depression, childhood trauma, high adult attachment anxiety and avoidance).

METHOD

  1. Top of page
  2. ABSTRACT
  3. METHOD
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Participants

The participants consisted of 13 mothers with a diagnosis of BPD, who were between 17 and 40 years of age (M = 24.15, SD = 6.72), and 13 healthy mothers (control group), who were between 18 and 31 years (M = 28.31, SD = 3.45). Healthy control mothers were recruited from Hunter New England Area Health Service (HNEAHS) staff bulletin boards and child and family health services. Mothers with BPD were recruited from parenting and mental health services located within the HNEAHS. The research team was independent of the health services targeted for recruitment and was not responsible for the clinical care of any of the participants. Posters about the study specifically targeting mothers with BPD were displayed in health service waiting rooms, and treating clinicians were asked to refer potential participants to the study. All women with BPD were receiving treatment for either mental health or parenting concerns at the time of recruitment. The average treatment duration for BPD participants was 25 months (range = 4–96 months, SD = 23.53). Eight of the women with BPD were prescribed antidepressant medication (including sertraline, escitalopram, duloxetine, citalopram, desvenlafaxine, or venlafaxine) at a therapeutic dose. Another woman was prescribed a combination of antidepressants (mirtazapine and venlafaxine) and antipsychotic medication (low-dose olanzapine at an anxiolytic dose). All of the healthy control women and 4 women with BPD were not prescribed psychiatric medication. A high proportion of the mothers with BPD (76.9%), but none of the healthy control women, also had community services involvement due to child-protection concerns.

Study inclusion criteria for participants were age 17 years or older and a healthy infant between 3 and 14 months old. Exclusion criteria for participants were mild to severe developmental disability (IQ <70), current Axis I disorder, intoxication at time of assessment, or substance use in the last 30 days. Exclusion criteria for infants were a known neurological condition, severe delay, and prematurity or traumatic head injury.

Clinical and Parenting Measures

All participants completed a clinical assessment consisting of the Diagnostic Interview for Psychosis (DIP; Castle et al., 2006) to screen for a history of psychosis, major depression, or mania; the Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD; Zanarini et al., 2003) to measure borderline personality psychopathology as described by the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text-revision (American Psychiatric Association, 2000); the Edinburgh Postnatal Depression Scale (EPNDS; Cox, Holden, & Sagovsky, 1987) and the Beck Depression Inventory, second edition (BDI-II; Beck, Steer, & Brown, 1996) to measure self-reported depressive symptoms; and the Alcohol Use Disorders Identification Test (AUDIT; Saunders, Aasland, Babor, de la Fuente, & Grant, 1993) to screen for alcohol abuse and/or dependency. Intoxication also was assessed on the day of testing using clinical observation. Parenting measures consisted of the Childhood Trauma Questionnaire (CTQ; Bernstein & Fink, 1998) to assess sexual, physical, and emotional abuse and neglect; the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) to assess six facets of difficulties in regulating emotion; the Experiences in Close Relationships Scale (ECRS; Brennan, Clarke, & Shaver, 1998) to measure adult attachment patterns; the Parenting Stress Index–Short Form (PSI-SF; Abidin, 1995) to assess overall parenting stress; and the Parental Cognitions and Conduct Toward the Infant Scale (PACOTIS; Boivin et al., 2005) to assess parenting beliefs, attitudes, and behavior toward the infant.

Infant Face Stimuli

Infant-face stimuli consisted of 30 color images of male and female infant faces (650 × 650 pixels); 15 images were of the participant's own infant (comprising five happy, five sad, and five neutral images), and 15 were of unknown infants (five happy, five sad, and five neutral images). Infant images were rated for valence and grouped into one of three emotion categories (happy, sad, neutral) by two independent raters not associated with the project. Only those images where raters were in 100% agreement about valence were retained. Own and unknown images were matched as closely as possible for gender, age, ethnicity, and intensity of emotion. To counter the possibility that infants of mothers with BPD may display atypical emotional expression as a result of chronic exposure to negative parental affect and that BPD mothers may find it easier to identify expressions displayed by healthy control infants, near-equal numbers of images of infants of healthy control mothers (seven images total) and of mothers with BPD (eight images total) were selected for inclusion in the unknown-infant images for both BPD and control mothers. Images of unknown infants were therefore not consistent across participants. All images were standardized to remove any obvious gender differences (e.g., earrings, clothing), identifying features or distractions (e.g., birthmarks, scratches, saliva) before resizing, orientating, and image presentation.

