Borderline personality disorder is a complex phenotype and etiological trajectories leading to BPD are invariably characterized by equifinality and multifinality (Beauchaine et al. 2009
). However, affective dysregulation is a core feature of BPD, with one study finding that 90% with the diagnosis meet this criterion (Zanarini et al. 2004
). Moreover, affective instability is the diagnostic feature associated most strongly with treatment utilization among those with BPD (Bagge et al. 2005
). For these reasons, trajectories leading to dysregulated emotion likely overlap with those that lead to BPD (Lieb et al. 2004
). Social baseline theory couches emotion dysregulation within an interpersonal context. This is consistent with findings that the developing frontolimbic system is affected by environmental inputs, such as the availability of caregivers to skillfully co-regulate a child’s emotional distress (e.g., Belsky and de Haan 2011
; Roth and Sweatt 2011
; see also Mead et al. 2010
for a review).
Current understanding of borderline personality development could be enriched by incorporating some of the key assumptions of SBT. First
, the early attachment relationship serves as the initial source of co-regulation. Filial bonding typically occurs quickly and unconditionally during a period of rapid neural development. During this time, neural links are formed between the PFC and structures that underlie emotion and memory (e.g., amygdala, nucleus accumbens, hippocampus; Coan 2008
; Hofer 2006
). This may contribute to early variability in assumptions about the social world and the perceived likelihood of experiencing co-regulation during emotional situations.
, these and other neural structures likely support attachment and maintenance of friendship bonds in childhood and adolescence (Porges et al. 1996
; Rockhill et al. 2009
; Shields et al. 1994
; Shipman and Zeman 2001
; Snyder et al. 1997
). Ideally, effective co-regulation will have occurred earlier in life and, consequently the neural structures implicated in self-control will have already begun to form and strengthen. Children with better self-control are more likely to be accepted by peers (Shields et al. 2008
) and may be less likely to use problematic strategies for obtaining emotional support (Crick et al. 2005
), which could further alienate potential sources of co-regulation (e.g., relational or physical aggression, self-injury, substance use).
, frontolimbic circuits are also implicated in adult attachment formation, trust, affiliation, and attraction (Fertuck et al. 2006
). By adulthood, people who have low expectations for co-regulation may be more likely to chronically deplete PFC resources by using independent self-regulation as the baseline strategy. Unfortunately, such chronic self-reliance often leads to “regulatory failure” (Gailliot and Baumeister 2007
), which can potentiate ineffective and/or impulsive emotion regulation strategies. Differences in frontolimbic circuits are found consistently in fMRI studies of BPD (Bohus et al. 2004
; Domes et al. 2009
In this review, we examine each of the three assumptions of SBT and how insufficient co-regulation across the life span may lead to behavioral and interpersonal difficulties, such as those seen in BPD. Social baseline theory integrates attachment theory with an emerging literature on neural depletion following independent self-regulation (Coan 2008
). Most attachment researchers propose a developmental trajectory of increasing behavioral control such that, by adulthood, nearly all self-regulation is independent (e.g., Mikulincer and Shaver 2008
). In contrast, SBT suggests that the process by which early attachment leads to adaptive adult outcomes is via healthy expectations of continued co-regulation in adulthood. The inability to identify, utilize, and maintain these social supports may contribute to emotional fatigue, dysregulated behaviors, and differences in frontolimbic circutry observed among adults with BPD (Bohus et al. 2004
; Domes et al. 2009
). From this perspective, borderline pathology can be understood not only as a disorder of emotion dysregulation but also one of insufficient co-regulation across the life span.
Biological Mechanisms of Early Attachment
Attachment figures serve as the initial source of social affect regulation and load sharing. Thus, these early relationships appear to have a lasting effect on attachment style and emotion regulation (Coan 2008
; Diamond and Hicks 2005
; Meany 200; Schore 1996
). Animal research with rat pups and their mothers has found that maternal care affects DNA methylation, altering future cortisol regulation and temperament of the pups (Meany 2001
). For humans, the caregiver-child relationship is the first experience whereby the child learns to influence the caregiver’s emotions and behaviors. Similarly, the child experiences self-regulation of behavior and emotions through the caregiver’s actions (Cole et al. 2004
). Researchers find that compared with adults, children recruit greater regions of the PFC during independent affect regulation tasks (Levesque et al. 2004
). This is likely due to immature emotion regulation skills and the considerable neural resources required for self-control in the absence of parental support. Thus, the attachment relationship appears to offset the extremely high cost of independent emotional and behavioral control.
