Distal Antecedents: Developmental Factors

This section outlines studies suggesting developmental risk factors that may facilitate impairments in trust processes in adult individuals with BPD. Orme et al. [43•] tested whether a relative lack of epistemic trust in childhood was associated with BPD symptoms in a sample of adolescent BPD inpatients admitted to a psychiatric unit. They found a significant negative correlation between BPD symptoms at admission and self-reported trusting state toward participants’ mothers and fathers.

Ebert et al. [44] hypothesized that childhood trauma is a risk factor for developing dysfunctional behavioral manifestations during a trust game (TG) procedure in individuals with BPD. Additionally, they were interested in the role of oxytocin. While oxytocin administration is usually associated with higher interpersonal trust [45, 46], the administration of this neuropeptide seems to have a paradoxical effect in BPD (e.g., see below [47]). Increased oxytocin seems to reduce trusting behaviors in individuals with BPD (but not controls), and such a decrease is more important for individuals with BPD who reported greater early parental emotional neglect. Overall, results suggest that developmental factors, such as emotional neglect experiences and lack of trust in parents (as a proxy of epistemic trust), are distal risk factors for trust issues in adults with BPD.

Proximal AntecedentsPrior Beliefs and Dispositions

According to investigations on core dysfunctional beliefs in personality disorders [48], mistrust represents a specific feature of BPD. Butler et al. [49] used the Personality Belief Questionnaire (PBQ) [48] and found that the item that best distinguished BPD from other personality disorders was, “I cannot trust other people.” Such a belief represents a dysfunctional global expectation of others’ trustworthiness. BPD patients self-reported lower interpersonal trust levels than non-clinical controls and patients with MDD or seasonal affective disorder [29]. Similarly, comparing adolescents with BPD with adolescents with other psychiatric conditions and a non-clinical sample, the BPD group self-reported the lowest level of emotional trust [50].

Other studies investigated the role of other dispositions, such as rejection sensitivity (RS, [51]), whose high levels may influence the positive association between BPD and impaired trust processes. Individuals with strong BPD features might be less inclined to trust others because of their concerns and anxiety about the possibility of being rejected or abandoned. In a sample of undergraduates, RS fully mediated the negative association between BPD features and facial appraisal of untrustworthiness [52]. More precisely, only emotional RS components (anger and anxiety about being potentially rejected), not the cognitive one (expectations) (see [53]), mediated the association between BPD features and facial appraisal of others’ trustworthiness [54].

Some studies show that oxytocin administration results in increased trusting behavior during economic games [45]. Bartz et al. [47] hypothesized that individuals with BPD might show an altered response to intranasal oxytocin because its effects on trust and prosocial behavior vary as a function of the relationship representations one possesses. Oxytocin administration in individuals with BPD resulted in expectations of lower cooperation. Conversely, healthy controls showed higher trusting expectations following oxytocin doses than placebo controls. Anxiously attached and rejection-sensitive participants predominantly accounted for these divergent results. Whereas oxytocin generally fosters trust exchanges, the effect is reversed when oxytocin interacts with attachment insecurities and personality traits of rejection sensitivity that are common in BPD patients.

In summary, studies support the model’s stage about the role of dysfunctional prior beliefs regarding others’ trustworthiness for trust impairments in BPD, indicating greater mistrustful beliefs in BPD trustors and pointing to the role of prior dispositions in the impairments in trust processes in BPD.

Perception of the Situation

Two studies using different methodologies demonstrate that BPD patients show state-dependent trust impairments connected to specific situations.

On the one hand, women with BPD did not differ from healthy controls on their partners’ perceived trustworthiness after a neutral conversation, but their trust in their partners decreased after personal or relationship-threatening discussions [55••]. On the other hand, Preuss et al. [56] tested whether social, compared with non-social, situations could activate untrustworthiness appraisal bias more easily in individuals with BPD, using different tasks. BPD participants demonstrated significantly less-consistent behavior (i.e., more investment variability) than the healthy control and MDD group in the social conditions (trust game and punishment game). However, the BPD group did not exhibit such volatility in the non-social conditions.