Procedure

Ethics approval for the study was obtained from the Hunter New England Human Research Ethics Committee and the Human Research Ethics Committee of the University of Newcastle. Written informed consent was obtained from all participants prior to entry into the study. Mothers who met study inclusion criteria were invited to complete the study in two phases: Phase 1, clinical assessment and videotaping of infant expressions, and Phase 2, infant emotion-recognition assessment.

In Phase 1, participants completed a clinical assessment involving mental health and parenting assessments. In a separate room, with the mother's permission, each infant was recorded during play with a trained staff member to obtain varying facial expressions (happy, sad, and neutral). Images were later standardized in accordance with a facial-affect coding scheme similar to that used by Cole, Barrett, and Zahn-Waxler (1992). Mothers were not present during the recording of their infant to ensure that all facial images would be novel when presented to mothers in Phase 2.

Approximately 4 weeks (M = 25 days, SD = 15.06) after Phase 1, mothers were invited to return to complete Phase 2 of the study and participate in the infant emotion-recognition assessment task. T tests revealed no significant differences between groups for length of time between participating in Phase 1 and Phase 2 of the study, t(24) = −1.97, p = .061, with the BPD group averaging 19.46 days (SD = 15.44) and the control group averaging 30.46 day (SD = 12.98). To reduce mothers' distraction, a qualified staff member cared for infants in a separate room during all assessment tasks. Mothers were seated 60 cm from a computer screen. Randomized, full-screen infant-face images were presented on the computer for 6 s each, using Experiment Builder software (Eyelink, Ontario, Canada). The 6-s presentation time was selected based on previous research (Campbell, McCabe, Leadbeater, Schall, & Loughland, 2010; Loughland, Williams, & Gorden 2002a; Loughland, Williams, & Gorden, 2002b; McCabe, Rich, Loughland, Schall, & Campbell, 2011; Williams, Loughland, & Gordon, 1999; Williams, Loughland, Green, Harris, & Gordon, 2003) and to minimize issues of impulsivity that are thought to affect emotion-recognition performance in mothers with BPD (e.g., Dyck et al., 2009). Each infant-face image was followed by a separate screen displaying a thumbnail image of the previously presented image (to control for memory effects) along with the three possible emotion-selection options (happy, sad, neutral) displayed below the image. Participants were instructed to provide a verbal response, and the researcher manually recorded responses. Image order for each category (own/happy, own/neutral, own/sad, unknown/happy, unknown/neutral, unknown/sad) was pseudorandomized, but remained consistent between each participant.

Data Analysis

SPSS 18.0 was used to analyze all data. Parametric and nonparametric statistics were used; however, given that the patterns of results did not differ with either statistic, the results for parametric tests are presented for the purposes of simplicity. Independent samples t tests were conducted to examine differences between groups for continuous sociodemographic and clinical variables, and chi-square tests were undertaken for categorical variables. For the t tests, Welch–Satterthwaite corrected degrees of freedom and significance values were reported, as appropriate. Emotion-recognition accuracy data were analyzed using a mixed between-/within-subjects analysis of variance (ANOVA), with group (BPD, control) as the between-subjects factor and emotion type (happy, neutral, sad) and picture type (own infant, unknown infant) as the within-subjects factors. When significant overall F statistics were obtained, planned comparisons were performed to examine the differences for independent variables between groups. Finally, correlation analysis was conducted to explore the relationship between recognition accuracy and clinical and parenting variables.

