BPD is characterized by a pervasive pattern of emotional instability that is directly related to many aspects of patients’ psychopathology, e.g., self-injury, interpersonal disturbance, and identity disturbance. Given the pivotal role of emotional instability in BPD symptomatology, this endophenotype is a major target for current treatment approaches, which aim to provide patients with techniques supporting the identification and regulation of negative emotional states. These include mentalization-based therapy (Bateman and Fonagy 2008) and dialectical behavior therapy (for a meta-analysis see Kliem et al. 2010). Emotional instability is theorized to be the result of two factors: (a) an enhanced emotional reactivity and (b) an inability to control intense negative emotional states (Koenigsberg 2010; Minzenberg et al. 2008; Posner et al. 2003). The review of available neuroimaging findings will be accordingly subdivided, although one should keep in mind that the line between emotion processing and emotion regulation is blurry at best as the generation of emotional responses is already subject to (implicit) modulations by regulatory-oriented processes (Gross 1998, 2002).

In accordance with two-factor models, BPD patients described more intense responses to emotionally evocative stimuli and exhibited greater instability of emotions in both laboratory and naturalistic settings (Ebner-Priemer et al. 2007; Herpertz et al. 1997; Rosenthal et al. 2008; Stiglmayr et al. 2005). Experimental studies provided further support for impaired inhibition of negative stimuli (Domes et al. 2006; Silbersweig et al. 2007), deficits in the disengagement of attentional resources from negative facial expression (von Ceumern-Lindenstjerna et al. 2010), and a hypervigilance to emotional cues (Arntz et al. 2000), especially in the presence of BPD-specific cues (Sieswerda et al. 2007).

At the neural level, most studies find support for abnormalities in the emotion processing circuitry of patients with BPD (for a detailed description of brain regions underlying emotion processing, see, e.g., Kober et al. 2008; Murphy et al. 2003; Tsuchiya and Adolphs 2007). An initial study by Herpertz et al. (2001), for instance, presented patients with aversive social scenes and found a link between patients’ experience of intense emotions and enhanced activations of the bilateral amygdala. Stronger neural activity was also found in the bilateral fusiform gyrus, which might reflect an increased vigilance for negative emotional stimuli, as the visual cortex has dense connections to the amygdala (Amaral et al. 2003). Subsequent work found enhanced reactivity of the left amygdala in BPD in response to pictures of facial expressions of emotions, regardless of their specific valence (Donegan et al. 2003), whereas another study demonstrated a more specific pattern of limbic hyperarousal. In this study, enhanced activity of the right amygdala in BPD was only observed in response to fearful compared to neutral faces, whereas attenuated activation of the bilateral amygdala was found during the presentation of angry facial expressions (Minzenberg et al. 2007).

Overall, available evidence in BPD is consistent with the conclusion that limbic and paralimbic hyperreactivity holds particularly for negative and neutral stimuli (e.g., Jacob et al. 2012; Niedtfeld et al. 2010; Schulze et al. 2011), whereas the existing results regarding the processing of positive stimuli are to date still inconclusive. Koenigsberg and colleagues (2009b) found that compared to healthy controls, BPD patients exhibited heightened activity of the limbic system only in response to negative stimuli; they failed to find group differences for positive images. They also observed greater activation of the dorsolateral prefrontal cortex during the presentation of negative stimuli in healthy controls. In contrast, Hazlett et al. (2012), as well as Donegan et al. (2003), found greater amygdala activity in BPD patients across all picture types, including positive stimuli. Additionally, recent findings suggest abnormalities in emotion processing might affect the functioning of the reward circuitry in BPD (i.e., the ventral striatum, the pregenual anterior cingulate, and the ventromedial prefrontal cortex) and the functioning of the cortical midline structures during the anticipation of reward (Enzi et al. 2013).