Early attachment status also predicts emotion regulation abilities and attachment style in adulthood (Diamond and Hicks 2005
; Diamond et al. 2006
, Lyons-Ruth 2008
). Thus, co-regulation not only conserves resources, but may also teach independent self-regulation and strategies for utilizing other individuals as regulatory resources; processes that are likely subserved by increased axonal connectivity within frontolimbic substrates (Posner and Rothbart 2007
). Functional connectivity between the dorsomedial PFC, amygdala, and the hippocampus may also be one mechanism by which early relationships become an “internal working model” of attachment (Hofer 2006
). In other words, the developing frontolimbic system is sensitive to social inputs and structurally encodes expectations of distress alleviation and security provision from attachment figures (Coan 2010
). Across development, the amygdala tags emotional stimuli while the hippocampus consolidates the associated contextual cues into long-term memory. Through this process, the behavior of attachment figures becomes stored as neural representation. The amygdala is also sensitive to signs of threat and, via input to the hypothalamus, functions to regulate stress hormones and facilitate social soothing (Kemeny 2003
; McEwan 2007
Reciprocal projections between the PFC and the amygdala, hippocampus, and hypothalamus contribute to memory formation and conditioned learning, including the appraisal of emotional stimuli and activating the appropriate motivated behavior (Davidson and Irwin 1999
; Rolls 2007
). Through this process, conditioned responses to attachment figures are encoded within medial, orbital, and dorsolateral circuits of the PFC, serving as markers of threat or protection. These associations strengthen (i.e., become internal working models) through dopaminergically mediated experiences of security (Coan 2010
). Similarly, oxytocinergic activity in the hypothalamus, nucleus accumbens, ventral tegmentum, and amygdala appear to influence attachment security (see Coan 2008
). Researchers also find that increased endogenous opioid activity in the ACC underlies sensitivity to and greater distress from social rejection, which may be yet another mechanism by which attachment experiences are encoded within frontolimbic circuitry (Eisenberger et al. 2007
; Way et al. 2009
). According to SBT, the development and reinforcement of these circuits likely shape interpersonal relatedness across the life span.
Middle Childhood and Adolescence
Throughout childhood, the regulatory effects of the child-caregiver bond occur both in the presence of the attachment figure and through the mental representation of caregiver availability in response to threat. Under optimal conditions, the child and caregiver form a secure attachment. This leads to reasonable assumptions of co-regulation and a confident sense of “the self” in relation to key attachment figures. Developmental psychopathologists conceptualize this as an “average expectable environment” (Cicchetti and Valentino 2006
). Such environments include a range of conditions that promote normative developmental processes. Differences within this range serve as opportunities for individual development and promote variability in the phenotypic expression of genes. In some cases, however, there is a “failure” of the average expectable environment, as in cases of neglect or abuse. Under such circumstances, normal development is threatened.
A large number of adults with BPD report a history of neglect or abuse (e.g., Goldman et al. 1992
; Soloff et al. 2002
; Zanarini et al. 1989
), although it is now clear that abuse is not a necessary antecedent to later borderline pathology (see Goodman et al. 2004
). However, research on children with abuse histories can inform our understanding of BPD because such youth also manifest poor interpersonal relatedness and deficits of emotional and behavioral control (Shields et al. 1994
Maltreated children are more likely than non-abused children to engage in antisocial, aggressive, withdrawn, and disruptive behaviors during play interactions, even when children are matched on key demographic variables (see Cicchetti et al. 1992
; Shields et al. 1994
). Moreover, children who display antisocial or aggressive behaviors are subsequently viewed as mean or attention seeking and tend to be less highly regarded by their peers (Asher and Coie 1990
; Denham and Holt 1993
; Salzinger et al. 1993
). This can lead to rejection and enduring negative reputations within social groups (Shields et al. 1994
). In turn, peer rejection produces functional changes in insular, ventrolateral PFC, ACC and ventral striatum activation among adolescents (Masten et al. 2009
). Thus, problematic behaviors and peer affiliations may reinforce maladaptive assumptions about attachment and decrease the likelihood of receiving effective co-regulation through friendships. In addition to changing social expectations and behaviors, peer rejection may also produce lasting biological adaptations within these frontolimbic circuits.