RESULTS

  1. Top of page
  2. ABSTRACT
  3. METHOD
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Sample Characteristics and Sociodemographic Profile

Sample and sociodemographic characteristics for the BPD and control groups are reported in Table 1. Chi-square analyses of sociodemographic variables revealed significant differences between groups, with BPD participants less likely to be married or in a de facto relationship, p < .001, to report significantly fewer years of education, p < .001, and have a lower level of family income, p = .003, relative to the control group. No significant group differences were observed for maternal age, infant age, infant birth order, or infant gender.

Table 1. Sample Characteristics and Sociodemographic Profiles of Borderline Personality Disorder (BPD) and Control Participants
Sample CharacteristicsBPD GroupControl GroupStatistical Difference
  1. a

    Welch–Satterthwaite corrected. bExact p value. cAll monetary amounts are expressed in Australian dollars.

  2. *p < .05, two-tailed. **p < .01, two-tailed. ***p < .001, two-tailed.

N (all participants female)1313 
Maternal Age (years)
Range17–4018–31 
M (SD)24.15 (6.72)28.31 years (3.45)t = −1.98
Infant Age (months)
Range3–143–12 
M (SD)7.73 (3.98)6.50 (2.65)t = 0.93a
Infant Sex  χ2 = 0.62
Female, n75 
Infant Birth Orderb  χ2 = 4.14
Firstborn610 
Second born43 
Third born20 
Fourth born10 
Maternal Education levelb  χ2 = 16.62***
<Year 1070 
Completed Year 10, 11, or 1252 
Tertiary Qualification111 
Yearly Family Incomeb  χ2 = 11.59**
<$35,000c71 
$35,001– $80,00065 
> $80,00007 
Maternal Marital Statusb  χ2 = 19.11***
Married18 
De facto15 
Divorced/Separated20 
Single90 

Infant Emotion Recognition Accuracy

A three-way, mixed between-/within-subjects ANOVA was conducted examining the relationship between group (BPD, control), emotion (happy, sad, and neutral), and image type (own vs. unknown infant image). No main effect for image type (own vs. unknown infant), p = 1.0, or interaction between image type and group, p = .08, was observed, suggesting that BPD and control mothers did not differ from each other in their ability to recognize their own infant's emotional expressions, as compared with those of unknown infants. However, a significant main effect for group was observed for recognition accuracy of infant emotions, F(1, 24) = 14.39, p = .001, suggesting that overall (i.e., across all emotions and image types), mothers with BPD were significantly less able to accurately recognize infant expressions of emotion, as compared with the control mothers. Cohen's d for the three emotion effects were .15, 1.02, and .26 for happy, neutral, and sad faces, respectively. Follow-up contrasts revealed no significant difference between BPD and control mothers when viewing sad infant faces, as compared to happy, F(1, 24) = .24, p = .63, or neutral faces, F(1, 24) = 3.5, p = .07, when collapsed across own and unknown infant image categories. However, relative to the control group, mothers with BPD were significantly less accurate at identifying neutral expressions, as compared with happy ones, F(1, 24) = 9.58, p = .01.

There also was a significant interaction between emotion type and group, F(1.26, 30.17) = 4.67, p = .03. Additional repeated measures ANOVAs were conducted using separate analyses for the BPD group and the control group, with the two within-subject variables (emotion and image type) forming the model. The main effect of image type in each of the subanalyses tests the significance of the image effect within each group. For the BPD group, the image effect was not significant, F(1,12) = 1.0, p = .34. However, the finding was significant for the control group, F(1,12) = 4.6, p = .05, indicating that image type played a role in recognition accuracy for the control group, although only at a marginal level. This finding also was in an unexpected direction; that is, control mothers were slightly more accurate in recognizing emotional displays for unknown infants, as compared to the emotional displays of their own infant.

Regarding misattribution, Table 2 provides the mean and standard deviations for errors made for each emotion. Misattribution of emotions was examined using t tests and showed that the groups differenced significantly from each other with respect to neutral being misidentified most often as sad, BPD = 27 vs. HC = 4 neutral images as sad; t(17.1) = 3.37, p = .004, but did not differ with respect to the number of happy images misidentified as neutral, BPD = 3 vs. HC = 0 happy images as neutral; p > .05, or sad images misidentified as neutral, BPD = 7, HC = 7. No other significant misattributions were observed.