Although most studies primarily investigated the processing of facial expressions or social scenes, heightened negative emotionality in patients with BPD has been demonstrated across a wide range of different stimuli. For instance, Jacob et al. (2012) reported stronger left-lateralized activation of the amygdala, along with attenuated activation of the subgenual anterior cingulate during an auditory anger induction procedure. Similarly, BPD patients showed stronger activation of the amygdala and insular region while recalling unresolved autobiographic life events (Beblo et al. 2006) or reading negatively valenced words (Silbersweig et al. 2007). Exaggerated and temporally prolonged amygdala responses were also observed in BPD patients during a procedure in which patients were presented with simple squares paired with painful electrodermal stimulation while being scanned (Kamphausen et al. 2013).

However, as outlined above, heightened emotional response represents only one facet of emotional instability in BPD. Impairment in the cognitive control of intense emotional states is argued to be equally important for understanding the disturbance patients experience in their daily lives, as the ability to regulate emotions has important effects on physical and mental health (Davidson 2000; Eftekhari et al. 2009; Gross and John 2003). As a consequence, studies of neural mechanisms associated with emotional regulation have rapidly increased in the last years. These studies have examined several different forms of cognitive control, ranging from attentional control processes, such as (in)voluntary shifts of attentional focus or limited allocation of resources necessary for the processing of external stimuli, to cognitive change via reappraisal (for reviews, see Ochsner and Gross 2005; Ochsner et al. 2012). The impact of cognitive reappraisal on neural activity in the limbic and paralimbic regions is observed across a complex prefrontal network involving multiple neural regions; however, the impact is particularly observable in the dorso- and ventrolateral, orbitofrontal, and anterior cingulate regions (e.g., Banks et al. 2007; Domes et al. 2010; Johnstone et al. 2007; Ochsner et al. 2004; Wager et al. 2008). The variability of prefrontal localizations associated with cognitive reappraisal might be attributable to temporal aspects of the reappraisal process, which first involves implementation and subsequently involves maintenance and monitoring of the effects of reappraisal on emotional responses (Kalisch 2009).

Recent studies of BPD patients investigated the effects of two different strategies of cognitive reappraisal: psychological distancing and reinterpretation of emotional situations. In one of the first neuroimaging studies of this phenomenon conducted by Koenigsberg et al. (2009a), patients were asked to decrease their emotional response to negative visual scenes using psychological distancing strategies. Compared to healthy controls, BPD patients showed difficulties engaging the dorsal anterior cingulate cortex and inferior parietal lobe during conditions of psychological distancing while showing a paradoxical increase of neural activity in the right amygdala. In another study conducted by Schulze et al. (2011), a delayed reappraisal paradigm was used to more clearly distinguish between the initial emotional reactivity to a negative stimulus and the subsequent modulation of emotional responses by cognitive reappraisal (Jackson et al. 2000). Statistical analyses during the initial viewing phase replicated previous results of limbic and paralimbic hyperreactivity during the presentation of negative and neutral stimuli in BPD. With regard to the regulatory phase, BPD patients showed attenuated activations of the left (more lateral) orbitofrontal cortex, along with attenuated reduction of activity in the bilateral insula during attempts to voluntarily decrease their emotional reactions. The orbitofrontal cortex is thought to play an important role in successful reappraisal via altering and updating the context-sensitive relevance of stimuli (Ochsner et al. 2004; Rolls 2000), suggesting specific deficits in BPD patients’ ability to implement new appraisals for the presented stimuli. Overall, the results by Schulze et al. (2011) support two-factor models of emotional instability in BPD, i.e., enhanced emotional responding and deficits in cognitive regulation of emotions. Interestingly, neurofunctional abnormalities in regions associated with the processing or regulation of emotions were only prominent when participants were instructed to attenuate negative emotions, whereas no group differences were found when participants attempted to increase their emotions. Additionally, recent work aimed to disentangle the specific effects of BPD from those of traumatization on the neural correlates of cognitive reappraisal (Lang et al. 2012). In this study, healthy participants without any history of traumatization showed an initial increase in prefrontal activation (e.g., dorsolateral and dorsomedial prefrontal regions, anterior cingulate cortex), whereas trauma-exposed BPD patients and a comparison group of healthy individuals with traumatic experiences showed initial deactivations in those regions. Intriguingly, the group differences in the initial phase of emotion regulation might also indicate specific impairments in the implementation and generation of alternative appraisals (Kalisch 2009). In addition, the results highlight the necessity to more adequately address individual differences in BPD, e.g., trauma exposure, PTSD, or treatment experience.

