Borderline personality disorder (BPD) is a psychological condition characterized by interpersonal dysfunction and problems, unstable relationships, emotional dysregulation, suicidal behavior, impulsive aggressions such as verbal outbursts, and rapid switches between idealizing and devaluing relationships [American Psychiatric Association (APA), 2000; Lieb et al., 2004]. These dysfunctional thoughts and behaviors may bring patients into conflict with other people, may lead them to experience rejection and abandonment more often than non-patients (Stepp et al., 2009; Renneberg et al., 2011; Lis and Bohus, 2013).

One central feature of the interactional problems of BPD patients is their altered emotion perception. Despite a broad body of studies, findings on this topic are not always consistent. On the one hand, people with BPD are assumed to have negative evaluation styles and beliefs about themselves, others, and the world; and that they make negatively biased assessments, are overly sensitive, and pay more attention to negative stimuli. Especially when exposed to neutral or ambiguous stimuli, patients—compared to healthy controls (HCs)—evaluate others as being more negative, aggressive, or extreme [American Psychiatric Association (APA), 2000; Arntz and Veen, 2001; Sieswerda et al., 2005; Barnow et al., 2009; Domes et al., 2009; Baer et al., 2012; Daros et al., 2014]. For example, people with BPD have been shown to perceive neutral or ambiguous pictures or videos of faces or whole body displays in a negatively distorted way (Dyck et al., 2009; Baer et al., 2012), pay more attention to emotions such as anger but are equally sensitive when recognizing emotions (Domes et al., 2008; Robin et al., 2012; Hagenhoff et al., 2013).

On the other hand and in contrast to findings on deficient emotion recognition, the literature also reveals results where no differences occur and even evidence of better performance of BPD patients on emotion recognition processes (Lynch et al., 2006; Minzenberg et al., 2006; Domes et al., 2008; Merkl et al., 2010; Franzen et al., 2011; Unoka et al., 2011; Robin et al., 2012; Schilling et al., 2012; Schulze et al., 2013; Sieswerda et al., 2013).

In addition to research on emotion evaluation biases within this patient group, several studies have shown that people with BPD recognize emotions presented as pictures or videos of faces or body displays less accurately than controls, and that this accuracy decreases with increasing affect intensity. They recognize negative emotions, particularly anger, disgust, and sadness, less well than controls; show deficits in recognizing neutral facial expressions; and perceive and evaluate the personality traits of film characters as being more extreme (Levine et al., 1997; Veen and Arntz, 2000; Bland et al., 2004; Domes et al., 2009; Dyck et al., 2009; Merkl et al., 2010; Unoka et al., 2011; Arntz and ten Haaf, 2012; Robin et al., 2012; Daros et al., 2013).

In patients with BPD, clinical observations and practice let us presume a lack of confidence regarding recognizing emotion of other people. This often refers to emotions, which are expressed moderately without extreme intensity, facial expressions or gestures. Since patients often had been exposed to the experience of violence and strong negative emotions during their life (Zanarini, 2000; Lieb et al., 2004; Barnow et al., 2005) one may hypothesize that this way of expressing emotions is more familiar to patients than a quieter and more moderate way of expressing emotions. Literature on confidence in emotion perception among patients is not numerous. Schilling et al. (2012) showed that patients were significantly more confident in their decision while performing a theory of mind paradigm wherein they should identify the emotions of facial expressions. Also, patients were more confident compared to controls in a false-memory paradigm in which participants were asked to recognize details from scenes they had watched earlier (Schilling et al., 2013). Due to the small number of reports in this research area, we were interested in whether and to what extend patients would differ in their feelings of confidence during emotion perception.

It has to be noted that most research on emotion perception in patients with BPD has focused on emotions expressed via basic facial expression and prosody. Until now, few studies have used whole body displays or entire movie scenes or film clips showing one person or even two people in an interaction (Domes et al., 2009). However, in everyday interactions, people do not present static facial expressions but rather a complex set of expressions, gestures, and body language. When people perceive their interaction partners, they use multiple modalities to infer their emotional states. Thus, focusing on facial expressions probably neglects an important human system for expressing emotion: the human body and its movements. Human body movement can also convey emotions, and observers can infer the emotional state of an individual or of interacting persons solely from their movements, even when they are far away and their faces of are not clearly visible (de Gelder, 2006). Furthermore, body movements do not only provide information on a possible unspecific threat as facial expressions do. Body movements or postures of anger implicate a more specific physical threat, they bind attention and also they also give a direct cue regarding an adequate behavioral response (Johansson, 1973; Sevdalis and Keller, 2011; Atkinson et al., 2012; Ennis and Egges, 2012; Lorey et al., 2012; Barliya et al., 2013; Kret and de Gelder, 2013; Gelder and Hortensius, 2014). Moreover, social interaction and social context seems to facilitate the emotion recognition process, so that observing two people interacting rather than just one person alone also enhances the validation of perceived emotion and the perception of emotions and confidence in this perception. This emphasizes the relevance of social context for emotion perception (Clarke et al., 2005; Kret and de Gelder, 2010; Lorey et al., 2012).