Table 2. Patterns of Misattribution for Infant Emotion Recognition Accuracy
 Borderline Personality DisorderHC 
Type of ErrorM (SD)M (SD)Statistic
  1. a

    These categories are the number of total errors made to faces depicting the listed emotion.

  2. *Significant within-group differences: p < .05. **p < .01.

All Errors for Sad*0.46 (1.13)0.77 (0.28) 
Sad as Neutral0.46 (1.13)0.77 (0.28) 
Sad as Happy0 (0)0 (0) 
All Errors for Neutrala1.77 (1.36)0.31 (0.63)**t(16.9) = 3.51, p = .003
Neutral as sSad1.54 (1.26)0.23 (0.60)**t(17.1) = 3.37, p = .004
Neutral as Happy0.23 (0.60)0.08 (0.28) 
All Errors or Happy*0.23 (0.60)0 (0) 
Happy as Neutral0.23 (0.60)0 (0) 
Happy as Sad0 (0)0 (0) 

Negative misattribution of neutral faces as sad was only significantly related with two largely interchangeable variables in the BPD group, mothers' age, r = .66, p = .01, and duration of the BPD, r = .7, p = .01, but not with level of education, p = .62, medication status, p = .52, or treatment duration, p = .71, suggesting that the findings were independent of treatment effects and sociodemographic variables. Although mothers with BPD displayed poorer emotion recognition accuracy overall, as compared to the control mothers, recognition accuracy levels nonetheless remained relatively high across both groups (i.e., minimum accuracy for the BPD group was 78.5%; minimum accuracy for the control group was 93.8%), suggesting performance was well above chance (i.e., 33.33%; see Table 2) and not the result of guessing or poor motivation.

image

Figure 1. Recognition of infant image type (own, unknown) by emotion (happy, sad, neutral, and total) by group (BPD, Control).

Download figure to PowerPoint

Clinical and Parental Profiles

Clinical and parenting assessment measures are presented in Table 3. The BPD group produced higher scores across all BPD clinical assessment measures, all ps < .001. Scores for depression, BDI-II: p < .001, EPNDS: p < .001, total emotional dysregulation, DERS: p < .001, and each DERS subscale (Difficulties with Emotional Clarity, Emotional Awareness, Impulse Control, Emotional Nonacceptance, Difficulty Engaging in Goal-Directed Behavior, Limited Access to Emotion-Regulation Strategies), all ps < .001, also were significantly higher in the BPD group. Scores for adult social attachment (ECRS) measures, including attachment anxiety, attachment avoidance, and combined attachment avoidance and anxiety, all ps < .001, were also significantly higher. However, substance use did not differ between groups based on AUDIT scores, p = .76, although significantly more BPD (38.46%) than control participants (0%) reported a history of substance use, p = .04.

Table 3. Means (SDs) for Borderline Personality Disorder Symptoms, Emotional Regulation, and Childhood Trauma Assessments for Borderline Personality Disorder (BPD) and Control Groups
 BPD GroupControl Group  
Clinical ProfileM (SD)M (SD)Statistical Difference, tdf
  1. ZAN-BPD = Zanarini Rating Scale for Borderline Personality Disorder; BDI-II = Beck Depression Inventory (2nd ed.); EPDS Edinburgh Postnatal Depression Scale; AUDIT = Alcohol Use Disorders Identification Test; ECRS = Experiences in Close Relationships Scale; DERS = Difficulties in Emotion Regulation Scale; CTQ = Childhood Trauma Questionnaire; PACOTIS = Parental Cognitions and Conduct Toward the Infant Scale; PSI-SF = Parenting Stress Index–Short Form.

  2. a

    Welch–Satterthwaite corrected. bExact p value.

  3. *p < .05, two-tailed. **p < .01, two-tailed. ***p < .001, two-tailed.