Several studies examined attentional control (i.e., behavioral inhibition) of BPD patients in the context of negative stimuli (e.g., Domes et al. 2006; Fertuck et al. 2006; Rentrop et al. 2008). At the neural level, deficits in the inhibition of behavioral responses in a linguistic Go/No-Go task, for instance, were related to relatively decreased activity in the ventromedial prefrontal cortex (including orbitofrontal and subgenual anterior cingulate regions) compared to healthy controls (Silbersweig et al. 2007). Attenuated activations of the anterior cingulate cortex during response inhibition were also observed in an emotional Stroop paradigm (Wingenfeld et al. 2009). While these studies assessed inhibition and negative emotions in general, recent work also addressed how performance on an impulse control task is modulated by more specific emotional states (Jacob et al. 2012). Listening to an anger-inducing story, BPD patients showed a stronger activation of the right amygdala and the nucleus subthalamicus, whereas healthy controls showed a stronger recruitment of the subgenual anterior cingulate cortex. The subsequent performance on a Go/No-Go task was accompanied by attenuated activation of the left inferior frontal cortex in BPD.

Finally, two studies investigated the influence of emotional distraction on working memory performance. In both studies, BPD patients showed longer reaction times during emotional distraction along with enhanced activations of the amygdala compared to healthy controls (Krause-Utz et al. 2012; Prehn et al. 2013b). Moreover, limbic and paralimbic activity was negatively correlated with current states of dissociation, suggesting that dissociation has a dampening effect on emotional reactivity (Krause-Utz et al. 2012). Dissociation is considered a regulatory strategy to cope with overwhelming emotions, presumably via prefrontal inhibition of limbic regions (Lanius et al. 2010; Sierra and Berrios 1998). While dissociation represents a highly efficient way to shut down the emotional system (Ebner-Priemer et al. 2005, 2009), such states also prevent patients from learning to implement new strategies in emotionally arousing situations and represent a major challenge for psychotherapeutic treatment of BPD (Kleindienst et al. 2011).

The study by Prehn et al. (2013b) further assessed the role of working memory load on emotion processing. Despite heightened distractibility by highly negative stimuli in general, healthy controls and criminal offenders with BPD showed equivalent declines of limbic activity associated with manipulations affecting working memory load. In other words, the engagement in a secondary cognitively challenging task downregulates amygdala activity, presumably because these tasks draw resources necessary for the processing of emotional stimuli (see also Kanske et al. 2011; Van Dillen et al. 2009). Hence, attentional deployment strategies might be particularly useful for attenuating emotional reactions in BPD.

To summarize, findings of functional neuroimaging studies support two-factor conceptualizations of emotional instability by illustrating abnormalities in the processing and regulation of emotions. Comparing negative emotionality between patients with BPD and healthy controls, enhanced activation of the amygdala and insular region, along with attenuated activation of the rostral part of the anterior cingulate cortex, was consistently reported. As stated initially, the boundary between processing and regulation of emotions is blurry at best (Gross 1998, 2002), and quite congruently, self-reports of emotional regulation capabilities correlate positively with neural activity in limbic and paralimbic regions (Niedtfeld et al. 2010; Schulze et al. 2011). Preliminary results also suggest dialectical behavior therapy—focusing particularly on improving emotion regulation—attenuates neural activity of limbic regions in response to emotionally arousing stimuli (Schnell and Herpertz 2007), providing a promising starting point for assessing the effects of specific treatment modules on affective hyperarousal in BPD. With regard to the cognitive control of emotions, BPD seems to be characterized by an attenuated functioning of the dorsal part of anterior cingulate cortex and medial and lateral parts of the orbitofrontal cortex and deficient activation of the dorsolateral prefrontal cortex. Therefore, dysfunctional behaviors, such as self-injury or dissociation, might be best understood as an attempt to initiate inhibitory control of intense emotions via increased activation of the prefrontal regions, e.g., dorsolateral prefrontal cortex (Krause-Utz et al. 2012; Niedtfeld et al. 2012; see also next paragraph).