The Impact of the Trustors’ Emotional State on Trust Processes

King-Casas et al. [57••] used a functional magnetic resonance imaging approach to examine differences between BPD and control groups both behaviorally and neurally during a trust game. Hula et al. [58] analyzed the King-Casas’ data set, adopting an alternative computational model that allows for inferences about three experiences relevant to trust: risk aversion, irritation, and guilt. The authors showed that BPD trustors experienced less guilt and more irritation than healthy control trustors during the economic game. The authors labeled trustees with low guilt-proneness and high irritation as “perilous individuals” who deliberately exploit the trustee and create problematic interactions. Perilousness was more common in the BPD sample compared with controls. Furthermore, perilous individuals were more likely to interpret cooperative situations negatively and were less prone to establish cooperative interchanges or repair cooperation ruptures. Moreover, like perilous individuals, individuals with BPD showed increased irritation from unpleasant interactions during economic exchanges. These results highlight the effects of the trustor’s emotional state in determining mistrust in BPD.

Roberts et al. [59] reported that the administration of acetaminophen, a pain reliever, reduces behavioral mistrust (i.e., low investment) exhibited by participants with high levels of BPD features during a Trust Game procedure. There were similar rates of untrustworthy expectations in individuals with high and low BPD features regardless of acetaminophen or placebo administration. The authors speculated that the decrease in behavioral mistrust in the acetaminophen condition among individuals with high BPD features was due to a reduction in negative emotional affect related to possible unpleasant outcomes in interpersonal interactions (and not due to changes in expectations). Previous evidence showed that acetaminophen reduced negative affective responses (e.g., to rejection; [60, 61]).

Masland and Hooley [62] examined the influence of an emotional prime on trustworthiness appraisal. Non-clinical participants with high versus low borderline features rated unfamiliar faces’ trustworthiness after an affective priming paradigm that exposed them to negative, neutral, or positive images. High-BPD-features individuals showed significantly lower trust appraisal after exposure to negative, neutral, and positive primes than the low-BPD group. However, low- and high-BPD groups showed a significant decrease in trust appraisal after negative emotional primes. Compared with the low-BPD group, negative affective primes influenced appraisal more in the high-BPD group.

To summarize, three studies with different methodologies and samples indicate that individuals with BPD (or higher BPD features) show greater trust impairments with increased state negative affect.

Trust Appraisal

Considering trust appraisal, the human face is a salient source of interpersonal information. The appraisal of others’ trustworthiness is such a relevant judgment for interpersonal exchanges that people, on average, make initial trust appraisals of others based on visual facial morphology after only 100 ms [40].

Fertuck et al. [63] compared facial trustworthiness and fear appraisal in BPD and healthy controls. Compared with controls, BPD participants rated the trustees’ faces as more untrustworthy, indicating the presence of an untrustworthy response bias to all trust faces. By contrast, no significant differences in sensitivity, discriminability, or bias in fear appraisal emerged. Moreover, BPD participants showed slower RT trustworthiness ratings than controls, especially toward more ambiguously trustworthy faces, while there were no differences in RTs for fear ratings between groups.

Nicol et al. [64] found similar results comparing participants with BPD vs. controls when assessing whether facial stimuli appraisal of age, distinctiveness, attractiveness, intelligence, approachability, and trustworthiness. BPD participants showed a significantly larger untrustworthiness bias effect than controls in social dimensions appraisal, such as unknown faces’ approachability and trustworthiness. There was no difference in non-social aspects of the appraisal of others between groups.