By late adolescence, many youth with BPD can be diagnosed reliably (Miller et al. 2008
). Moreover, mood dependent, impulsive behaviors (e.g., self-injury, substance use) often emerge during adolescence (Crowell et al. 2008
). Recently, we have suggested that repetitive self-injurious behaviors in adolescence may represent one stage in a trajectory leading to adult BPD (Beauchaine et al. 2009
; Crowell et al. 2009
). From the perspective of SBT, self-injury likely occurs in the context of co-regulatory failure. This is consistent with findings that interpersonal stressors, such as conflict with parents, peer problems, or the end of a romantic relationship, often precipitate self-injurious behavior (Berman et al. 2006
; Brown et al. 2002
). Chronic interpersonal stress across both family and peer systems may ultimately be a strong predictor of later borderline pathology (Kobak et al. 2009
). To our knowledge, however, there are no data that test this hypothesis sufficiently.
At each stage of development, the child is actively shaping her own trajectory through behavioral response patterns and via active attempts to resolve stage-salient tasks (Cicchetti and Cohen 2006
). Thus, attachment formation is not merely a product of parental availability and support. Rather, child temperament and Temperament × Parenting interactions influence attachment patterns across the life span and shape the emerging adult personality (Bates et al. 1985
; Rothbart et al. 2000
). However, even though there is widespread acceptance of transactional/biosocial explanations of personality development, nearly every study of adolescent or adult BPD examines independent functioning, without any interaction with attachment figures (e.g., Buchheim et al. 2008
; Silbersweig et al. 2007
; Wingenfeld et al. 2009
). Moreover, very few researchers have attempted to investigate individual differences in adult attachment using measures of neural activity.
A growing literature suggests, however, that adult attachment relationships are represented biologically and can be assessed using fMRI. In our work (Coan et al. 2006
), we have examined neural correlates of adult co-regulation. Specifically, we have used fMRI to study women’s brain activation in response to intermittent, mild electric shocks. Threat cues and shocks were delivered during three separate conditions: holding a husband’s hand, no hand, or the hand of an anonymous male experimenter. The findings confirmed that social contact functions to regulate emotional responding during stress. Relative to the other two conditions, when women held their husband’s hand they showed attenuated activation in the neural circuits that subserve emotional and behavioral threat responses. Moreover, marital quality predicted neural responses, such that women with healthier marriages showed lower threat-related activation in the right anterior insula, superior frontal gyrus, and hypothalamus while holding their husband’s hand. In other words, women in happy relationships were able to more effectively “outsource” their emotion regulation to an attachment figure, thereby reducing their own metabolic demands (see Coan 2010
This type of co-regulation can be thought of as a form of “on-line” social support (cf., Coan 2011
). However, internal models of attachment are also crucial to successful adult functioning. For example, during an fMRI task that involved contemplating negative relationship scenarios, adults with insecure attachment styles showed greater activation in emotion-related brain regions when compared with securely attached adults (Gillath et al. 2005
). Furthermore, there was an inverse relation between activation of the anterior temporal pole (a paralimbic structure) and the orbital frontal cortex, such that those with the greatest attachment anxiety showed the least frontal activation and the greatest paralimbic activation. The authors hypothesize that insecurely attached individuals may be less able to modulate emotional processes via top-down control from frontal regions.
In theory, attachment history provides a working template of interpersonal behaviors, which can be accessed to regulate emotions at a lower metabolic cost, even when the attachment figure is not readily available. This extends to pair-bond formation, where attachment history likely influences present-day emotion regulatory activities, including the decision to cede some of the regulatory effort to the potential mate (Coan 2010
). Among adults with an insecure attachment history, these processes may be disrupted.