Deliberate self-injurious behavior, such as cutting or burning, is one of the most prominent clinical symptoms observed in BPD patients, with rates as high as 90 % observed in some studies (Skodol et al. 2002; Zanarini et al. 2008). Patients frequently state that they use non-suicidal self-injurious behavior to escape from undesired or extreme negative emotions (Chapman et al. 2006; Kleindienst et al. 2008). They also experience an immediate relief of tension, a decrease of dissociative symptoms, and elevations of mood (Herpertz 1995; Kemperman et al. 1997; Reitz et al. 2012). Self-injury was therefore proposed to represent a dysfunctional attempt to regulate negative emotional states directly corresponding to emotional instability in BPD (Klonsky 2007; Welch et al. 2008).

Patients with BPD display a reduced sensitivity to painful stimulation across different nociceptive modalities (e.g., Cardenas-Morales et al. 2011; Magerl et al. 2012; Russ et al. 1992; Schmahl et al. 2004), which is further decreased under conditions of elevated stress or states of dissociation (Bohus et al. 2000; Ludascher et al. 2007). Conversely, discontinuation of intentional self-injurious behavior results in a normalization of pain perception (Ludascher et al. 2009). The reduced sensitivity to pain, however, is not attributable to impairments in the sensory-discriminative component of pain processing, as BPD patients did not exhibit abnormal performance in the spatial discrimination of nociceptive stimuli or in the detection of painful stimulation (Ludascher et al. 2007; Schmahl et al. 2004). Therefore, BPD patients’ reduced sensitivity might be better explained by abnormalities in the affective-motivational component of pain sensation, which in turn can be modulated by the context and cognitive appraisals of pain. Importantly, recent work by Magerl et al. (2012) found that endogenous antinociception in BPD is best predicted by the recency of self-injurious behavior, whereas psychometric measures of BPD psychopathology were unrelated to pain sensitivity. Thus, the authors suggested reduced pain sensitivity might not be a consequence of BPD psychopathology per se, but rather the result of neurofunctional rearrangement of brain circuits involved in the processing of pain due to BPD-related self-injurious behavior.

Neural models of pain sensation identified two anatomically distinct neural pathways that might be implicated in BPD patients’ self-injurious behavior—sensory discriminative and affective motivational (for a review, see Treede et al. 1999). The sensory-discriminative system, necessary for stimulus localization and perception of pain quality and intensity, projects from the lateral thalamic nuclei to the primary and secondary somatosensory cortices. In contrast, the affective-motivational pathway, which underlies evaluation of pain and subsequent emotional or behavioral reactions, projects from the medial thalamic nuclei to the anterior cingulate cortex and insula, which is in turn reciprocally connected to the brainstem and the amygdala (Augustine 1996; Barbas et al. 2003).