To investigate the perception of emotional body movements while completely excluding facial information and other distracting variables, this study exploits the advantages of point–light displays (PLDs). These are recordings of the kinematics of a few dots placed on a model’s body. This technique has already been used to detect gender differences in the perception of emotional body language (Alaerts et al., 2011; Sokolov et al., 2011), or for the investigation of differences in emotion perception of various patients groups with somatic and psychological impairments (Pavlova, 2012). Henry et al. (2012), for example, found an impaired processing of socially relevant motions and emotions among patients with dementia. Nackaerts et al. (2012) showed a lower accuracy among persons with autism spectrum disorder whilst recognizing a person’s movements or emotional state and Kaletsch et al. (2014) discovered that patients with major depressive disorder rated emotional body movements more negatively and with higher confidence compared to HCs. Also for patients with schizophrenia deficits in emotion perception of biological and emotional body movements and facial expressions are discussed (Tomlinson et al., 2006; Spencer et al., 2013).

The advantage of using such highly simplified representations is that they provide only kinematic movement information. This ensures that the perception process is not influenced by confounding variables in the stimulus material such as attractiveness, sympathy, and other cultural aspects found in the complex and natural stimuli of faces or whole-body presentations (Hoffmann et al., 2010).

Against this background, we examined the possible differences between patients with BPD and HC. More precisely, we hypothesized that patients with BPD would (a) rate the emotional valence of the depicted interactions more negatively compared to HC, (b) differ in the experienced intensity of depicted emotions, especially when emotional scenes are negative, and (c) differ from controls in how confident they are about their ratings of depicted emotions.

The present study was conducted to investigate differences in the perception of emotional body movements in patients with BPD. Body language is a reliable source of information for assessing emotional states. It can convey emotions and may trigger more complex behavioral responses within the observer than facial expressions might do (de Gelder, 2006). Thereby it plays a significant role in the context of social interacting and functioning which are domains in which many psychiatric patients encounter difficulties. However, until now, no study using PLDs has examined the relationship between BPD and the perception of emotions presented solely through body movements without additional facial expressions.

This study provided one major finding: In comparison to HC, patients with BPD felt less confident when evaluating the depicted emotional scenes. Although not statistically significant, however, due to the exploratory nature of this work it is still noteworthy that patients tended to become less confident the more difficult the videos became to evaluate. Also not statistically significant but still noteworthy was that there might be the effects that firstly, patients with BPD rated monadic but not dyadic scenes more negatively; and secondly, patients rated monadic scenes as being more intense. These indications could be investigated in a subsequent study. To sum up, the present work provides evidence that patients with BPD differ in their feelings of confidence during perception of and response to emotional body movements when they are presented within a PLD paradigm. Less confidence in emotion perception may lead patients to refrain from approaching social situations or interactions or to behave less confidently within such situations. This behavior, in turn, could have adverse effects on such or future interactions.

When examining the first hypothesis that patients with BPD would show a general bias when rating the emotional valence of the depicted interactions, we found no differences between groups. Thus, patients rated the emotions shown just as well as HC. This was also true in regard to the factors difficulty, social context and depicted emotion, which not lead to any significant group differences or interaction effects. Thus, and in contrast to findings that reflect a biased or distorted emotion perception, our findings are similar to those that patients did not show general deficits in emotion recognition or a perception bias, especially toward negative emotions (Minzenberg et al., 2006; Domes et al., 2008; Unoka et al., 2011; Robin et al., 2012; Schilling et al., 2012; Sieswerda et al., 2013). Contrary to previous findings of a negatively biased perception of emotions in patients with BPD (Arntz and Veen, 2001; Barnow et al., 2009; Dyck et al., 2009; Baer et al., 2012) we were unable to detect such a clear effect in this study as that found when applying the same paradigm to a group of patients with major depressive disorder (Kaletsch et al., 2014). Whereas patients with major depressive disorder may tend to perceive emotional interactions with a negative bias on a more general level in line with a negative view of the world, people with BPD might show differential effects not when observing but only when directly involved in an interaction and therefore making the interaction more personally relevant.