ZAN-BPD
Total Scorea20.31 (4.03)2.00 (1.83)14.92***16.73
Affective Disturbance7.38 (1.56)1.08 (1.5)10.53***24.00
Cognitive Disturbancea4.38 (1.5)0.08 (0.28)10.17***12.82
Impulsivitya3.31 (1.84)0.46 (0.52)5.36***13.98
Disturbed Relationshipsa5.15 (1.46)0.38 (0.65)10.74***16.56
BDI-II Symptom Severitya29.08 (14.83)4.00 (4.26)5.86***13.97
EPDS Symptom Severity15.92 (5.74)3.23 (3.14)7.00***24.00
AUDIT Symptom Severity3.77 (3.86)3.38 (2.36)0.3124.00
History of Substance Abuse, n (%)b5.00 (38.46)0.00χ2 = 6.19*1.00
ECRS
Anxiety96.46 (22.46)44.77 (19.44)6.27***24.00
Avoidancea72.69 (22.28)29.15 (7.87)6.64***19.95
Combined Anxiety/Avoidance Scorea169.15 (21.51)73.92 (19.5)11.82***23.77
DERS
Total Scorea112.08 (31.18)55.54 (10.88)6.17***14.88
Emotional Claritya15.77 (3.85)7.77 (1.59)6.92***15.97
Emotional Awarenessa19.62 (5.19)10.31 (2.98)5.61***19.15
Impulse Controla17.31 (7.66)7.38 (2.40)4.46**14.33
Emotional Nonacceptancea17.62 (8.5)7.85 (1.77)4.06**13.04
Difficulty Engaging in Goal-Directed Behavior18 (5.4)10.69 (3.15)4.22***19.30
Limited Accessa23.77 (8.9)11.54 (4.29)4.46***17.30
CTQ
Total Scorea72 (28.33)32.69 (9.89)4.72***14.88
Physical Abusea12.54 (7.71)6.69 (2.53)2.60*14.55
Physical Neglecta10.38 (5.5)5.23 (0.44)3.37**12.15
Emotional Abusea18 (5)6.92 (3.3)6.67***20.80
Emotional Neglecta16.46 (6.81)7.08 (2.81)4.59***15.97
Sexual Abusea15.38 (8.34)6.77 (3.9)3.37**17.00
PACOTIS
Parental Perceived Self-Efficacya7.24 (1.62)8.83 (1.08)−2.94**20.89
Perceived Parental Impacta6.95 (1.85)9.54 (0.77)−4.66***16.02
Parental Overprotection6.42 (1.95)3.45 (1.6)4.26***24.00
Parental Hostile/Reactive Behavior2.24 (2.21)2.18 (1.87)0.0824.00
PSI-SF
Total Scorea89.31 (30.39)52.46 (11.08)4.11**15.13
Parental Distress37.85 (11.44)19.46 (6.5)5.04***24.00
Difficult Parent–Child Interactiona25.15 (11.76)15.62 (3.78)2.78*14.45
“Difficult Infant” Perceptionsa26.31 (10.66)17.38 (3.97)2.83*15.26

For trauma, mothers with BPD reported significantly more incidents of childhood trauma based on the total CTQ score, p < .001, and on each of the CTQ subscales, as compared to control mothers, with scores falling into the mid to high range (CTQ subscales scores ≥10) for physical abuse, mothers with BPD n = 7 (53.85%) versus healthy control mothers n = 2 (15.38%), p = .02, physical neglect, n = 6 (46.15%) versus n = 0 (0%), p = .01, emotional abuse, n = 13 (100%) versus n = 2 (15.38%), p < .001, emotional neglect, n = 10 (76.92%) versus n = 3 (23.08%), p < .001, and sexual abuse, n = 9 (69.23%) versus n = 2 (15.38%), p = .004. Overall, 12 mothers with BPD reported experiencing mid to high levels of more than one type of trauma during their childhood, as compared with only 2 mothers in the control group.