Neurobiological correlates of altered pain processing in BPD were initially investigated by Schmahl and colleagues (2006). In this study, heat stimuli of two different intensities were delivered via a thermode. Participants either received a standardized temperature of 43 °C or an individual temperature adjusted for equal subjective pain in all individuals. Here, BPD patients again showed much higher pain thresholds than healthy controls. In response to individually adjusted heat stimuli, patients showed neural deactivation in the perigenual anterior cingulate cortex and the amygdala, while greater BOLD responses were found in the dorsolateral prefrontal cortex. Therefore, antinociceptive mechanisms might be attributable to increased top-down control closely associated with deactivations of the limbic system, particularly in patients with comorbid posttraumatic stress disorder (Kraus et al. 2009). Subsequent work combined thermal sensory stimulation with the prior induction of negative affect to further elucidate the potential role of pain with respect to regulation of affective states in BPD (Niedtfeld et al. 2010). In line with previous work, the BPD sample showed enhanced activations of the amygdala, insula, and anterior cingulate cortex in response to negative (but also to neutral) stimuli. The subsequent administration of heat stimulation again served to suppress neural activation of the amygdala and anterior cingulate cortex. Interestingly, there was only limited support for a BPD-specific role of pain in affect regulation, as both groups showed comparably attenuated activity in the amygdala. As a result, the authors proposed a general mechanism to explain the soothing effects of pain. In particular, they argued that painful sensory stimulation might draw resources from the processing of negative emotions, similar to cognitive strategies of “attentional deployment” (Gross 2002; Ochsner and Gross 2005). Additional connectivity analyses indicated that inhibitory influence on limbic regions during painful stimulation is particularly attributable to activation of the medial and dorsolateral parts of the prefrontal cortex (Niedtfeld et al. 2012).

In a different study, participants were presented with an audiotape describing an act of self-injury comprising different aspects of the situation, such as the respective trigger, related cognitive and emotional reactions, the act of self-harming behavior, and the subsequent relaxation (Kraus et al. 2010). While listening to emotional reactions, BPD patients displayed attenuated activation of the orbitofrontal cortex in comparison to healthy controls, presumably due to impaired ability to inhibit or modulate the elicited emotions. The description of the self-injurious act itself was again accompanied by decreased activity of the mid-cingulate cortex in BPD.

Alternative accounts of self-injury highlight the importance of the endogenous opioid system (for reviews, see Bandelow et al. 2010; New and Stanley 2010; Stanley and Siever 2010). The human opioid system consists of β-endorphin, met-enkephalin, and dynorphin, with all three neurotransmitters broadly involved in antinociception and stress-induced analgesic states (Fields 2004; Flor et al. 2002). More specifically, it is suggested that several forms of dysfunctional behavior in BPD (e.g., self-injury, risk-seeking behavior, promiscuity, substance abuse) might reflect attempts to raise relatively low levels of opioids (Bandelow et al. 2010; New and Stanley 2010). Reduced availability of dynorphin might also explain feelings of chronic emptiness and dysphoria (Land et al. 2008; for a discussion, see Stanley and Siever 2010). An initial investigation into this phenomenon found lower levels of cerebrospinal fluid β-endorphins and met-enkephalins in individuals with a history of self-injury, mainly patients with BPD (Stanley et al. 2010). These findings were paralleled by a PET study measuring μ-opioid receptor availability during neutral and dysphoric states (Prossin et al. 2010). At baseline, greater μ-opioid receptor availability in BPD was found in the orbitofrontal cortex, caudate nucleus, nucleus accumbens, and amygdala, presumably reflecting lower levels of endogenous opioids. The induction of sad affect led to a greater activation of the opioid system in the anterior cingulate cortex, orbitofrontal cortex, ventral pallidum, and amygdala, indicating a compensatory response to basal levels of μ-opioid in BPD.

To summarize, theoretical conceptualizations propose that self-injury is a dysfunctional, but highly efficient strategy for the regulation of negative emotional states. Higher pain thresholds in BPD are not attributable to impairment in the sensory-discriminative component of pain processing, but rather to abnormalities in affective-motivational components. The results of imaging studies further bolster such conceptualizations. In response to pain, patients with BPD show enhanced activation of the prefrontal cortex, along with attenuated activation of the anterior cingulate cortex and limbic system, whereas no functional abnormalities are found in somatosensory cortices. Therefore, the soothing effects of self-harm might be best explained by an increased top-down control of limbic regions through medial and dorsolateral prefrontal regions. Further emphasis must be placed on understanding associations between self-injury and the opioid system. Given that recent findings indicate that BPD patients exhibit baseline deficits in endogenous opioids, this area might hold some promise as a target for pharmacotherapy in BPD.