Even though it failed to attain significance, the two-way interaction between social context and group indicated that patients with BPD rated monadic scenes but not dyadic scenes more negatively than HC. This effect, however, only indicated a trend and future research could investigate the relationship between aversive feelings when being lonely, empathy, and the number of persons observed.

We also could not confirm our second hypothesis that patients with BPD would perceive emotional interactions as being more intense than HC would. This is in line with research reporting an inability to detect a dichotomous or extremely intense evaluation style in BPD patients (Veen and Arntz, 2000; Sieswerda et al., 2013). Contrary to previous findings of a more extreme evaluation of personality traits we did not detect such an effect in this study as that found when applying the same paradigm to a group of patients with major depressive disorder (Kaletsch et al., 2014). Whereas patients with major depressive disorder may tend to perceive negative emotional interactions more intense, people with BPD might show differential effects not when observing but only when directly involved in an interaction and therefore making the interaction more personally relevant. The non-significant interaction for the intensity of ratings indicated that patients with BPD rated monadic scenes as being more intense, but only when the depicted emotional scene was negative.

The most important finding in the present study is that patients with BPD were significantly less confident when perceiving and evaluating the depicted emotional scenes. This lack of confidence in perceiving emotional body movements may be based on a history of invalidating, instable, and confusing emotional experiences gained with, for example, attachment figures who provided unreliable emotional responses. This insecurity, in turn, could lead to less confidence in evaluating personal interactions (Fonagy, 2000). Such a pattern of emotion perception combined with a high need for self-protection could lead to insecure or less competent behavior within social interactions, to withdrawal, or to avoidance behavior (Newman et al., 2007). In the long term, feelings of insecurity may lead to lower self-confidence and self-efficacy, negative and instable affect, or states of emotional dysregulation (Johnson et al., 2002; Fruzzetti et al., 2005).

Interestingly, even though it failed to attain significance the two-way interaction between difficulty and group indicated that patients with PBD became less confident as the depicted emotion became more difficult to recognize. This could imply that patients are more confident when depicted emotions are strong and expressive, as they often experience them to be for themselves, or people in interactions with patients often might have expressed them, but become less confident, the more subtle and sensitive the expressed emotions are. This might be understood against the background that patients often have a history of violence, abuse and extreme emotions in the past, so that they might be more familiar with this kind of emotion information and therefore might have difficulties in assessing ambiguous, difficult or rather less extreme emotions (Zanarini, 2000; Lieb et al., 2004; Barnow et al., 2005).

Contrary to our results Schilling et al. (2012, 2013) found that patients with BPD showed more confidence than controls when completing the Reading the Mind in the Eyes Test (Baron-Cohen et al., 2001) and during a false memory paradigm whose stimuli differed greatly from those used in the present experiment. However, they assessed confidence in a similar way to our paradigm only in the Reading the Mind in the Eyes Test, and compared their answers only within the highest confidence category between groups, whereas, in our analysis, we compared mean confidence values. Perhaps patients are much less confident when observing a social and emotional interaction compared to a static image. A comparison study could shed light on this issue. Schilling et al. (2012) discussed a possible overconfidence bias or a possible tendency to respond in extreme ways. We could not find evidence for either an overconfidence bias or en extreme response or evaluation style while rating emotional valence or in regard to intensity of ratings.

This is the first study to investigate differences between patients with BPD and HCs when perceiving emotions expressed solely via body movements conveyed by PLDs. First and foremost, we have demonstrated that patients with BPD are less confident when perceiving and evaluating emotional scenes, and that their confidence declines the more difficult the emotional scenes become. Additionally, unlike patients with major depressive disorder (Kaletsch et al., 2014), BDP patients do not reveal an altered emotion perception; that is, any biased perception in either a negative or a positive direction. However, they do show a tendency to evaluate monadic stimuli more negatively. Further, in general, BDP patients do not perceive and evaluate depicted emotions more extremely in the sense of a dichotomous thinking but tend to evaluate monadic scenes as being more intense when the depicted emotion is negative.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.