Fertuck et al. [65•] replicated the behavioral findings in an fMRI study. Furthermore, BPD participants during untrustworthiness ratings evidenced less activity in the anterior insula and lateral prefrontal cortex than the controls. Such a decrease was proportional to the degree of untrustworthiness bias and impaired discriminability demonstrated by BPD patients and the controls. Individuals with BPD did not show amygdala hyperactivation relative to healthy controls during trustworthiness or fear appraisal. Thus, impaired probabilistic reasoning (linked to prefrontal cortex activity) might be more relevant than hypersensitivity to threatening stimuli (traditionally linked to hyperactivity in the amygdala) in playing a role in trustworthiness appraisal impairments in BPD.

Houben et al. [66••] examined momentary appraisals in a group of individuals with BPD compared to healthy controls in an Ecological Momentary Assessment (EMA) study. Participants were prompted 10 times per day for 8 days to answer questions about their momentary emotions and appraisal of their living situations. Compared to healthy controls, BPD patients experienced lower levels of trustworthiness in trustees in their daily lives.

Finally, Biermann et al. [67] investigated the impact of face masks on trust judgment of faces. Face masks were associated with an overall drop in trust. Irrespective of the presence of a face mask, BPD participants reported lower trust ratings of faces than healthy controls.

In summary, five studies consistently documented a specific bias in lower trust appraisals of trustors’ facial stimuli among BPD patients compared with healthy controls. Furthermore, in our previous section (prior beliefs and dispositions), three additional studies reported results that point in the same direction [52, 54, 62]. This effect is also present when examining BPD patients’ daily lives [66••].

Behavioral Manifestations: Interpersonal Cooperation and Repair of Ruptures in Cooperation

Several researchers have investigated atypical trust manifestations in the BPD population using game theory procedures. King-Casas et al. [57••] used a trust game procedure recording neural activation in a BPD and control group. The authors focused on the capacity to sustain a mutually rewarding, cooperative social exchange (vs. cooperation ruptures) and the ability to repair non-cooperative interactions (“coaxing” behaviors, i.e., when a trustor repays a large part of the investment to the trustee to signal their trustworthiness and gather more substantial investments on subsequent rounds from trustees). Compared with the controls, BPD patients were more likely to initiate cooperation ruptures by sending rejecting social signals. Moreover, BPD trustors had lower rates of coaxing behavior to repair the cooperation ruptures than the controls. Furthermore, anterior insula activity is related to the violation of social norms perception in non-clinical samples, but this was not the case in the BPD sample. Because BPD group showed no insula activation, the authors attributed BPD patients’ low investment behaviors to a lack of sensitivity (assessed via insensitive insula activity) from social norms violations. Furthermore, a lack of insula activation in the BPD group may occur because of dysfunctional beliefs’ top-down influence on neural activity, such as holding negative expectations about social partners.

Unoka et al. [68] replicated King-Casas et al.’s [57••] behavioral effect. The authors used a single-trial TG and a risk game in three trustor groups: BPD, major depression disorder (MDD) individuals, and controls. Additionally, before playing the trust and risk games, participants shared their expectations about the games’ outcomes. In the single trial TG, the trustee can share a fair number of monetary units with the trustor (i.e., the participant) or an unfair amount (violating the investor’s trust). By contrast, in the risk game, the number of monetary units the trustees return to the trustor is determined randomly. The BPD group evidenced lower investment rates in the TG procedure than the MDD and control groups but comparable investment rates to the other groups in the risk game. Moreover, BPD had more skeptical forecasts about the TG outcomes and more accurate estimates about the risk game compared with MDD individuals and controls.

Niedtfeld and Kroneisen [69] used a single-trial TG and tested whether a BPD group show altered memory for cooperative versus non-cooperative interaction partners relative to a control group. Female BPD patients and healthy controls played 40 rounds of single-trial trust games interacting with trustworthy and untrustworthy faces in a source memory paradigm. Half of the rounds resulted in cooperative interactions from trustees, whereas target faces (trustees) behaved uncooperatively for the other half. The BPD group invested lower amounts of money for trustworthy targets than controls. For untrustworthy faces (trustors), on the contrary, no differences emerged between the groups. Moreover, the BPD group had significantly more difficulty recalling cooperative targets than controls. No differences emerged for uncooperative targets.