Disturbed relatedness was acknowledged in earliest descriptions of BPD (Kernberg 1967), and patients’ difficulties in interpersonal contexts are currently considered the most unique and discriminant feature of BPD psychopathology (Gunderson et al. 1995). Up until the present, studies in this area have focused predominantly on abnormalities in the perception and recognition of others’ mental and emotional states as a social-cognitive factor contributing to disturbance in BPD patients’ social interactions (for a review of social cognition in BPD, see Roepke et al. 2012). Consistent with this notion, most behavioral studies found that BPD patients exhibit impairment in basic emotion recognition (e.g., Bland et al. 2004; Levine et al. 1997; Merkl et al. 2010; Ritter et al. 2011; Unoka et al. 2011) as well as a bias to interpret ambiguous stimuli more negatively (Domes et al. 2008; Meyer et al. 2004; Wagner and Linehan 1999), although a number of studies also reported superior attribution and detection of mental states in BPD (Fertuck et al. 2009; Lynch et al. 2006; Schulze et al. 2013a). Deficits in the recognition of emotions were particularly prominent in more complex and naturalistic settings (for a discussion of the role of ecological validity, see Roepke et al. 2012), e.g., when subjects have to integrate facial and prosodic information (Minzenberg et al. 2006), are required to rapidly discriminate facial expressions (Dyck et al. 2009), or have to judge video-based social interactions (Preissler et al. 2010; Sharp et al. 2011). In sum, these studies point predominantly toward impairments in BPD patients’ capacity to infer the emotional states of other individuals.

However, cognitive empathy represents only one facet of social cognition (Decety and Meyer 2008; Singer 2006); a second component represents the emotional response to the observed emotional state of another person, i.e., emotional empathy (Eisenberg and Miller 1987; Mehrabian and Epstein 1972). The few experimental studies available on the subject found that BPD patients exhibit a reduced capacity to experience empathic concern for other people’s observed distress (Dziobek et al. 2011; Ritter et al. 2011). In combination with aforementioned findings of deficits in cognitive empathy, these findings suggest that misperceptions of others’ emotional states might lead to dysfunctional emotional responses in social interactions (New et al. 2008).

Consequently, recent neuroimaging studies have aimed to identify functional brain abnormalities associated with impaired cognitive empathy in BPD. An initial study by Guitart-Masip et al. (2009) investigated patients’ brain responses while discriminating emotional facial expressions from neutral expressions. Compared to healthy controls, BPD patients made more mistakes in discriminating between negative and neutral facial expressions. More specifically, patients less accurately discriminated between disgusted and neutral expressions and showed a trend for similar difficulties in distinguishing between fearful and neutral faces. These behavioral impairments in BPD were accompanied by greater activation of the left inferior and middle temporal cortex (including parts of the superior temporal gyrus), possibly reflecting abnormalities in the perceptual processing of visual stimuli.

Subsequently, Dziobek et al. (2011) used a more ecologically valid paradigm to measure functional neural correlates of social cognition in BPD. The Multifaceted Empathy Test (MET; Dziobek et al. 2008) depicts people in emotionally charged situations and allows a separate assessment of cognitive and emotional aspects of state empathy. Participants are required to either infer the mental states of the depicted person (cognitive empathy) or to rate their level of empathic concern for the displayed individual (emotional empathy). Behaviorally, patients showed impairments in cognitive and emotional aspects of empathy compared to healthy controls. At the neural level, both groups activated a similar neural network during empathic processes, which comprised the temporal pole, the temporoparietal junction, the orbitofrontal cortex, and the superior temporal gyrus for cognitive empathy (for a review, see Frith and Frith 2005) and, in addition, the insula and medial prefrontal cortex for emotional facets of empathy. Subsequent group comparisons revealed significantly reduced activation of the left superior temporal sulcus and gyrus of BPD patients while inferring others’ mental states. Moreover, this reduction was related to symptoms of intrusion in the BPD group. In contrast, enhanced activation of the right mid-insula was found when patients were emotionally attuned to another person. This brain abnormality was additionally associated with emotional arousal, as assessed by skin conductance responses.