With respect to parenting measures, mothers with BPD scored significantly higher on parental overprotection, p < .001, total parenting stress, p = .001, and on each subscale of the PSI-SF: Parental Distress, p < .001, Difficult Child, p = .01, and Difficulty in Parent–Child Interaction, p = .01, as compared with healthy control mothers. Mothers with BPD also reported significantly lower levels of perceived self-efficacy as a parent (PACOTIS: p = .01), and perceived parental impact on child emotions and behavior, p < .001, as compared with control mothers. No differences between groups were observed for scores on parental hostility toward their infant, p = .94.

Pearson's correlations were conducted to examine relationships between infant emotion recognition and clinical and parenting variables within the BPD and control groups. No significant correlations were observed between infant emotion recognition and any of the clinical or parenting variables. With respect to clinical and parenting variables, BPD symptom severity was significantly correlated with hostile/reactive behavior toward one's infant, r = .71, p = .01, and parental overprotection, r = –.63, p = .02. Maternal depression was significantly correlated with total parenting stress, r = .7, p = .01, and distress in the parenting role, r = .82, p = .001. Moreover, total emotional dysregulation was significantly correlated with total parenting stress, r = .81, p = .001, distress in the parent role, r = .88, p < .001, parent–child relationship/interaction difficulties, r = .65, p = .02, and maternal perceptions of the infant as being difficult, r = .64, p = .02. However, no significant correlations were observed between parenting disturbance and adult attachment styles or childhood trauma within the BPD group, p > .05. No correlations were observed between clinical and parenting variables for the control group.

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. METHOD
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The present study investigated the ability of mothers with BPD and healthy controls to accurately interpret their infant's emotional displays of happy, sad, and neutral, relative to those of unknown infants, and examined the relationship between infant emotion-recognition ability and key clinical and parenting variables. The findings extend previous research, which has examined emotion perception performance in women with BPD using only novel (unknown) adult-face images, and is the first to examine if emotion perception is influenced by the use of attachment-related stimuli (own vs. unknown infant images).

Infant-Face Emotion-Recognition Accuracy

Mothers with BPD were significantly poorer than control mothers at accurately identifying infant displays of emotion. They also displayed markedly reduced recognition accuracy for neutral infant expressions, demonstrating a strong negative misattribution bias for neutral, mistaking it 84.8% of the time as sad. The finding is consistent with that reported previously for BPD based on adult-face emotion images (Daros et al., 2012; Domes et al., 2008; Dyck et al., 2009; Meyer et al., 2004; Murphy, 2006; Wagner & Linehan, 1999). To our knowledge, this study is the first to demonstrate a negative misattribution bias for neutral infant-face images within the BPD literature. Previous research has suggested that negative misattribution of neutral face stimuli in BPD is associated with chronic emotional, physical, and/or sexual abuse perpetrated in childhood by attachment figures, resulting in a tendency for people with BPD to more negatively interpret others' (including infants') emotions and intentions (e.g., Arntz & Veen, 2001). Certainly, a large proportion of the mothers with BPD in this study reported experiences of childhood abuse and trauma that scored in the mid to high range (92.31%), as compared to mothers in the control group (15.38%). However, as no significant relationship was observed between negative misattribution of neutral faces and indices of childhood trauma in mothers with BPD, this suggests that another factor was involved. An alternative explanation is that maternal reflective functioning is a mediating factor between the experience of childhood trauma and the capacity to objectively identify and accurately interpret infant emotional displays (e.g., Fonagy & Target, 1997). That is, impaired infant-emotion recognition may not be the direct result of experiencing childhood trauma; rather, childhood trauma may affect a mother's reflective capacity and her ability to perceive the emotional and relational cues of her infant. This could be due, potentially, to altered neuronal responses for affective stimuli due to early traumatic experiences. Studies of neuronal functioning among people with BPD have reported atypical activations of the amygdala in particular during presentations of negative scenes and also when viewing affective and neutral facial stimuli (Donegan et al., 2003; Herpertz et al., 2001; Koenigsberg, Fan et al., 2009; Koenigsberg, Siever et al., 2009; Minzenberg, Fan, New, Tang, & Siever, 2007). Although preliminary, collectively the results tentatively demonstrate support for previous research that indicate a negative bias for the perception of emotion in people with BPD, rather than a face emotion processing deficit, per se.