Applications to Borderline Personality Disorder
Emotion related neural activity is a focus of BPD research. These studies find that when compared with typical controls, those with BPD have increased amygdala activation and decreased ACC activation in response to fear, emotional pictures, fearful faces, and abandonment scripts (Donegan et al. 2003
; Herpertz et al. 2001
; Minzenberg et al. 2007
; Schmahl et al. 2003
). During emotionally challenging tasks, such as the recall of unresolved life events or facing attachment threats alone, adults with BPD show increased activation of the amygdala, insula, and parahippocampal regions relative to controls. (Beblo et al. 2006
; Buchheim et al. 2008
; Schnell et al. 2007
). In some of these studies, adults with BPD also show increased activation in cortical regions involved with regulatory efforts, including the ACC, medial PFC, and OFC compared with typical participants. Those with BPD also differ from controls in response to neutral stimuli, activating similar regions of the brain in response to both neutral and emotional stimuli (Schmahl et al. 2004
; Schnell et al. 2007
; Wingenfeld et al. 2009
During tasks that tap behavioral control, especially within the context of negative emotion, adults with BPD show absent or decreased activation of the cingulate, ventromedial PFC, medial OFC, and subcortical reward regions, along with increasing amygdala activity (Kraus et al. 2010
; Silbersweig et al. 2007
; Völlm et al. 2007
). Adults with BPD also recruit a larger number of brain regions during memory tasks compared with typical controls, suggesting that cognitive tasks are more taxing or effortful for those with the diagnosis (Mensebach et al. 2009
). Thus, fMRI studies reveal broad disruption of frontolimbic circuitry during emotional, cognitive, and behavioral tasks. Importantly, activation within frontolimbic circuitry is intricately related to neurotransmitter and neuropeptide functioning (see Coan 2008
; Eisenberger et al. 2007
). Although beyond the scope of this review, both neurotransmitters and neuropeptides are involved in attachment formation, mood regulation, and behavioral control (see Crowell et al. 2009
; Stanley and Siever 2010
for recent reviews). Serotonin, dopamine, endogenous opioids, oxytocin, and vasopressin play a central role in the development of BPD.
Finally, co-regulation in the partner relationship is an act of trust. Research on the development of trust emphasizes mutual interdependence, which includes both a cooperative intention and expectation (Loomis 1959
). Mutual interdependence occurs when both partners have a shared goal, a desire to reach the goal, and awareness that collaboration is essential. Among adults with BPD, there is evidence that trust is impaired, both behaviorally and in terms of neural response patterns (King-Casas et al. 2008
). Specifically, compared with typical controls, those with BPD are more likely to rupture trust during an economic investment game. This break down of trust occurs in two ways. First, those with BPD are less likely to repay investors in a manner that leads to profit for both players. Second, participants with BPD are less likely to engage in “coaxing behaviors” (i.e., viewing small investments as a sign of broken cooperation and repairing via large repayments).
In contrast, typical controls are more likely to coax the investor, which serves to repair trust and increase the overall amount of money earned for both participants (King-Casas et al. 2008
). The authors report differences in anterior insula activity during this task. While typical controls had strong anterior insula activation when offering a small repayment (i.e., a norm violation), those with BPD did not show such activation when engaging in trust-rupturing behavior. The authors suggest that “the norms used in perception of social gestures are pathologically perturbed or missing altogether among individuals with BPD” (p. 806).
An alternative interpretation is that those with BPD are inexperienced at trust. This includes trusting that another’s behavior is well-intended and also trusting that one’s own behavior will be well received. Based on the findings of King-Casas and colleagues, control participants appear to experience distress when failing to repay investments whereas those with BPD do not. The lack of anterior insula activation may suggest that those with BPD do not experience conflict when cheating the other participant out of money, perhaps because they assume the partner is also attempting to cheat them. Although the authors did not assess motivations or cognitions, those with BPD appear to struggle with representing other people’s motives accurately (Linehan 1993
). They may therefore feel justified in taking as much money as possible, whenever it is available. In other words, those with BPD behave as though the world is scary, awful, and untustworthy. In real-world situations this likely manifests as an inability to trust potential co-regulators. Linehan (1993)
suggests that such doubt is a consequence of an invalidating family environment—one which inconsistently rejects, ignores, invalidates, or rein-forces emotional responses. Implicit in this description of an “invalidating environment” is insufficient co-regulation by the attachment figure.