Abnormalities in the left superior temporal sulcus of BPD patients were also found by Mier et al. and by Frick et al. (2012). In the former study, neural responses were measured while participants performed basic or more complex social-cognitive tasks. Healthy controls showed stronger activation of the superior temporal sulcus in response to greater demands on social-cognitive processes; such a pattern, however, was not seen in BPD patients. Hypo-activation of this brain region was consequently particularly prominent when patients performed complex social-cognitive tasks, i.e., attributing others’ intentions. Furthermore, the authors found that BPD patients exhibited hyperresponsiveness of the limbic system independent of the specific social-cognitive process investigated. In line with these findings, Frick and colleagues (2012) also observed attenuated activation of the superior temporal sulcus and enhanced activation of the right amygdala during a mental state discrimination task (based on the eye region of individuals only). In this study, patients with BPD were more accurate in their ability to discern others’ intentions than nonclinical controls (see also Scott et al. 2011).

Taken together, available studies illustrate abnormalities in BPD patients’ ability to infer others’ mental and emotional states, with traumatization being a negative predictor of social-cognitive abilities in BPD (Dyck et al. 2009; Preissler et al. 2010). Neuroimaging studies of mentalization found consistently altered functioning of the superior temporal sulcus in BPD. The superior temporal sulcus might be particularly vulnerable to the impact of traumatic experiences, as this brain region matures rather late in adolescence (Paus 2005). Alternative accounts propose that emotional hyperreactivity in concert with deficient functioning of the prefrontal cortex might interfere with social-cognitive processes (for a discussion, see Domes et al. 2009). The data provided by Dziobek et al. (2011) support the hypothesis that emotional arousal interferes with empathic processes in BPD. Thus, abnormalities related to the processing and regulation of emotions might be linked to altered social-cognitive processes in BPD patients.

As illustrated above, up until now, studies investigating social cognition in BPD primarily focused on the ability of BPD patients to accurately recognize others’ emotions and intentions, i.e., cognitive empathy. Such abilities, however, only represent a small part of successful interpersonal interactions, which are characterized by an ongoing exchange of social signals between individuals (cf. Dziobek 2012). For that reason, recent studies adopted multiplayer economic exchange games to model interactional behavior in BPD (see also Seres et al. 2009). In multi-round trust games without any feedback about the back transfer of the other partner, BPD patients transferred smaller amounts of monetary units compared to nonclinical controls or patients with major depression (Unoka et al. 2009). Impaired trust during these sequential interactions was also related to individual difficulties in interpersonal relationships.

Despite the prominence of interpersonal disturbances in BPD, there is to date only one study available that examined the underlying neurofunctional correlates. King-Casas and colleagues (2008) used a multi-round economic exchange game with functional neuroimaging to investigate interactional behavior in patients with BPD. Their findings suggested that interpersonal disturbances in BPD are related to difficulties in maintaining cooperation with a healthy partner as well as an impaired capacity to restore broken cooperation. Neurologically, abnormal activation patterns of the bilateral anterior insula differentiated healthy controls from individuals with BPD. While activation of the insula in healthy controls tracked changes in cooperative behavior, in patients with BPD, activity of the anterior insula was unrelated to the magnitude of offers received (input) in BPD. Importantly, no insular abnormalities were found to be related to repayment of partners (output). Thus, the attenuated activity observed in the anterior insula during input led the authors to propose that “anomalous [perception of] social ‘input’ … (rather than ‘output’) is a potential mechanism of interpersonal dysfunction” (King-Casas and Chiu 2012; p. 121), particularly with regard to disturbed perception of social norms and an inability to recognize cues signaling waning trust of partners.