Negative misattribution of neutral infant expressions has been reported in other clinical populations (e.g., depression and anxiety: Gil, Teisssèdre, Chambres, & Droit-Volet, 2011). Studies of mothers with postnatal depression have revealed that they tend to rate negative infant faces as more negative (Stein et al., 2010), and neutral infant faces as more sad and less neutral; further, with increasing maternal anxiety, angry infant faces are more frequently recognized as disgust (Gil et al., 2011). These findings have suggested that maternal mental state and mood influence emotional interaction with an infant and the perception of infant affective communications. Depression is a comorbid symptom in BPD, so measures of depression (BDI and EPNDS) were included in the current study and were used as a covariate in all analyses. However, no association was observed between assessments of depression and infant emotion perception. Instead, negative misattribution was significantly associated with mothers' age and duration of BPD. There are a number of possible explanations for this finding. Older age and longer illness duration may be associated with greater exposure to psychotropic medications, affecting performance; however, as negative misattribution was not associated with medication status, p = .52, or treatment duration, p = .71, this assumption is unlikely. Another possibility is that older women with longer illness duration experience a greater number of adverse social experiences that reinforce expectations of future negative encounters with others, including one's own infant, producing the negative misattribution of neutral infant emotion observed in this study (Arntz & Veen, 2001). But since adverse social experiences were not assessed in this study, this assertion remains speculation. Future research should further investigate this result in treated samples with the equivalent years of active illness, but different maternal ages.

Previous face-perception research has demonstrated that happy facial expressions are more salient and thus easier to identify than are other expressions (e.g., Elfenbein & Ambady, 2002; Kirita & Endo, 1995). In the current study, mothers with BPD and control mothers did not differ from each other on the identification of happy (BPD = 97.7%, controls = 100%) or sad (BPD = 95.4%, controls = 99.25%) infant expressions, suggesting mothers with BPD were not measurably impaired in their perception of, and potential sensitivity to, the happy and sad emotional cues of their infants. This finding is somewhat at odds with the fact that all BPD mothers were engaged in mental health programs specifically targeting atypical parenting. Treatment and engagement in mental health programs may have helped to mediate infant emotion perception performance in this group, potentially normalizing it. More likely though, ceiling effects within the perception tasks and the use of only three infant emotions (happy, sad, and neutral) may have resulted in the infant emotion perception task being insufficiently difficult to differentiate group performance. A more complex task with a greater range of emotions may remedy this.

Several key BPD clinical variables were associated with a number of parenting behaviors. In the BPD group, BPD symptom severity was associated with increased hostile/reactive behavior toward one's infant and parental overprotection. Both emotion dysregulation and maternal depression were associated with higher levels of parenting stress and distress in the parenting role. Emotion dysregulation also was associated with parent–infant relationship difficulties and maternal perceptions of the infant as being difficult. These findings support previous research that has shown mothers with BPD are prone to significant levels of stress in their parenting role and that this may manifest in behavior toward their infant, either by perceiving the infant as difficult to care for or by alternate displays of hostility and/or reactive behavior toward the infant, then overprotectiveness (e.g., Crandell et al., 2003; Hobson et al., 2005; Hobson et al., 2009; Newman et al., 2007; White et al., 2011).

Recognition of Own Versus Unknown Infant Images

Previous research has shown that mothers with secure infant attachment preferentially process the emotional cues of their own, over unknown, infants (Strathearn et al., 2009). An unexpected finding in this study, however, was that control mothers were marginally, p < .05, better at recognizing the emotional displays of unknown infants, relative to that of their own infant. One explanation is that control mothers might have been motivated to spend more time attending to unfamiliar infant images because of the increased processing load associated with novel stimuli. Previous visual scanpath research has supported this view (Loughland et al., 2002a). However, as expected for mothers with BPD, enhanced accuracy was not observed for either type of infant image (own or unknown), suggesting that they similarly processed familiar and unfamiliar (novel) images. Daros et al. (2012) previously reported this finding, but only in relation to adult faces. This is the first study to demonstrate this finding for biologically relevant (i.e., known infant) stimuli.