Saunders et al. [70] found cooperation impairments in BPD patients with an iterated form of the prisoner’s dilemma game [71]. The original prisoners’ dilemma comprises participants who choose to cooperate or defect (i.e., keep all monetary units for themselves) for their sole or joint benefit. The iterated version allows for measuring how individuals acquire and maintain reciprocal altruistic behavioral patterns in multiple exchanges. In the iterated version, the rational strategy is to seek cooperation that maximizes both players’ gains. To get the maximum mutual benefit, the trustor should systematically repeat the trustee’s last choice, undertaking a “tit-for-tat” approach to elicit cooperation from social partners [72]. BPD patients were less able to form reciprocally cooperative relationships with social partners (i.e., they did not assume a tit-for-tat strategy) than the controls [70].

However, Hepp et al. [73] did not find significant differences between a group of individuals with BPD and a group of healthy controls in a dictator game in which participants acted as allocators, thus showing no differences in cooperative behavior in BPD.

Lévay et al. [74] found that BPD patients did not differ in their prosocial disposition from healthy controls in a Social Value Orientation task in which participants must decide how to divide sums of money between themselves and a fictive unknown person. Interestingly, BPD participants were more prone to anticipate a selfish decision in money division from the other.

In summary, four out of six studies support that, compared with controls, BPDs show reduced behavioral cooperative and reparative interactions with trustees.

Trust Learning

In a study [75], trustees expressed happiness or anger during a TG. Trustee fairness was manipulated in the game to display fair vs. unfair behavior. BPD patients showed better performances than control since they could adapt their investments to the actual fairness of the trustee even in the condition with emotional cues, whereas control participants did not show differences in their endowments between fair and unfair trustees when emotional cues were presented. The authors interpreted these results regarding the superiority of Theory of Mind in BPD patients.

Fineberg et al. [42•] investigated the different weights of social and non-social cues in a learning task for BPD participants compared with controls. The task design included five different subphases that varied in terms of volatility and reliability of cues. The participant’s task was to learn the reward probability of social (i.e., partner’s advice during the game) and non-social (i.e., computer’s advice) cues. Learning rates were modeled based on the number of trials occurring between the start of the phase and the engagement of choices consistent with the ongoing condition (i.e., following the advice during stable and reliable stages and not following the advice during volatile or unreliable phases). The authors, examining the transcripts of participants’ answers to debriefing questions, counted the number of times each participant mentioned the confederate. Compared with controls, the BPD group mentioned the confederate more frequently, suggesting more attention paid to and dependence upon social cues. Looking at learning rates during the task, the BPD group learned more slowly than control subjects during all three phases. Furthermore, the BPD group showed slower learning rates in the volatility conditions than the control group.

Abramov et al. [76] implemented a 15-round trust game manipulating trustees’ investment rates with three separate phases: formation of trust, dissolution of trust, and trust restoration. Individuals with high BPD features showed declining trust toward trustees only during the formation of the trust phase. Surprisingly, following trust violation and during the restoration phase with a trustee, trustors with high BPD features showed higher investment rates than individuals with low BPD features. In a reanalysis of these data [77], the authors found that feelings of rejection and self-protective beliefs partly explained these paradoxical effects. These results point to the role of previous beliefs and dispositions in trust learning.

A novel way of considering trust learning processes in BPD is to assess changes in trust appraisal as a consequence of effective psychotherapy. Only one study addresses this issue. Bo et al. [78] investigated changes in trust toward parents and peers as treatment outcomes of MBT group treatment in adolescents with BPD. Although there was no control group, there was significant pre-post changes in trust in peers and parents. Although we should apply caution because these results do not come from a randomized controlled trial, this is the first partial demonstration that BPD adolescents might learn from psychotherapy to trust more their peers and their parents.