Limitations

Several limitations are associated with this research. Although the sample size recruited for this study was small, it was still larger than those in many previous studies (e.g., Crandell et al., 2003), demonstrating the difficulties involved in recruiting mother–infant dyads with a diagnosis of BPD due to the nature of the illness, the demands of child rearing, and fear associated with child protection service involvement. However, one strength of the current study was that no participants dropped out between Phases 1 and 2, suggesting that the methodology was well-tolerated. Clearly, though, a larger sample would have provided better power for detecting further group differences. In addition to sample size, a large percentage of the current BPD sample was composed of individuals in active treatment with psychotropic medication. Previous research has found no significant effect of antidepressant (Tranter et al., 2009) or antipsychotic (Hempel, Dekker, van Beveren, Tulen, & Hengeveld, 2010) medication on emotion-recognition performance, and it was not associated with performance in this study. However, as the effects of medication on social neurocognitive functioning are yet to be fully characterized, future studies may benefit from distinguishing groups of patients with BPD on the basis of current psychiatric medication status.

A related issue is that of comorbid substance use and intoxication. The current study relied on self-reported substance dependence and observation of intoxication by a trained mental health professional at the time of participation. However, it is possible that study participants, despite screening, still may have been affected by substances that impacted their performance. Future research might consider a more objective measure of substance use and intoxication by including urine toxicology screening, should funding be available.

BPD also is comobid for depressive disorder (e.g., Zanarini et al., 1998). Individuals with BPD and comorbid depressive symptomatology were included in this study to reflect clients who typically present to clinical services for treatment. While people with a history of Axis I disorders were excluded from the study, self-reported depression based on BDI (Beck et al., 1996) and EPNDS (Cox et al., 1987) measures were included and controlled for in all analyses, and thus did not contribute to emotion recognition performance. Future studies, however, may benefit from the inclusion of a psychiatric control group of people diagnosed with depression to examine issues of clinical specificity. Objective assessments of clinical symptoms and parenting also are required to overcome the limitations associated with self-report measures, particularly in samples such as this where the participants may be anxious about scrutiny from child protective services.

The present study also was limited by not including a measure of maternal reflective functioning, such as the Parent Development Interview (Aber, Slade, Berger, Bresgi, & Kaplan, 1985), to investigate the potential mediating role of reflective functioning between early trauma, emotion recognition, and parenting variables in women with BPD. The current study also had a high-functioning control group in terms of years of education, employment, and socioeconomic and marital status. Although attempts were made to match participants on these variables, difficulties with recruiting mothers with infants made this problematic, and so the variables were controlled for statistically in all analyses. These variables are therefore unlikely to explain group differences in emotion recognition performance.

Last, mothers were asked to verbally identify infant emotions, which preclude any assessment of reaction time. While reaction time was not the focus of this study given the fixed presentation time (i.e., 6 s each) of face images, future studies may want to examine reaction time by including a button response method.

Clinical Implications

The research findings from this study have important implications for clinical practice and targeted remediation for mothers with BPD. The research is novel, identifying a particular subgroup of mothers with BPD who experience difficulty identifying neutral infant emotion displays and tend to misattribute neutral negatively as sad. This is important, as many research and clinical programs focus on delivering interventions targeting recently diagnosed adolescent mothers due to concerns surrounding the ability of these mothers to provide sensitive infant care as a result of a range of psychosocial issues (e.g., competing developmental tasks, low education and income levels, single-parent status) (see Letourneau, Stewart, & Barnfather, 2004). In addition, the compounding impacts of adverse social experiences in older women with a longer duration of BPD potentially reinforce fears of rejection and/or abandonment in this group and are overlooked as a relational risk factor, along with its impact on the sensitivity of maternal responses and implications for an infant's developing emotion regulation and attachment system. No known programs exist that target mothers with chronic BPD. The findings from this study have the potential to aid the development of specific and efficacious remediation interventions focused on improving maternal interpretation and response to infant-face stimuli as well as emotional interaction and reflective capacity in mothers with chronic BPD.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. METHOD
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES