significant development changes in bipolar disorder...of both adult and child psychopathology and...

Significant Development Changes

in Bipolar Disorder

REVIEW Open Access

What we have changed our minds about:Part 1. Borderline personality disorder as alimitation of resiliencePeter Fonagy1* , Patrick Luyten1,2, Elizabeth Allison1 and Chloe Campbell1

Abstract

This paper sets out a recent transition in our thinking in relation to psychopathology associated with personalitydisorder, in an approach that integrates our thinking about attachment, mentalizing (understanding ourselvesand others in terms of intentional mental states) and epistemic trust (openness to the reception of socialcommunication that is personally relevant and of generalizable significance) with recent findings on the structureof both adult and child psychopathology and resilience. In this paper – the first of two parts – we review evidencesuggesting that a general psychopathology or p factor underlies vulnerability for psychopathology. We link thisp factor to a lack of resilience using Kalisch and colleagues’ positive appraisal style theory of resilience (PASTOR).We argue that vulnerability for (severe) psychopathology results from impairments in three central mechanismsunderlying resilience – positive situation classification, retrospective reappraisal of threat, and inhibition ofretraumatizing triggers – which in turn result from a lack of flexibility in terms of social communicative processes.We suggest that, from this perspective, personality disorders, and borderline personality disorder (BPD) in particular,can be considered to be the prototype of disorders characterized by a lack of resilience. Part 2 proposes anevolutionary developmental psychopathology account linking this inflexibility in social communication to problemswith the development of epistemic trust – that is, an evolutionary pre-wired social communication system thatnormally facilitates resilience through salutogenesis, that is, the capacity to learn and derive benefit from the (social)environment.

Keywords: Borderline personality disorder, Resilience, Epistemic trust, Mentalizing, Attachment, Psychopathology

BackgroundA challenge for contemporary thinking about psychopath-ology arises from a general neglect by adult psychopatholo-gists of the developmental psychopathology traditionestablished by Sroufe and Rutter [1] over 30 years ago. Spe-cifically, the fact that when we consider an individual’s psy-chiatric history over their life course, it rarely follows thediscrete, symptom-led and time-limited categories thattraditional models have used in conceptualizing mental dis-order. This has increasingly come to be regarded as consti-tuting a slow-burning crisis in the way we understand, andby extension treat, mental disorders. There is a heightenedrecognition of the salience of transdiagnostic features in

clinical presentations as well as across treatment protocols[2, 3]. Particularly in cases of more severe and persistentmental health difficulties, an individual’s clinical presenta-tion changes over time, one typical example being progres-sion from conduct disorder to depression [4], or theextensive comorbidity between traditional ‘symptom’ disor-ders and personality disorders (PDs) (e.g. [5]).Here we posit a reconceptualization of psychopath-

ology associated with PD that speaks to these conceptualand diagnostic enigmas, in an approach that integratesour thinking about mentalizing (i.e. understanding our-selves and others in terms of intentional mental states)and epistemic trust (i.e. openness to the reception of so-cial communication that is personally relevant and ofgeneralizable significance) with recent findings on thestructure of both adult and child psychopathology [3]and resilience [6].

* Correspondence: [email protected] Department of Clinical, Educational and Health Psychology,University College London, London, UKFull list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Fonagy et al. Borderline Personality Disorder andEmotion Dysregulation (2017) 4:11 DOI 10.1186/s40479-017-0061-9

http://crossmark.crossref.org/dialog/?doi=10.1186/s40479-017-0061-9&domain=pdf

http://orcid.org/0000-0003-0229-0091

mailto:[email protected]

http://creativecommons.org/licenses/by/4.0/

http://creativecommons.org/publicdomain/zero/1.0/

Angela Lovelace

At the core of the thinking set out here is anemphasis on the relationship between the social envir-onment as a system on the one hand, and individualdifferences in the capacity for social cognition (as de-fined below) on the other. We argue that the pres-ence or absence of resilience is the outcome of thedynamics of this relationship. Understanding thenature of resilience, we suggest, requires engagementat the level of the mechanism that channels the rela-tionship between the social layer of communicationand the individual’s capacity for reorganizing mentalprocesses. Attempts to intervene at the level of non-resilient responses, we suggest, can be of only limitedeffectiveness. This, we argue, explains the lack of clin-ical response of patients with BPD features to manytraditional psychotherapeutic interventions.A further informing principle is that the type of func-

tioning associated with many forms of psychopathologymight best be understood as an evolutionarily drivenform of entrenched adaptation to stimuli from the socialenvironment – often in interaction with genetic propen-sity [7] – rather than as a mere deficit. It is this adaptiveimperative that underpins the enduring quality that iscentral to definitions of PD. The ‘borderline mind’, andrelated severe problems with social communication typ-ically observed in what we commonly refer to as ‘per-sonality pathology’, may therefore best be understood asa socially triggered outcome, a learned expectation aboutcultural context. Hence, while the processes we describein this paper may be implicated in most, if not all, typesof psychopathology, we consider severe PD, and BPD inparticular, to be prototypical of the type of social com-munication problems that we now see as lying at theroot of vulnerability for severe psychopathology.Finally, in terms of clinical implications, we will indi-

cate how this change in perspective drives a shift in clin-ical focus beyond the consulting room to the widersocial systems that can promote resilience.In the first part of this paper we review emerging evi-

dence that a general psychopathology (or ‘p’) factorunderlying psychopathology provides a comprehensiveexplanation for the extensive comorbidity among disor-ders, as well as many of the other features of individualswho we traditionally consider to be ‘hard to reach’. Wethen argue that this p factor should not be primarilyseen in terms of the presence of specific vulnerabilityfactors (although these may well play an important role,and may be primarily responsible for the phenomeno-logical heterogeneity observed among and within disor-ders), but in terms of the absence of resilience. Weoutline the recently formulated comprehensive positiveappraisal style theory of resilience (PASTOR), and applyit to BPD as the prototype of disorders characterized bythe absence of resilience. We argue that the absence of

resilience in BPD results from an inflexibility in the hu-man capacity for social communication, and in problemswith recalibrating the mind in the face of adverse experi-ences in interaction with others in particular.In the second part of this paper, we will relate this lack

of social communicative flexibility to impairments inepistemic trust from an evolutionary and developmentalpsychopathology perspective, and discuss the clinical im-plications of this shift in our views.

A general factor in psychopathologyOur starting point is the challenge presented to the trad-itional taxonomic structure of psychopathology by co-morbidity (concurrent and sequential over time),recurrence and the unwieldy proliferation of diagnosticdisorders. In our opinion, this challenge has been com-pellingly met by the suggestion that there is a generalfactor of psychopathology – in the words of Caspi andcolleagues, ‘one underlying dimension that summarizedindividuals’ propensity to develop any and all forms ofcommon psychopathologies’ ([3], p. 131). In theiranalysis of the Dunedin longitudinal study, Caspi et al.examined the structure of psychopathology from adoles-cence to mid-life, considering dimensionality, persist-ence, co-occurrence and sequential comorbidity. Theyfound that vulnerability to mental disorder was moreconvincingly described by one general psychopathologyfactor – labelled the ‘p’ (for pathology) factor – than bythree high-order (spectral) factors (internalizing, exter-nalizing and thought disorder). A higher p factor scorewas associated with ‘more life impairment, greaterfamiliality, worse developmental histories, and morecompromised early-life brain function’ ([3], p. 131). Inthe meantime, several studies have replicated thishigher-order p factor [8–11]. Importantly, the p factorconcept may thus also explain why discovering isolatedcauses, consequences or biomarkers and specific, tai-lored treatments for psychiatric disorders has proved soelusive for the field [3].This work on a general factor of psychopathology has

recently also been extended to childhood and adolescence.A longitudinal study of 2450 girls aged 5–11 years, forinstance, has further indicated the criterion validity of thep factor construct, and found it a significantly better fitthan a correlated two-factor (internalizing and externaliz-ing) model [9]. These findings weaken the argument thatthe p factor is a statistical artefact and reinforce the im-portance of further consideration of what the p factormight substantively represent [9]. In a large (n = 23,477)community-based sample aged 11–13.5 years, Patalay etal. investigated the traditional two-factor (internalizingand externalizing) model and a bi-factor model with ageneral psychopathology higher-order model [12].Both models were found to fit the data well; however,

Fonagy et al. Borderline Personality Disorder and Emotion Dysregulation (2017) 4:11 Page 2 of 11

the general psychopathology however better predicted fu-ture psychopathology and academic attainment 3 yearsfrom the time of original assessment; with individuals withhigh p scores being 10 times as likely to have diagnosabledisorder 3 years from assessment than individuals withlower p scores (see also [8]).More specifically in relation to PDs, Sharp and col-

leagues have considered the question of whether a gen-eral factor for psychopathology exists in the context ofPD diagnosis [13]. In a series of exploratory factor ana-lyses based on a sample of 966 inpatients, only four ofthe six PDs (avoidant, schizotypal, narcissistic, and anti-social) examined formed factors with 75% of the criteriathat mark their respective factors. Half the obsessive-compulsive PD criteria loaded with the narcissistic PDcriteria, and the other half split across two other factors.However, Sharp et al. found that (a) a BPD factor in-cluded primary loadings from just over half (55.6%) ofthe BPD items, of which three had notable cross-loadings, each on a different factor; (b) nearly half(44.4%) of BPD items loaded most strongly on threenon-BPD factors (although two had notable cross-loadings on the BPD factor); and (c) the BPD factor wasalso marked by a narcissistic PD item and had notableadditional cross-loadings by other narcissistic as well asavoidant and schizotypal PD items. In the same study,Sharp et al. evaluated a bi-factor model of PD pathologyin which a general factor and several specific factors ofpersonality pathology account for the covariance amongPD criteria. In the bi-factor model, it was found that allBPD criteria loaded only on to the general factor. OtherPDs loaded either on to both the general and a specificfactor or largely only on to a specific factor. The impli-cation of this is that BPD criteria may capture the coreof personality pathology, or may be most representativeof all PDs. To compound more widely the salient statusof BP traits, Caspi et al., in their work on the p factor,found that in terms of personality information, individ-uals who scored highly on the general psychopathologyscale were characterized by ‘three traits that compromiseprocesses by which people maintain stability – lowAgreeableness, low Conscientiousness and high Neuroti-cism; that is, high-p individuals experience difficulties inregulation/control when dealing with others, the envir-onment and the self ’ ([3], p. 131). Such a profile appearsto capture the core features of BPD – emotion dysregu-lation, impulsivity and social dysfunction – and speaksto trait profile approaches to PD [14]. Yet, to claim thatsuch a profile in itself in some sense explains the devel-opmental and life-course forecast that comes from ‘p’obviously risks approaching circularity.The question that then remains is: what is the mean-

ing of the general psychopathology factor at the level ofmental mechanisms? Currently, we can only speculate

about the nature of this generic aetiological factor, butone association to be investigated may be childhoodmaltreatment. Studies indeed suggest that maltreatment,like p, increases the chance of most types of mental ill-ness in adulthood [15] and worsens the course of mentalillness [16]. It has been recently suggested that child-hood maltreatment may be an ecophenotype associatedwith an earlier age at onset of psychopathology, greatersymptom severity, higher levels of comorbidity, greaterrisk for suicide and, importantly, a poorer response totreatment [17].In our opinion, research findings on maltreatment, al-

though still too narrow, indeed point the way to under-standing some of the mechanisms underlying theassociation between the p-factor and vulnerability to(severe) psychopathology This emphasis on the role ofadversity should not be associated with a narrowly envir-onmental position on the relationship between adversityand BPD. Such a position would stand counter to grow-ing evidence for a genetic determinant of BPD. Researchshowing the familial nature of BPD [18, 19], and classicaltwin studies that place heritability of BPD at around40–50% [20–23], have been borne out further by morecomplex behaviour–genetic models that take into accountsiblings, spouses and twins [24]. Although a geneticanomaly associated with BPD has not so far beenidentified, it appears that an endophenotype for thedisorder may be recognized. For example, impulsiveaggression and suicidal behaviour have been linked tothe tryptophan hydroxylase (TPH) gene, and patientswith BPD have a higher frequency of two out of eightpolymorphisms in one of the two known isoforms ofthe TPH gene [25].Impulsive aggression has also been connected with re-

duced serotonergic responsiveness and the inefficient(short or ‘s’) allele of 5-HTTLPR. This has been identi-fied in patients with BPD [26] in some but not all ac-counts (e.g. [27]). There are suggestions that the s allelemarks a vulnerability to stressful life events [28] on theone hand, and the positive influence of maternal sensi-tivity [29] on the other. Accumulating evidence supportsthe view that the s allele, in combination with secure at-tachment, increases agreeable yet autonomous social be-haviour in adolescents [30]. In the context of attachmentinsecurity, this polymorphism is linked with poor self-regulation [31] and impulsiveness [30]. The implicationmay be that the s allele increases social sensitivity, mak-ing a child both more and less prosocial in response todifferent environmental stimuli.Furthermore, the methylation of certain genes could

mediate the long-term effects of adversity [32]. The gluco-corticoid receptor gene promoter, for instance, has beenshown to be more methylated in samples of brain tissueof individuals who had experienced adversity and suicide


[33]. The methylation of NR3C1 is associated with severityof maltreatment from DNA samples collected from per-ipheral blood leucocytes in bipolar disorder [34] and alsoin BPD [35]. In general, inherited differences in specificgenes thus may moderate the effects of adversity and de-termine who is more resilient [36].Interactional models of biological vulnerability com-

bined with psychosocial risks are therefore being in-creasingly considered in relation to BPD (e.g. [37, 38]).The emphasis placed on social adversity in this papershould not be regarded as a statement of the exclusivepre-eminence of the environment in understanding thedevelopmental origins of PD. Rather, the assumptionthat should be understood to underpin our discussion ofthe role of maltreatment and adversity is that such expe-riences in individuals who are biologically susceptible(and there may be different genetic routes that lead tothis susceptibility) cumulatively strain the viability of re-silience and, as we shall demonstrate, epistemic trust.

BPD as a limitation of psychological resilienceIn further clarifying the relationship between BPD andthe p factor, Kalisch and colleagues’ [6] conceptualframework for the neurobiology of resilience is enlight-ening. Kalisch et al. [6] argue that psychological resili-ence is not an absence of disease processes, but areflection of the work of active, biologically based mech-anisms. In considering the relationship between PD andadversity, we have similarly tended to focus on identify-ing the characteristics of the patient who is experiencingmental health difficulties rather than attempting to de-lineate the competencies or capacities of the person whohas remained functional and free of disorder despitesubstantial hardship. In fact, studies suggest that only aminority of individuals develop persisting trauma-relatedpathology as a result of experiencing or witnessing a sin-gle extreme or life-threatening event (e.g. Type Itrauma). The majority of people have a remarkable cap-acity for resilience when faced with such events [39, 40].Rather than searching for the clinical indicators of a

transdiagnostic concept such as p, we may be wiser con-ceptualizing p as an indication of the absence of resilienceand focusing on identifying mechanisms that ‘normally’protect individuals from harsh conditions. Perhaps p maybe more appropriately considered as pointing to protec-tion (or rather the absence of protection).Resilience has always been an important theme in dis-

course on mental health [41, 42] but recent concernsabout healthcare costs have led to the concept increas-ingly occupying centre stage [43]. Work on the topiccovers myriad different factors and explanations associ-ated with psychological resilience, such as living in a stableand comfortable neighbourhood, family resources andfamily support, participating in community sporting or

extracurricular activities, racial or gender socialization, be-ing securely attached, being able to regulate one’s emo-tions, exposure to a sensitive style of parenting, or geneticfactors. Many of these factors overlap conceptually as wellas statistically. They are not explanations for resilience,but rather factors that predict the activation of psycho-logical or biological mechanisms that produce resilience(the absence of pathology in the presence of adversity) asan outcome. Sadly, this conceptual clarity is often lackingin writings about resilience, especially those that concerninterventions aimed at its promotion.The diverse accounts of resilience, often advanced at

radically different levels of explanation – from socioeco-nomic through to genetic – can be unified within thepositive appraisal style theory of resilience (PASTOR) con-ceptual framework presented by Kalisch et al. [6]. Accord-ing to this formulation, the process underlying resilience isdriven by top-down processes in the form of the appraisalthat is made of a stressful stimulus. The external and so-cial factors that have been associated with resilience (suchas social support or a secure attachment history) affectresilience either directly or indirectly in that they shapethe individual’s appraisal approach, or minimize exposureto stressors. This is not to deny the role of socio-environmental factors in determining an individual’sresilience, or to deny the importance of interventionsat a social or community level; it is to suggest thatthe mechanism by which these distal social factorsaffect individual resilience is via their impact on theindividual’s appraisal style.

Resilience and reappraisalThe appraisal theory of resilience is based on a specificunderstanding of the nature of higher-order cognition[44]. The theory is that the resilience process is as follows:a potentially stressful stimulus is perceived and mentallyrepresented by the individual. The mental representationis then appraised using higher-order cognition, understoodin terms of an ensemble of psychological mechanisms andphenomena, including executive function, attention, gen-eral intelligence and self-awareness. This in turn deter-mines the emotional response of the individual – theirresilience.We consider this an important perspective but a narrow

interpretation of what may be considered higher-order cog-nition. The outputs of neural processing intrinsically de-pend on the processing units that take input from theoutput of other units, perform specific functions, and gen-erate output that in turn becomes the input of other pro-cesses. In most models of brain function, any psychologicalcapacity is underpinned by a large number of such hypo-thetical processes [45, 46]. In this context, the nature of theorganization of processing units, or indeed the system thatdetermines their relative activation, may be either a simple


function of the efficiency of processing or, within a hier-archical system, determined by the functioning of a higher-order system. The higher-order meta-system monitors theperformance of lower-order systems to ensure optimal per-formance within a particular context. These components ofhigher-order cognition are what constitute the core of anormal wakeful and wilful mind in the process of consciousperception, imagination, decision-making and action plan-ning. These functions, taken together, create an opportun-ity for the internal reorganization of neural structureswithin the human brain. A consistently ‘self-observing’process, which monitors the quality of outcome of neuralprocessing units, enables the individual to reorganize theway neural structures subserve cognitive function. Menta-lizing is a key facet of this self-observational process, andthe extent to which intentionality fulfils expected behav-ioural outcomes is a critical indicator of the efficiency ofneural processing and guides the way information pro-cessing is organized within available pertinent neuralunits. We assume that an efficiently functioning humanbrain representing a resilient system achieves such robust-ness because mentalizing provides a clear window on theefficiency of brain functioning. Multiple processing unitscover similar functions in the brain. Some units, beingmore efficient than others, are more likely to be providersof output that is taken forward to other units. But circum-stances change, and demands for adaption may reversethe hierarchy of efficient functioning of these processingunits. Resilience is the appropriate appraisal and monitor-ing of the external social environment and internal func-tioning of processing units. Thus, as we will explain inmore detail in Part 2 of this paper, higher-order cognitionis the developmental capacity, based on early relationshipsand constantly renewed in changing social contexts, to ap-praise the efficiency of functioning, which in essence isintersubjective in its nature. The capacity to anticipate thereaction of another person, to regulate attention or to im-plement action plans are all shaped by the overarchingneed for survival in the context of social interaction. Afailure of resilience arises when the individual is unable tochange processing systems in a sufficiently flexible mannerto maintain optimum outcome despite changed circum-stances. When an individual cannot disengage a process-ing system that is no longer appropriate to the task – forexample, a child whose perfectionistic attitudes servethem well during a period of knowledge acquisition andrelatively simple tasks, but cause great problems whentask complexity has increased to a point where perfectionis impossible – the lack of flexibility is what creates vul-nerability. Insensitivity does not create risk; the sensitivityof higher-order cognition is what provides protectionthrough the appropriate appraisal of the functioning ofneural structures relative to the environment. This is howthe resilient brain functions; it is not a model that skirts

reification – it is a description of our assumptions of thenature of brain function.Higher-order cognition appears to be more flexible

within the brain than other, more specialized modalforms of cognition such as basic vision and hearing. Forexample, brains are able to preserve core aspects of thefunctional architecture of the information processingthat sustains higher-order cognition in spite of substan-tial structural damage [47]. Higher-order cognition is aform of information processing, therefore, that does notcompletely rely on one single, static or fixed set of spe-cialized brain regions and anatomical connections,within certain limits of course. It works by exploitingavailable neural resources and possible routes betweenthem; it seems to use degenerated and pluripotent brainsystems flexibly, enabling higher-order cognition toemerge as one of the most robust brain functions. Inthat sense, the mind does not exist in one physical loca-tion within the brain; rather, it is an abstraction, or code,and the brain is the code interpreter. Basic conscious-ness – the mechanism for the resiliency of cognitive andcontrol systems – is thought to have evolved to be max-imally resilient itself: ‘consciousness itself can be inter-preted as a general algorithm for resilience selected byevolution’ ([47], p.22). This decoupling of higher-ordercognition from a single location appears to be highlyadaptive: its relatively abstract and algorithmic naturemakes it more robust in the face of any localized damageor degeneration within the brain.The algorithmic quality of consciousness may be

regarded as a pinnacle of human evolution, but thisshould not detract from its highly pragmatic, adaptivepurposes. This resilient framework is an essential condi-tion for functioning autonomy and the capacity to adaptto the world’s demands – particularly the highly complexdemands of the human social world. As Paradiso andRudrauf [48] have argued in their article on social cogni-tion and social neuroscience, tellingly entitled ‘Strugglefor life, struggle for love and recognition: the neglectedself in social cognitive neuroscience’, the self, self-awareness and intersubjectivity are integral to social cog-nitions and actions. As described above, the appropriatefunctioning of higher-order cognition crucially dependson appropriate judgements about social contexts. In thissense, social cognition is part of the mechanism ofhigher-order cognition, although social cognition itself ismade up of a set processes that are monitored by themetacognitive evaluations that higher-order cognitionperforms: as in any feedback system, there is an inherentcircularity in this conceptualization. This is inevitablegiven that we are describing the extent to which a sys-tem is capable of reorganizing its own functioning. Simi-larly, the modes of operationalizing the self and theidentification of self-awareness are strongly shaped by


developmental contributions from the social environ-ment – parents, sibling, peers and significant others. Inother words, the abstract algorithm that creates personalconsciousness cannot be separated from social interac-tions. This is what the algorithm was developed for, andwhat further shapes the algorithm of the self and its on-going relationship with the outside world.Although there are many factors at work in contribut-

ing to resilience, Kalisch describes the three underpin-ning appraisal mechanisms that determine resilientbehaviour and responses [6], as follows:

1 Positive situation classification. This refers to themanner of immediate appraisal of a situation at themoment of encountering it (e.g. ‘What is the personapproaching me carrying in their hand?’). In the caseof an insignificant threat, a positive appraisal styleenables the individual to view it in a manageableperspective. Clearly, in the context of an adverseevent, a negative appraisal and stress response arecalled for. In such situations, resilience can besubsequently promoted through the second andthird forms of appraisal.

2 The retrospective reappraisal of threat. Whether atraumatic event results in post-traumatic stressdisorder, for example, is dependent on how it isretrospectively reappraised [49, 50]. This, as Kalischet al. describe it, ‘shifts the emphasis from theexternal situation (or changes in the situation) tothe individual’s ability to flexibly adjust currentnegative appraisal or to implement new, morepositive appraisals and then to maintain thoseappraisals. Both processes have to occur in the faceof interference from automatic and uncontrollednegative appraisals and the accompanying aversiveemotional states’ ([6], p. 14).

3 Inhibition of retraumatizing triggers. This mechanismenables the individual to inhibit the threat-associatedsensations that might beexperienced when remembering a traumaticevent and serve to reinforce, perpetuate andgeneralize the sense of threat.

BPD and the PASTOR model of resilienceTo return to BPD, we can follow the PASTOR model bydistinguishing between resilience factors and mecha-nisms. We suggest that a traditional clinical mistake inthe treatment of BPD has been to intervene at the levelof resilience factors rather than at the level of appraisal(i.e. mechanisms) – this in effect means that we havebeen working at the level of correlation rather thancausation. In BPD, the appraisal mechanisms are at fault,in large part because of mentalizing difficulties (e.g. inthe mistaken appraisal of threat at the moment of its

presentation) or a breakdown in epistemic trust,which damages the capacity to relearn different waysof mentalizing – or appraising – situations (i.e. the inabilityto change our understanding of the threat after the event).The outcome is the lack of resilience that is highly charac-teristic of BPD, regardless of its clinical presentation.

BPD and positive situation appraisalMentalizing has an interpretive role and allows us to ex-plain and predict behaviour; in this sense it also has asocial regulatory role [51]. Behaviour can be producedby rational interactions among beliefs and desires,which, when interpreted (appraised) according to spe-cific culturally determined expectations, generate mean-ing (a meaning assigned to the observed action) in termsof putative mental states that could have engendered theperceived behaviours. Therefore, for our behaviour to besocially meaningful (predictable), it can and should obeythese same conventions. Frequent behavioural deviationsfrom these expectations may be considered as being coreto PD. This is confusing and stressful for the observerbecause the normal process of reconstructing mentalstates from actions is disrupted.The great importance of this process of meaning

generation has been powerfully illustrated by studiesin which participants were led to believe that deter-ministic neurological processes, rather than mentalstates, control behaviour: in other words, they werediscouraged from believing in free will. Introducingan abstract disbelief in free will led to an observedweakening of neural signals associated with readinessplanning; subjects became less prepared to act volun-tarily [52]. Setting up a deterministic neurological biasalso appeared to ‘free’ individuals from a sense of per-sonal responsibility and generated more antisocialcheating and aggression [53].If mentalizing is assumed to have such an interpret-

ative and regulatory role, then individuals with BPD whohave limited capacity to exercise this regulative functionare at least partially deprived of the appraisal processesneeded to reduce the stress of any social experience.This leaves them at times confused and vulnerable inboth the interpretation and the convention-governed ex-pression of mental states in behaviour. To put it plainly,they are frequently puzzled by others’ actions, andequally find themselves victims of misattributions byothers. There is ample clinical evidence of limitations ofappraisal in BPD (for examples, see [54–57]) althoughundoubtedly, as would be predicted by the p factormodel, they are by no means the only clinical group toshow concerning limitations in this area. Poor appraisalmay be more severe and pervasive in BPD than, forinstance, in major depressive disorder or generalizedanxiety disorder without PD comorbidity.


Individuals with BPD tend to be very prone toautomatic, non-reflective mentalizing; they often basetheir inferences on the immediate exterior features ofothers, and rely on affective rather than cognitivementalizing. This has clear implications for the stylein which they are likely to appraise social situations.As a result of their mentalizing tendencies, individualswith BPD tend to appraise situations and read others’expressions quite quickly: they may show a hypersen-sitivity to facial expressions [58, 59] and higher-than-normal sensitivity to non-verbal communication [60,61]. For example, individuals with BPD have been found tooutperform non-BPD comparisons on the Reading theMind in the Eyes Test [62] or to be at least as good as nor-mal controls on the same test [56, 63]. However, this em-phasis on external and immediate cues in appraisalsituations is accompanied by difficulties in making more re-flective judgements based on what might be going on insidepeople’s minds – so, for example, individuals with BPDtend to perform more poorly in social exchange tasks [55,64]. They have also been found to be more likely to viewcharacters/behaviours as negative or aggressive [65]; to havean impaired view of neutral faces in the context of anger ordisgust [66]; and to react with hostility to neutral social in-teractions [67] – all suggestive of the negative appraisalstyle described by Kalisch and colleagues [6]. The emphasison affective mentalizing also results in a heightened sensi-tivity to emotional cues [59], especially in cases of angerand fear [68, 69]. Furthermore, unbalanced mentalizing onthe self–other dimension can cause individuals with BPDto experience severe difficulties in separating the self fromthe other [70–73] and to be unduly emotionally affected byothers’ affective states. This often leads to the experience ofemotional contagion, which has clear implications for socialappraisal situations [74, 75]: BPD individuals can feel forcedto be rigid and highly controlling in order to maintain asubjective sense of coherence and integrity [76].The mentalizing profile characteristic of an individual

with BPD, in sum, results in an oversensitivity to possiblydifficult social interactions (because distortions in menta-lizing are more likely to result in mistaken interpretationsof others’ behaviour and motivation). In the aftermath of achallenging or stressful interaction, it is difficult for the in-dividual to make sense of, contextualize or put aside po-tentially upsetting memories of experiences, leaving themmore vulnerable to emotional storms. A capacity for expli-cit, reflective mentalizing in particular serves a dual inter-pretive (appraisal-strengthening) and self-regulatory role.The absence of this capacity deprives the individual of afundamental tool in reducing stress.However, one can see that in certain situations, for ex-

ample, an emergency milieu characterized by high levelsof interpersonal aggression, the heightened and immediatesensitivity and seemingly instinctive and physically

charged form of appraisal characteristic of BPD might infact be adaptive, at least in the short term. In such an en-vironment, extreme vigilance is a potential advantage, andsimilarly, the ability to form intense emotional relation-ships quickly might elicit resources or protection. Thementalizing profile associated with BPD and the appraisalstyle this generates is maladaptive in most stable socialcontexts, but we postulate that this mentalizing profilemay be a response to cues suggestive of an unreliable andpotentially threatening social environment. We thusshould be wary of seeing apparent dysfunctions of theclinically ‘hard to reach’ as indicative of a deficit or anykind of sub-optimal functioning (as, indeed, we have donepreviously [77]). We would now consider that what mayappear to us as dysfunction is an evolutionarily primedadaptation to specific environmental and social contexts.As a genetically triggered adaptation, the individual is bio-logically programmed to resist change in a behaviour pat-tern that signals increased chances of selection. Webelieve that enduring mental disorders (including BPD)are nested in the context of the evolutionary priorities ofthe human condition.

BPD and retrospective reappraisalThe mentalizing difficulties of BPD patients have alsoconsiderable implications for understanding the difficul-ties with retrospective reappraisal that may undermineresilience. Reappraisal can attenuate ongoing stress re-sponses by appropriately adjusting negative appraisalsand/or generating complementary positive appraisals. Instrongly aversive situations the stress response is essen-tially unavoidable: the experience is automatically classi-fied as negative and requires ‘after the event’ changes inthe meaning of the stimuli. This is often achievedthrough reappraisal in terms of the mental states of theprotagonists. To retrospectively appraise an event orsituation in a way that promotes resilience, an individualneeds to be able to reappraise it in a way that involvesreflective, cognitive mentalizing. Such reappraisal willoften also depend upon a capacity to mentalize the in-ternal states of both the other and the self. In otherwords, the mentalizing strengths that this form of retro-spective reappraisal requires are not congruent with thementalizing profile typical of BPD, which is character-ized by (a) a tendency to focus on the external ratherthan internal states of others; (b) the dominance of auto-matic, intuitive mentalizing over controlled, reflective,mentalizing that could help to put the potentially trau-matic event into perspective;(c) an imbalance betweenaffect and cognition in favour of the former, leading toself-perpetuating persistence of negative affect; and, fi-nally, (d) difficulties in coherently representing the selfindependently of the other, undermining the potential tocontextualize and make proportionate an event.


The mentalizing model for trauma has reappraisal ofphysical and psychological experience at its core [78, 79].Similarly, trauma-focused cognitive-behavioural therapyand other exposure-based therapies (e.g. eye movementdesensitization and reprocessing therapy) enhance menta-lizing of the trauma experience, creating a second-orderrepresentation of the event in terms of greater coherenceof the subjective experience of the victim and often alsothe perpetrator. Patients with BPD have a specific problemin relation to reappraisal proper because they find it chal-lenging to generate second-order representations of men-tal states that might be modified to constitute morepositive reappraisals of experiences or modify and thusmitigate (adjust) negative appraisals. In essence, this lies atthe core of Gunderson and Lyons-Ruth’s interpersonalhypersensitivity theory of BPD [80]. Interpersonal hyper-sensitivity is the likely consequence of a failure of re-appraisal following stressful social interactions. In theabsence of being able to mentalize in a balanced way, anevent or a relationship can be endlessly discussed and dis-sected in an apparent attempt at reappraisal, but such at-tempts have an unreal quality. Complicated inferencesabout mental states are made, but they might have littleconnection with reality. We term this pseudomentalizing,or in extreme, hypermentalizing; it is a state of mind thatcan be clinically misleading in that it may present as astrong attempt at reflection and engagement, but it willultimately be circular and unproductive. Hypermentalizingof trauma, the failure to move on from it, may be inevit-able if individuals cannot reliably access and use socialcommunication that could enable them to resolve or con-tain the sense of threat associated with a trauma (or if aperceived threat that has been misinterpreted as such,owing to problems in the first resilience mechanism).However, as our understanding of this state of ‘petrifica-tion’ has deepened [81], we also have come to recognizethat mentalizing is not everything, or rather, that bodilyexperience has an important role in enabling access to fur-ther resilience strategies. This brings us to the importanceof inhibition mechanisms.

BPD and the interference inhibition mechanismAccording to Kalisch et al.’s conceptual framework [6],the final level of appraisal underpinning resilience is aninhibition mechanism based on interference. As men-tioned above, a strongly aversive event naturally gener-ates powerful negative appraisal responses. The ability tomoderate and regulate such negative responses after theevent can further determine the extent to which theevent continues to cause difficulties in psychologicalfunctioning. This implies the inhibition of conflictivenegative appraisals and acting deliberately to interferewith emotional reactions to information processing. Theinhibition of negative and disruptive responses through

distraction or interference can enable the individual tobegin the process of reappraisal proper, allowing a moreresilient response to emerge. An individual’s inhibitorycapacity may to a large part be a trait-like characteristic,with some genetic basis. However, the extent to whichthe inhibition mechanism can be overwhelmed and howits restoration can be managed may be malleable tosome degree.Although much has been written on the nature of

traumatic experiences, within the view outlined in thispaper, an aversive event becomes traumatic in its after-math when it is accompanied by a sense that one is notaccompanied – that one’s mental experience is notshared and the ‘mind is alone’ [78, 82]. Trauma obtainsfrom a primitive, adaptive human terror of isolation.Here, again, we run into the key importance of socialreferencing to calibrate the mind. In the process of re-appraisal, the social referencing provided by being ableto access another mind enables us to frame and put intoperspective an otherwise overwhelmingly frightening ex-perience. This process, which drives a so-calledbroaden-and-build cycle [83], is far more available to in-dividuals who are open to the benign social influence ofother minds. As outlined in more detail in Part 2 of thispaper, those who are able to manifest sufficient levels ofepistemic trust to embark on the mutually mentalizingstance that is essential in soliciting other minds in sup-port of one’s own, are therefore more likely to be resili-ent. The commonly observed vicious cycle of BPD,comorbid trauma and the acute subjective experience ofisolation captures the implications of the failure of thisinhibition reappraisal mechanism.Individuals with the diagnosis of BPD have been

shown to have serious limitations in their capacity forthe inhibition of conflictive negative appraisals and forinterfering emotional reactions to information process-ing. They cannot cognitively inhibit retraumatizing trig-gers, leaving them vulnerable to threat-associatedsensations that might be experienced when rememberinga traumatic event, which serve to reinforce the sense ofthreat. It is not possible for these individuals to accessmentalizing if the self is overwhelmed by negative inter-ference that impairs normal cognitive function. This iscongruent with the view that emotional dysregulation isthe fundamental problem in BPD [84–86]. The idea of afailure of inhibition in BPD also echoes recently reportedfindings from Koenigsberg et al. concerning the failureof habituation in BPD [87, 88], which may have a geneticbasis [89].We have similarly (albeit not formulated in terms of

the failure of interference or habituation) described thephenomenology of the unyielding nature of trauma-linked subjective experience in BPD [90] in terms ofalien self experiences that consist of a sense of looming,


unmanageable anxieties that cannot be reappraised andcontained, as the subjective outcome of incorporating anexperience of overwhelming hostility into the self [91].In this context, the focus is not on the development ofthis experience but rather how it is so persistently main-tained despite intense and persistent efforts at re-appraisal. This shift in perspective involves a recognitionof the significance of the capacity for inhibition in thetreatment of BPD. Individuals who are very poor atmentalizing may require more than cognitive interven-tions (talking) to bring about the inhibition of the stressresponse. Interventions may have to relate to the bodymore directly. We have always had a view that mentaliz-ing was embodied [92], but we have not treated this factseriously enough. We now see a role for physical activityin strengthening the capacity for inhibition at the sametime as helping to restore mentalizing. In clinical terms,we suggest that physical activity has a role in strengthen-ing the capacity for inhibition at the same time as, or asa precursor to, helping to restore mentalizing. Perhapsnew areas for developing effective interventions may liein this direction. For example, if an adolescent cannotcommunicate, activating interference to permit re-appraisal via physical activity may well be more valuablethan spending hours attempting to activate mentalizingvia talking and reflection. The best initial approach maybe a physical one: running with them, and discussingwhat the running was like. Such a simple focus on theembodied experience can be used to begin to rehearsethe most basic principle of responding to and givingspace to a stimulus outside the negative responses thatnormally overwhelm other forms of social cognition.

ConclusionsAlthough we still consider attachment and mentalizingto be key in our understanding of personality pathology,and in earlier formulations we have always emphasizedthe importance of the absence of resilience in BPD, therehas been a notable shift in our views on the emergenceand nature of BPD. Rather than seeing BPD primarily interms of the presence of impairments in attachment andmentalizing, we consider the notable absence of resili-ence and the social communicative inflexibility thatseems to underlie this absence as an adaptive strategythat individuals with BPD acquire within a social contextwhere social inflexibility was often the only possible sur-vival strategy and had considerable advantages in theshort term.We will further elaborate on these issues in Part 2 of

this paper. Currently we are still faced with an importanttheoretical dilemma: from where does this absence ofpositive reappraisal mechanisms stem? How can weunderstand the inflexibility in social communicative pro-cesses in BPD and in all those suffering from serious

psychopathology, which seems to render these individ-uals so ‘hard to reach’? How did this inflexibility developover time? We believe that the answers to these ques-tions lie in an evolutionarily informed developmentalpsychopathology account of BPD and related disordersthat has considerable implications for prevention andintervention.

AbbreviationsBPD: Borderline personality disorder; PASTOR: Positive appraisal style theoryof resilience; PD: Personality disorder

AcknowledgementsNot applicable.

FundingPF is in receipt of a National Institute for Health Research (NIHR) SeniorInvestigator Award (NF-SI-0514-10157). PF was in part supported by the NIHRCollaboration for Leadership in Applied Health Research and Care (CLAHRC)North Thames at Barts Health NHS Trust. The views expressed are those ofthe authors and not necessarily those of the NHS, the NIHR or theDepartment of Health.

Availability of data and materialsNot applicable.

Authors’ contributionsAll the authors contributed equally to the writing of this article. All authorsread and approved the final manuscript.

Competing interestsThe authors declare that they have no competing interests.

Consent for publicationNot applicable.

Ethics approval and consent to participateNot applicable.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Author details1Research Department of Clinical, Educational and Health Psychology,University College London, London, UK. 2Faculty of Psychology andEducational Sciences, KU Leuven, Leuven, Belgium.

Received: 17 August 2016 Accepted: 21 March 2017

References1. Sroufe LA, Rutter M. The domain of developmental psychopathology. Child

Dev. 1984;55:17–29.2. Cuthbert BN, Insel TR. Toward the future of psychiatric diagnosis: the seven

pillars of RDoC. BMC Med. 2013;11:126.3. Caspi A, Houts RM, Belsky DW, Goldman-Mellor SJ, Harrington H, Israel S,

Meier MH, Ramrakha S, Shalev I, Poulton R, Moffitt TE. The p factor: onegeneral psychopathology factor in the structure of psychiatric disorders?Clin Psychol Sci. 2014;2:119–37.

4. Simic M, Fombonne E. Depressive conduct disorder: Symptom patterns andcorrelates in referred children and adolescents. J Affect Dis. 2001;62:175–85.

5. Luyten P, Fonagy P. Psychodynamic treatment for borderline personalitydisorder and mood disorders: a mentalizing perspective. In: Choi-Kain LW,G. GJ, editors. Borderline personality and mood disorders: Comorbidity andcontroversy. New York: Springer; 2015. p 223-51.

6. Kalisch R, Muller MB, Tuscher O. A conceptual framework for theneurobiological study of resilience. Behav Brain Sci. 2015;38:e92.


REVIEW Open Access

What we have changed our minds about:Part 2. Borderline personality disorder,epistemic trust and the developmentalsignificance of social communicationPeter Fonagy1* , Patrick Luyten1,2, Elizabeth Allison1 and Chloe Campbell1

Abstract

In Part 1 of this paper, we discussed emerging evidence suggesting that a general psychopathology or ‘p’ factorunderlying the various forms of psychopathology should be conceptualized in terms of the absence of resilience,that is, the absence of positive reappraisal mechanisms when faced with adversity. These impairments in thecapacity for positive reappraisal seem to provide a comprehensive explanation for the association betweenthe p factor and comorbidity, future caseness, and the ‘hard-to-reach’ character of many patients with severepersonality pathology, most notably borderline personality disorder (BPD). In this, the second part of the paper, wetrace the development of the absence of resilience to disruptions in the emergence of human social communication,based on recent evolutionary and developmental psychopathology accounts. We argue that BPD and related disordersmay be reconceptualized as a form of social understanding in which epistemic hypervigilance, distrustor outright epistemic freezing is an adaptive consequence of the social learning environment. Negativeappraisal mechanisms become overriding, particularly in situations of attachment stress. This constitutes ashift towards a more socially oriented perspective on personality psychopathology in which the absenceof psychological resilience is seen as a learned response to the transmission of social knowledge. This shiftin our views has also forced us to reconsider the role of attachment in BPD. The implications for preventionand intervention of this novel approach are discussed.

Keywords: Borderline personality disorder, Resilience, Epistemic trust, Mentalizing, Attachment,Psychopathology

BackgroundBringing together the threads of the argument we builtin Part 1 of this paper, we propose that the commonvariance revealed by bi-factor studies of psychopathologyindicates a shared variance in resisting socially expect-able adversity. Moreover, persistent psychological dis-tress associated with personality disorder (PD) has as acommon element diagnostic criteria that we may par-ticularly expect to see in BPD, making BPD features thecore features linked to persistence of psychiatric prob-lems. So far, we have outlined a model that inverses this

vulnerability from one focused on the common charac-teristics of the pathological condition to an alternativeperspective that highlights the absence of resilience asthe shared cause. Following Kalisch et al.’s persuasivemodel of resilience [1], we argued that the persistence ofpsychopathology, as observed prototypically in BPD,results from a pervasive limitation on the appraisal ofstressful social experience, which could be linked tolimitations in the capacity to mentalize.What may explain this absence of capacity to

reappraise stressful social experiences? Here, recent evo-lutionary and developmental accounts of the emergenceof epistemic trust in humans may provide importantanswers. These views also, as we will demonstrate,necessitate a shift in our perspective on the role of

* Correspondence: [email protected] Department of Clinical, Educational and Health Psychology,University College London, London, UKFull list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Fonagy et al. Borderline Personality Disorder and Emotion Dysregulation (2017) 4:9 DOI 10.1186/s40479-017-0062-8

http://crossmark.crossref.org/dialog/?doi=10.1186/s40479-017-0062-8&domain=pdf

http://orcid.org/0000-0003-0229-0091

mailto:[email protected]

http://creativecommons.org/licenses/by/4.0/

http://creativecommons.org/publicdomain/zero/1.0/

attachment in BPD. Put briefly, the theory of epistemictrust posits that the human infant – most usually firstwithin the context of early attachment relationships – isinstinctively inclined to develop openness to the recep-tion of social communications from their primary care-givers. Stated otherwise, epistemic trust is an adaptationallowing the infant to receive social knowledge fromtheir better-informed elders [2], enabling them to benefitfrom the complex edifice of human knowledge that theirimmediate culture has available to them.There are two possible bases on which cultural

knowledge can be accepted by a learner as credible:they can either work it out for themselves (which istime-consuming, difficult, and often impossible) orthey can rely on the epistemic trust they have in theauthority of the communicator [3, 4]. Trusting thecommunicator means that the learner does not haveto go back to first principles each time they encoun-ter novelty: a strange-looking tool without a self-evident purpose is accepted as being used asdescribed by a trusted elder, because they have saidso [5]. Being told in this way is enough, and saves anawful lot of time and effort, and indeed possiblyallows the infant to grow up and build upon orrevolutionize the use of the tool in question. Thiscapacity to teach and learn social knowledge largelyunderpins the evolution of human culture [6]: it hasbeen proposed that this form of cultural evolution,based on the transmission of knowledge via epistemi-cally trusted communication, emerged during the latePleistocene era [7].The internalization of knowledge about the social

world constitutes a particular kind of learning: it in-volves encoding the piece of knowledge as signifi-cant, relevant to the recipient and sociallygeneralizable – that is, as an accepted and reusablepiece of cultural currency. This specific form oflearning is stimulated by ostensive cues generated bythe communicator [8, 9]. Such cues trigger a peda-gogic stance in the recipient, priming them to regardforthcoming communications as significant. Humaninfants display species-specific sensitivity and defer-ence to non-verbal ostensive cues, such as eye con-tact, turn-taking contingent reactivity, being calledby their name, and the use of a special tone of voice(‘motherese’) by the communicator [10, 11]. Theseostensive cues have in common the quality that therecipient is recognized as a subjective, agentive self.Once epistemic trust is stimulated in this way, thechannel for the transmission of knowledge is opened.Mimicry may be protected by human evolutionbecause it generates epistemic trust, inevitably sig-nalling recognition in the child by the imitatingadult. A social smile (recognition of the self by the

other) probably increases the tendency for imitationbecause the smile generates epistemic trust andopens the communication channel to receiveknowledge.It has been argued that this mechanism for opening

the epistemic channel exists because it cannot be leftopen by default: it is adaptive for humans to adopt aposition of epistemic vigilance unless they are reassuredthat it is safe to do otherwise [4, 5]. The notion that chil-dren are promiscuously credulous to those around themhas been disproved by ample evidence suggesting theways in which dubious social signifiers and poor pastperformance may render a social communicator suspectand their assertions about the world regarded with scep-ticism [12, 13]. Epistemic vigilance is a necessary tool toprotect against misinformation, whether as a conse-quence of malicious intent or incompetence on the partof the communicator [4]. Therefore, although the pur-pose of epistemic trust is the transmission of data, itsapplication is a highly psychological and relationalprocess, dependent on calculations about who is trust-worthy, authoritative and knowledgeable –in otherwords, about whose information is worthy of beingencoded as relevant and culturally significant to the self.

Epistemic mistrust and developmentalpsychopathologyIn situations where a young learner’s early environmentis heavily populated by unreliable communicators, theopening of epistemic trust becomes problematic: it maybe more adaptive to remain persistently vigilant about,or even closed off to, the communication of social know-ledge. In the face of an abusive and hostile caregiver,whose intentions towards the infant or child are not be-nign, epistemic mistrust becomes entrenched as an ap-propriate adaptation that has been prepared by naturalselection.Consistent with these assumptions, an accumulating

body of evidence indicates that childhood maltreatment,broadly defined, can have a negative impact on severalaspects of social-cognitive competencies in individualswho have not yet been explicitly diagnosed with a men-tal disorder [14–17]. Young maltreated children displayimpairments with regard to several indices of mentaliz-ing: (a) they engage in less symbolic and less child-initiated dyadic play [18, 19]; (b) they sometimes fail toshow empathy when witnessing distress in other chil-dren [20]; (c) they have poor affect regulation, whichcontributes to psychopathology and peer rejection inlater life [21–24]; importantly, (d) they make fewer refer-ences to their internal states [25]; and (e) they struggleto understand emotional expressions, particularly facialexpressions [26, 27]; this latter feature has been observedeven in studies that controlled for verbal IQ [28, 29].


The impact of maltreatment reaches into adulthood. Alarge-scale study of 5000 adults [30] found that maltreat-ment by parents in childhood was strongly associatedwith adult variations in theory of mind, or mental-stateinferencing, as well as self-reported levels of social affili-ation (social motivation and social support). Interest-ingly, this study found that face discrimination and facememory abilities in adulthood were relatively unaffectedby early adversity. The findings confirm that social cog-nition may be the domain that it is particularly vulner-able to the effects of adverse childhood environments.Impairments in epistemic trust are a further, and per-

haps more damaging, long-term sequel of the experienceof childhood maltreatment. Epistemic hypervigilance canmanifest as the overinterpretation of motives, which cantake the form of hypermentalizing [31, 32], or pseudo-mentalizing [33]. There is significant evidence that thequality of the relationship of a child to a given commu-nicator determines the extent to which they acquire andgeneralize information from that communicator [34–36].When in a state of epistemic hypervigilance, the recipi-ent of a communication assumes that the communica-tor’s intentions are other than those declared, and theinformation is therefore not treated as being from a def-erential source. Most typically, epistemic mistrust mani-fests as the misattribution of intention and theassumption of malevolent motives behind another per-son’s actions, and therefore treating them with epistemichypervigilance (or conversely, in some instances, exces-sive inappropriate epistemic trust). There is evidence tosuggest that a hypermentalizing stance is more charac-teristic of BPD in adolescence [31, 32]. It is possible thatthis hypermentalizing typically subsides into a flatterprofile of outright epistemic mistrust as the individualmatures. We speculate that this pattern may partially ac-count for the common life-course history of BPD symp-toms, which demonstrates a reduction in impulsivesymptoms over time but no lessening of the affectiveand social symptoms associated with BPD.In a state of epistemic mistrust, the recipient of social

communication may well understand what is beingexpressed to him/her, but he/she cannot encode it asrelevant, internalize it, and appropriately reapply it. Theconsequence is that the regular process of modifyingone’s stable beliefs about the world in response to socialcommunication is closed down or disrupted. This gener-ates the quality of rigidity and being ‘hard to reach’ thattherapists have often described in their work in the fieldof PD [37]. Change cannot happen in the therapeuticsetting because, although the patient can hear andunderstand the communications transmitted to them bythe therapist, the information cannot be accepted asrelevant to them and generalizable to other social con-texts. The persistent distress and social dysfunction

associated with PDs is the result of the destruction ofepistemic trust in social knowledge of most kinds.PD may therefore be best understood as a failure of

communication arising from a breakdown in the cap-acity to forge learning relationships. We believe thatthis quality underlies the painful sense of isolationthat characterizes the subjective experience of anindividual with BPD.

Reconsidering the role of attachmentThe change of emphasis in relation to the role of attach-ment theory in the aetiology of PD we will consider inthis section speaks to some of the long-standing criti-cisms of attachment theory that emerged from twodirections: psychoanalysis and anthropology. The psy-choanalytic criticism of attachment has tended to takethe position that attachment theory is too mechanisticand reductionist; that its broad classifications leaveattachment unable to engage with the subtlety and com-plexity of individual human subjectivity. These argu-ments have been well rehearsed [38, 39]. Meanwhile,anthropologists have suggested that attachment theorydisallows other kinds of complexity: those that arrivefrom cultural differences and varying environmental im-peratives. Varying contexts might indeed generate differ-ent family configurations and caregiving expectationsand structure, for example, alloparenting [40]. As an-other example, the fluid capacity of caregivers to attach,disengage and reattach across their lives has been com-pellingly described by the anthropologist Scheper-Hughes in her work on mothering in an acutely impo-verished milieu, where she observed mothers facing thedeath of their infants with apparently little sorrow, butbecome loving mothers to subsequent children or chil-dren who, having previously been given up on, went onto survive [41]. Similarly, historians have traced highrates of infanticide in many cultures (30–40% in early19th century Milan, for example [42]). Indeed, early his-torians of childhood, such as Philippe Ariès [43] andLawrence Stone [44], characterized it as a state of unre-mitting abuse and brutality. Stone argued that the highlevels of infant and child mortality in the pre-industrialera precluded the investment of love and affection inchildren that we would now consider normative [44].More recently, this depiction of the experiences of chil-dren in the past has been displaced by a more subtleand complex portrait of how parents have historicallyperceived and related to their children [45]. Ample ex-amples have been found of the ways in which childrenwere recognized, loved, protected and mourned for bytheir caregivers (e.g. [46, 47]). These academic skir-mishes over the sameness and difference of being aparent and a child across time, and the co-existence oflove and violence in human experience, should not


surprise us from a clinical point of view: they are inkeeping with our understanding of attachment as a uni-versal human (and indeed mammalian) instinct, whilestill allowing us to recognize, for example, the high ratesof infanticide that historians have traced in some periods[42]. In all but the most cases extreme childrearingscenarios, attachments of some style do form; but it ispossible that different social environments are likely totrigger different attachment styles as being more adap-tive to each environment.The attachment style to which the child is exposed

may be protective of the child, even if it is harsh orcruel. We thus suggest that attachment styles are them-selves one piece of social communication that the famil-ial context is promoting about the most effective way tofunction in the prevailing culture. Attachment is part ofa social signalling system telling the infant or youngchild to prioritize developing specific mentalizing capaci-ties and particular patterns of behaviour. The familyenvironment associated with BPD may entail triggeringa particular style of adaptation to ensure survival toreproduction, albeit one that causes pain to the individ-ual and is challenging to the immediately surroundingenvironment. For example, risky sexual behaviour in ad-olescents with a childhood history of neglect may be away of increasing the likelihood that they will contributeto the gene pool. Such behaviours are resistant to changebecause the adaptation is triggered by natural selection;the individual’s genes ‘communicate’ that this is mostlikely to ensure survival (of the genome) [48]. Lowerlevels of mentalizing, greater aggressiveness and highersensitivity to perceived threats may be adaptive re-sponses to certain cultural environments. Natural selec-tion has charged families with psychologicallyenculturating their children to maximize their likelihoodof survival. Social learning from the immediate familyand culture can help us account for the relationship be-tween individual behaviours and the culture that engen-ders them. Low levels of interpersonal understanding, oreven frank attacks on the self-awareness of individualfamily members, may be biologically successful, evolu-tionarily selected strategies. A stance of dismissing at-tachment and non-mentalizing is not experienced as adeficit by the person adopting this stance, but rather asthe most appropriate strategy to ensure their survival. Itfurther follows that if mentalizing interventions are tosucceed with children, they need to occur in the contextof the family [33] and enhance the quality of mentalizingwithin the family system to which children are orientedto acquire social expectations.At a theoretical level, this change in focus involves a

certain reconfiguration of the role of attachment in de-velopmental psychopathology. Like other authors [49],we have previously placed considerable weight on the

nature of attachment disorganization in our accounts ofBPD based on the mentalizing model [50]. We maintainthat the role of attachment is highly significant in thedevelopmental origins of PD. However, we argue that itsrole might perhaps be best understood as only one(albeit very important) form of content learned from thesocial environment. This is congruent with recent worksuggesting that the relationship between infant attach-ment status and later outcomes is more complicatedthan that suggested by early attachment studies [51].Other findings have suggested limited evidence for link-ing childrearing environments to later outcomes and thefluctuating significance of infant attachment style acrossthe life trajectory. For example, in infancy, the role ofgenes in determining security or insecurity of attach-ment is negligible and the childrearing environment iscritical [52]; however, in adolescence, the impact of gen-etic factors rises considerably, such that they predict 38and 35% of security and insecurity, respectively [51].Meanwhile, parental sensitivity – previously consideredkey for the transmission of attachment security in in-fancy (see a major meta-analysis by Verhage et al. [53])– may have other functions beyond ensuring secure at-tachment, although this function is, of course, an im-portant one. The relationship between parentalsensitivity and developmental outcomes, according to re-cent and highly compelling findings by Kok et al. [54],may be more general and structural than can be cap-tured by infant attachment status: these findings indicatethat normal variation in maternal sensitivity is related tomarkers of optimal brain development. This suggeststhat the parenting environment supports the neurobio-logical architecture of higher-order cognitive functionupon which the capacity to mentalize depends.We suggest that the relationship between parental sen-

sitivity, attachment and epistemic trust lies in the way inwhich epistemic trust in most normal circumstances de-velops in the context of attachment relationships. Secureattachment, which provides mostly consistent contingentparental responses to the child, also provides mostlyconsistent ostensive cueing and therefore the most fer-tile ground in which epistemic trust can emerge andsubsequently generalize to new relationships. This, ofcourse, follows Bowlby’s description of internal workingmodels [55]. Attachment to a safe, sufficiently reliableand mentalizing caregiver provides the child with a senseof agency that allows the child to have some confidenceboth in their own interpretation of the social world, andin the good faith and general accuracy of their care-givers’ communications [56].The role of attachment in our conception of personal-

ity has shifted as we have increasingly come to regardthe conceptualization of linear causation in psychopath-ology as unhelpful; instead, we conceptualize the


perpetuation of PD being driven by loosely coupledinteracting systems working in a circular way. A linearapproach would posit that the capacity for mentalizing isvulnerable because of the social-emotional quality ofearly attachment experiences; partial, erratic mentalizingturns into an interpersonal vulnerability whereby a per-son feels interpersonally brittle because they cannot reli-ably process the psychological meaning of socialexperience, and vulnerable because they cannot processtheir own emotional reactions to these experiences.Evidence suggests that attachment stress derails men-

talizing judgments [57]; working in the other direction,attachment schemas predict mentalizing in adolescence[58, 59]. According to this model, mentalizing and emo-tional regulation compete, and attachment insecurity hasa catalytic role in disrupting the development of optimalmentalizing capacity.Mentalizing difficulties lead to affect dysregulation,

which in turn further disrupts mentalizing. Whereverthis cycle starts, mentalizing problems lead to interper-sonal conflict and social difficulties, which generate in-tense (social) affect such as shame, which is inadequatelycontextualized because of the failure of social cognition.This affect further undermines the capacity to mentalize,which can then create further social challenges, generat-ing interpersonal conflict that will inevitably lead tohigher emotional arousal. The emotional arousal ispoorly modulated and causes further disruptions of so-cial cognition as part of a recursive process, the finaloutcome of which is an individual lacking the higher-order cognitive capacity necessary to withstand eveneveryday social adversity.The likely interaction between a history of adversity

that challenges epistemic trust and mentalizing failure asboth a cause and a consequence of emotion dysregula-tion culminates in a stance where the individual withlimited mentalizing capacity cannot reliably detect os-tensive cueing and adopts what is perceived to be a mal-adaptive pattern of rigidity – that is, inability to change.What emerges is an (implicit) attitude of mistrust in thesocial environment [60] and an incapacity to learn fromsocial experience or to modify one’s behaviour on thebasis of social learning. In our view, these individuals arethose with high ‘p’ scores whose disorders persist be-cause of their inaccessibility to normalizing social influ-ence. Their ‘impermeability’ to therapeutic influencecomes not from the deep-seatedness of the pattern butits central manifestation of epistemic mistrust born of adual core of a history of adversity and emotionally dis-rupted sensitivity to ostension. This is not a naive envir-onmental theory promoting the quality of socialinteraction at the expense of biological factors: there isevery reason to suspect that genetic predisposition, aswell as the normal mixture of early environmental

determinants, makes an individual more or less receptiveto ostensive cues. The fact that therapeutic interventionshave the capacity to promote sensitivity to ostensivecues in no way prejudges the balance of biological versuspsychosocial influences on sensitivity to social cues.Because clinicians have historically linked non-responsiveness to therapeutic intervention to character-istics of their patient rather than features of their ownrelationship to the patient, the pattern of epistemic mis-trust/hypervigilance was regarded as a feature of themost stable system they could identify in their patient –their personality. As ‘normal’ personality is in fact farfrom stable, consistent or unmalleable in relation to so-cial situations [61, 62], perhaps disorders of personalityare so called because, unlike normal personality, individ-uals with PDs have in common an absence of flexibilityand great difficulty in adapting to changing social situa-tions. Hence, epistemic mistrust may have its roots inpart in disturbed attachment experiences, but ultimatelyit is a disorder of social communication or social learn-ing. Its core is a compromised capacity for appropriatelyinterpreting social actions in terms of mental states,which is what normally bolsters resilience, leaving theindividual with dysfunctional social learning systems thatare inadequate to assure adaptation in the face of ‘nor-mal’ adversity.Although this perspective has considerable bearing on

our understanding of the subjective experience of BPD,it is also one that is consistent with a conceptualizationof the human mind as having evolved to be highly socialand culturally responsive. Therefore, it is a theory that isrelevant to how we think about the relationship betweenthe individual and culture, and it is of relevance to amuch broader and more interdisciplinary way of think-ing than our previous position was. This rather moresystemic, less intrapsychic approach involves a reposi-tioning of the role of attachment in developmental psy-chopathology to accommodate the imperatives of thewider social environment within which the dyadic rela-tionship is located. The anthropologist Thomas Weisnerexpressed it thus:

The question that is important for many, if notmost, parents and communities is not, “Is [thisindividual] child ‘securely attached?’” but rather,“How can I ensure that my child knows whomto trust and how to share appropriate socialconnections to others? How can I be sure my childis with others and situations where he or she willbe safe.” Parents are concerned that the childlearns culturally appropriate social behaviours thatdisplay proper social and emotional comportmentand also show trust in appropriate other people.([63], p. 263)


Our thinking has – albeit from a different direction –come to a similar conclusion.

The role of systemsIf the lack of resilience we associate with BPD is to beunderstood as an inability to access positive appraisaland the inhibition mechanisms owing to imbalances inmentalizing and the associated compromise of epistemictrust, this also has implications for the system inhabitedby that individual. As outlined earlier, we suggest that‘personality’ dysfunction persists through the self-perpetuating cycle of social dysfunction and mentalizingdifficulties. The resulting heightened affect disrupts theinterpersonal environment, creating social challengesthat derail mentalizing and in turn undermine socialfunctioning.A graphical display may help to illustrate these com-

plex interactions (see Fig. 1). Emotion dysregulation,disrupted attachment histories and the disorganizedinsecure attachment system interact to generate social/interpersonal dysfunction, a shared characteristic of PDs[64, 65]. Such dysfunctions are best understood ascommunication failures rather than as properties orcharacteristics of the individual suffering from PD.The failure of communication occurs at a number of

levels. First, the social disruption associated with inter-personal conflict will itself compromise the processes ofsocial learning and, in particular, of salutogenesis (thecapacity to learn and benefit from the (social) environ-ment). This is a systemic failure of communication thatmay characterize a family, the members of a social groupsuch as a gang, a social subculture, or indeed an entire

culture. We will discuss such systemic failures in moredetail below in terms of their impact on the network ofsocial influence within which all socialization occurs.Second, the loss of balanced mentalizing triggered by

interpersonal conflict generally lessens interest in thecontent of communication and social informationexchange. There is a pervasive loss of interest inintentionality; observable outcomes are gradually priori-tized as indicators of attitudes and the general tenorof verbal communication is perceived as meaningless‘psychobabble’ with few or no substantive implicationsfor the life of the individual.Third, social dysfunction, as well as the misinterpret-

ation of social signals associated with the loss of menta-lizing, leads to a probable failure to appropriatelyidentify ostension – the sense that a communication isof personal relevance.These factors (and probably many others) contribute

to the individual’s failure to develop epistemic trust inpersonally relevant communications. Again, we are keento point out that this is not inherently a maladaptiveprocess. The failure to develop epistemic trust leaves thenatural function of epistemic vigilance in place. It is infact an efficient adaptation and an indication that theindividual is exercising appropriate caution in relation tosocial influence, which we see as manifesting in theundesirable persistence of antisocial expectations orschemata and the individual’s relative imperviousness tosocial influence.However, the absence of epistemic trust sets a limit

upon social learning. This can render the individualincreasingly ill-suited to function effectively within their

Fig. 1 The Natural Pedagogy Model of Personality Disorder. Illustrates the interactions between social dysfunction, failure in social communication,epistemic mistrust, and imperviousness to social influence that underpin personality disorder. Emotion dysregulation, disrupted attachment historiesand the disorganized insecure attachment system generate social/interpersonal dysfunction. This undermines accurate social communication, causingsocial disruption, the misinterpretation of social signals, and difficulty in recognising ostensive cues from others. These difficulties in the area of socialcommunication can give rise to epistemic mistrust in relation to the social environment. This is not inherently a maladaptive process: epistemicvigilance has a natural function. However, the absence of epistemic trust sets a limit upon social learning. This can render the individual potentiallyunable to function effectively within their social environment and can lead to further disruption in the social network, leaving the individualincreasingly isolated and prone to further social/interpersonal dysfunction


social environment. Disruption of the social networkwithin which the individual could (or perhaps should)function leaves them increasingly isolated and prone tofurther social/interpersonal dysfunction.There are many levels at which systemic thinking ap-

plies to how we respond to PD. In terms of clinicalwork, a mentalizing team around the therapist is, weargue, essential for maintaining good practice. In thecontext of the persistent distress associated with PD,clinical encounters happen, by necessity, against thebackground of constant exposure to psychic equivalenceand pretend or teleological modes [33, 66]. We suggestthat it is the impact of non-mentalizing on the system ofsocial communication, and not the unchangeability ofnon-mentalizing per se, that makes PDs clinically chal-lenging conditions. One of the defining characteristics ofPD is that the patterns of social dysfunction shown bythe patient are enduring. Indeed, as mentioned above,BPD in particular has traditionally been regarded as analmost untreatable condition; this is one of the factorsthat have contributed to the stigma experienced by thosereceiving a PD diagnosis. However, effective therapiesfor BPD now exist: at least nine forms of treatment havebeen tested in at least 20 randomized controlled trials[67], and patients with BPD should no longer beregarded as ‘unhelpable’. We would argue that theapparent inconsistency that a condition has long beenbelieved to be untreatable, yet appears to be moreresponsive to therapy than most mental disorders, is tobe found to lie in the way the non-mentalizing actionsof BPD patients can create non-mentalizing socialsystems that sustain their condition – including in theconsulting room. We suggest that it is unrealistic toexpect a clinician working with such patients to themselfmaintain an effective mentalizing stance in the mediumto long term if they are not supported adequately tomaintain their capacity to mentalize, ideally by a sur-rounding team that is not directly exposed to (and is thusprotected from) the patient’s dysfunctional social system.Systemic interventions may be required to address

these problems [68]. In principle, the patient and therap-ist are isolated in a room, albeit with bidirectional socialinfluence – the therapist is, after all, in a position toenhance the patient’s capacity to reflect, to question andto focus simultaneously on both other and self, insideand outside. But the reality is that the therapist becomesembedded within the patient’s social survival mechan-ism, which subsumes the obliteration of balancedmentalizing (normally erring on the side of being unre-flective, externally focused, emotional and dominated byresonance rather than reflectiveness). The clinician’smentalizing, even if exceptional, is unlikely to be suffi-cient to be able to deal with such highly intense emo-tional situations and conflicts. Therapists require their

own system of support relationships, primarily fromother clinicians, in order to scaffold their capacity tomentalize and facilitate epistemic trust.The self-perpetuating cycle of sustained dysfunction

associated with BPD and a non-mentalizing social sys-tem reminds us of the international variability in theprevalence of BPD. It has been observed that BPD is lesscommon in non-Western societies, possibly as a resultof the fact that the lack of social capital and communitysupport characteristic of many modern or modernizingsocieties leaves the individual more vulnerable to impul-sivity and affective instability [69]. Available prevalencedata suggest that Western countries with higher levels ofinequality of wealth experience higher rates of BPD [70].The anomie of modern life – that is, a lack of socialconnectiveness leading to dysregulation –described byDurkheim [71], and connected by other authors with theconditions that might account for national variations inBPD [69], can be read as a description of a systemiccollapse of epistemic trust. This emphasis on the role ofthe social environment points to the value of thinkingabout ways in which a social climate can be encouragedto become more mentalizing to support a changeprocess. Families are one obvious example of a systemicarena for the promotion of mentalizing that reinforcesthe learning of epistemic trust. Bateman and colleagues’development of the Families and Carers Training andSupport programme (FACTS) for those supporting afamily member with BPD is one example of a mentaliz-ing intervention for the family [72].The school is another system that seems ideal as the

site for mentalizing interventions. Tellingly, evidencesuggests that, of the many interventions that now existto deal with bullying in schools, the most effective sharethe characteristic of involving the whole school [73]. Amentalizing-based approach, known as Creating aPeaceful School Environment (CAPSLE), is one ofthree bullying prevention strategies found by a largemeta-analysis to be most effective [74] (the other twoprogrammes were the Olweus Bully Prevention Pro-gram, whose generalizability has recently been ques-tioned by Bradshaw [75], and Finland’s KiVa nationalanti-bullying program [76]). The mentalizing approachof CAPSLE is a systemic one, which seeks to create amentalizing climate and a group dynamic that can re-sist and limit the potency and currency carried by theindividual acts of violence or aggression that are inev-itable in a school [77–80].AMBIT (adaptive mentalization-based integrative ther-

apy) is a third example of a clinical approach that com-bines mentalizing with thinking about the systems thatsurround an individual [81, 82]. Originally developed for‘hard-to-reach’ adolescents with complex needs, AMBITis now being applied to younger and older client groups.


Such clients present with multifaceted difficulties andso tend to attract complicated multi-agency andmulti-professional networks aiming to provide help.At the same time, these clients tend to be highlyalienated from conventional social networks, whileoften forming personal relationships that carry furtherrisks. AMBIT seeks to counter these difficulties byusing a main keyworker to, where possible, simplifythe individual’s experience of the complex networkthat surrounds them. The keyworker simultaneouslyseeks to support and encourage the non-professionalsocial networks that surround the individual (e.g. thefamily, friendship groups or extra-curricular/activity-based groups), while also serving as a secure attach-ment base from which the individual might explorethe social opportunities their environment presents. Afinal crucial component of this approach is its em-phasis on the need for a supportive mentalizing sys-tem around the keyworker, given the anxieties andpressures involved in such therapeutic work.The systemic mentalization-based interventions out-

lined above have in common their view of the individualas being temporarily separated from their social net-work, and of their capacity to form bonds of trust beingshaky and prone to disruption. Without intervention,the person loses their epistemic safety net; the sociallydefined network of meanings is under threat. The inter-ventions address the network, not just the individual orthe therapist. In AMBIT, the links between the keywor-ker and the ‘dis-integrating’ (the term used in AMBIT toindicate the frequency with which the various agenciesaround a client may pull in opposite directions in theirvarious attempts to work with the client) social care sys-tem around the family are an important focus. InCAPSLE, the non-mentalizing bully–victim–bystander isfocused on by everyone within the whole school. FACTSaims to address the non-mentalizing within the familysystem. Common to each of these approaches is its cap-acity to ensure that epistemic trust – the meaningfultransfer of information from one person to the other –is ultimately assured and protected. It is evident inCAPSLE where the disruption of epistemic functionmakes the intervention necessary; indeed, one of theoutcome measures for this intervention is the improve-ment of children’s scores in standardized assessments ofeducational attainment [79]. In AMBIT, meaningfulcommunication between different helping systems isresumed with the restoration of mentalizing. Similarly,in FACTS, with improved mentalizing the family canonce again take up its function of social informationtransmission. It is in our opinion thus not mentalizingitself that is of direct benefit; it is the normal socialfunctions that depend on mentalizing that bring the realtherapeutic benefit.

Non-mentalizing social systems present a powerfulcue that the individual is in an environment wheresocial relations are not operating on the principle ofshared goals, cooperation and interdependence. It isthese behavioural imperatives that are, as Tomasellodescribed, associated with our higher-order cognitivecapacities [83]. When presented with cues that signifythat we do not have access to collaborative social re-lations, we make cognitive adjustments, as evidencedby new research on Social Baseline Theory [84]. As asimple illustration, hills are judged to be less steepwhen one is standing next to a friend, and there is adose–response effect: the longer the friendship, theless steep the hill appears to be [84, 85]. Coan et al.state that ‘The human brain expects access to rela-tionships characterized by interdependence, sharedgoals, and joint attention’ ([84], p. 87). Violations ofthis increase stress and increase cognitive and physio-logical effort – ‘social relationships decrease the pre-dicted cost of the environment’ ([84], p. 87). Socialbehaviour is so closely at the heart of the human evo-lutionary story that it is a fundamental instrumentthat humans use to ‘mitigate risk and diminish thelevel of effort needed to accomplish goals’ ([84], p.87). In the absence of this social baseline, the envir-onment is perceived to be more risky and costly interms of effort. The accessibility of social support isone of the factors that humans – and other social an-imals – use in adjusting their appraisal bias.Literature relating to research in non-human ani-

mals shows that the capacity of an organism to regu-late its internal state according to evaluations of theexternal conditions (rather than through basic stimu-lus–response mechanisms) is fundamental to behav-ioural flexibility; it has been recently suggested thatappraisal theory can be fruitfully brought into thisthinking [86]. In particular, it has been suggested thatcognitive biases arising from the interference ofaffective states, as well as genetic and environmentalfactors, can affect the appraisal of ambiguous situa-tions, which subsequently shapes resilience to stress-ful events [86]. One example is Harding andcolleagues’ classic finding that rats exposed to un-stable housing conditions made more pessimisticevaluations of ambiguous stimuli, in a way that issimilar to how anxious or depressed people tend tomake negative judgments about ambiguous stimuli[87]. Whereas previously, as attachment theorists, wemay have made sense of the relationship between be-havioural flexibility, social stimuli and appraisal interms of internal working models, we now suggestthat epistemic trust is the mechanism via whichhumans’ behavioural flexibility arising from appraisalsbecomes compromised.


Implications for prevention and interventionDifferent approaches to BPD from a theoretical andpractical point of view appear to be embarrassinglysimilar in terms of outcome [88, 89] in BPD. Basedon the considerations outlined in this paper, we sug-gest that all effective treatments of BPD involve thesame structure, namely that the re-emergence ofepistemic trust requires three initially sequentiallyimplemented but, as treatments unfold, increasinglyconcurrent forms of communication.

Communication system 1This entails the communication of therapeuticmodel-based content that indicates to the patientthat the therapist has considerable knowledge as wellas personal characteristics that may be highly valuedby the patient. The knowledge communicated willnaturally vary according to the treatment model (e.g.Transference Focused Psychotherapy will communi-cate information about primarily subtle intrapsychicrelationships, while Dialectical Behavior Therapy willoffer broader psychological constructs and copingstrategies). Content analysis of all effective treatmentsreveal that the relationship of therapist and patient issupported by the former conveying a convincing un-derstanding of the patient as an intentional agentwhich generates a sense of self-recognition. Allevidence-based models of psychotherapy presentmodels of mind, disorder and change that are accur-ate, helpful to patients and increase patients’ capacityfor understanding. However, they also need to over-come the epistemic hypervigilance (‘not true’, ‘notrelevant to me’) presented by the patient. So, besidesthe content, this stage involves a subtle and richprocess of ostensive cueing. Thus, even at this rela-tively early stage the therapist must present theirinformation with mentalizing in mind, establishing collab-oration with the patient, demonstrating that they see thepatient’s problems from their perspective, recognizingthem as an agent, and with the attitude that the patienthas things to teach the therapist. Through this, the therap-ist responds contingently to the patient. From the struc-tural perspective we are presenting here, the therapist’sattempt to apply his/her model to interactions with thepatient serves as an ostensive cue, which increases the pa-tient’s epistemic trust and thus acts as a catalyst for thera-peutic success. It does so to the extent that (a) thetherapist is able to find and effectively transmit contentthat provides valuable ways for the patient to understand(mentalize) themselves and their reaction to others, and(b) the process of transmission involves the patient recog-nizing the truth and personal relevance of the content, sothey become able to relax their epistemic mistrust.

Communication system 2Mentalizing may be a common factor in effective psycho-therapies, but not in the sense that we originally intended[90]. It is not that, regardless of the therapeutic model, pa-tients learn the ‘Esperanto’ of mentalizing, or even thealtogether more appealing discourse of ‘plain old therapy’[91]. The constant engagement of the patient by the ther-apist has several key features that are relevant to the res-toration of epistemic trust. First, the therapist consistentlyrecognizes the patient’s agentiveness, focuses on him/heras an actor and negotiates from the perspective of the pa-tient’s self. Second, by marking the patient’s experiences,the therapist acknowledges the patient’s emotional state.Third, the therapist makes extensive use of ostensive cuesto denote the personal relevance of the information trans-mitted and its generalizable social value. By mentalizingthe patient effectively, the therapist models mentalizing,creating an open and trustworthy environment with lowarousal. Structurally, a ‘virtuous cycle’ is put into motion:the therapist responds sensitively to the patient, the pa-tient takes a step back from epistemic isolation, and thepatient gradually begins to exercise his/her mentalizingskills, which, step by step, extend from the confines of thetherapeutic context and generalize to his/her wider socialcontext. This elicits an emotional reaction by the patientto the social context, giving the therapist further oppor-tunity to respond sensitively. This process involves a com-plex and non-linear progression. Improving mentalizing isnot its main goal, but the improved mentalizing that re-sults from it enables the patient to start to approach andlearn from their wider social context. Answering the ques-tion of why patients with a better capacity for mentalizingimprove more in psychotherapy than those whose menta-lizing is poorer helps us to understand the process.Mentalizing moderates the impact of therapeutic commu-nications: a poorly mentalizing patient will frequently in-terpret the therapist’s ostensive cues erroneously, andepistemic trust is thus not established. With improvedmentalizing, the therapist’s communications are appreci-ated and interpreted as trustworthy – and have theintended influence on the patient. The experience ofhaving one’s subjectivity understood – of being mentalized– is a necessary trigger for being able to receive and learnfrom the social knowledge that has the potential to changeone’s perception of oneself and the social world. The ‘gift’of a mentalizing process in psychotherapy is to open up orrestore the patient’s receptivity to broader social influence,which is a precondition for social learning and healthy de-velopment at any age.

Communication system 3The greatest benefit from a therapeutic relationshipcomes from the generalizing of epistemic trust beyondtherapy, such that the patient can continue to learn and


grow from other relationships. Social learning in thecontext of epistemic trust is (re)established, and thisleads to salutogenesis. The third communication systemis a process of opening the person’s mind via establishingepistemic trust (collaboration) so he/she can once againtrust the social world by changing his/her expectationsof it. This means that it is not just what is taught intherapy that helps the patient, but that the patient’s cap-acity for learning from social situations is rekindled.Enhanced mentalizing allows the patient to achieveimproved social relationships and recognize who is areliable and trustworthy source of information – that is,who one can ‘be friends with’. The improved epistemictrust and abandonment of rigidity enables learning fromexperience once again. So, therapeutic change is prob-ably a consequence of how the patient comes to usetheir social environment, and not to what happens intherapy per se. The benefits of therapy remain contin-gent on what is accessible to patients in their particularsocial world. Therapeutic interventions are effectivebecause they open the patient to social learning experi-ences which feed back in a virtuous cycle. If the environ-ment is at least partly benign, therapy will ‘work’.This third system – social learning in the context of

epistemic trust – is, according to our thinking, themechanism at work in the circular and self-perpetuatingrelationship between BPD and the social context. Theconceptualization of the three communication systemsoutlined here involves an acknowledgment of the inher-ent limitations of clinical interventions in cases wherethe patient’s wider social environment does not supportmentalizing. The implication of this is that what happensin any therapeutic intervention cannot on its own beexpected to be enough to bring about any lasting signifi-cant improvement in the patient’s state. Indeed, incertain circumstances it would be maladaptive for theindividual to develop epistemic trust and lower theirsocial defences – for instance, in social environmentscharacterized by high levels of aggression or violence, inwhich an external, non-reflective, rapidly respondingaffective focus on others as opposed to the self would bebetter prioritized as a survival strategy.

ConclusionsSeveral features of the theoretical approach presented inthis paper await further empirical confirmation, but ac-cording to the theory of epistemic trust and social learn-ing, the lack of resilience, or positive appraisal,characteristic of individuals with BPD may be, in a sense,mislabelling. It may be more accurate to characterizeBPD as an ‘emergency mode’ form of social understand-ing in which epistemic hypervigilance, distrust, or out-right epistemic freezing is an adaptive consequence tothe individual’s social environment. For various possible

reasons, the individual has adopted negative appraisalmechanisms as a default. This is a highly socially ori-ented perspective on personal psychopathology. The keyargument is that BPD (or other manifestations of the ab-sence of psychological resilience) is the outcome of theways in which the individual has learned to respond tothe transmission of social knowledge within their ownsocial environment.Future research is needed to investigate these assump-

tions in more detail. This may also lead to the develop-ment of new prevention and intervention strategies,which are urgently needed, particularly given theincreasing recognition of the need for prevention strat-egies for BPD [92, 93].

AbbreviationsAMBIT: Adaptive mentalization-based integrative therapy; BPD: Borderlinepersonality disorder; CAPSLE: Creating a Peaceful School LearningEnvironment; FACTS: Families and Carers Training and Support programme;PD: Personality disorder

AcknowledgementsWe are indebted to many colleagues for inspiring and contributing tothe ideas presented in this paper: Professor Eia Asen, Professor AnthonyBateman, Dr Dickon Bevington, Professor Pasco Fearon, Professor GeorgeGergely, Professor Alessandra Lemma, Dr Trudie Roussow and ProfessorMary Target.

FundingPF is in receipt of a National Institute for Health Research (NIHR) SeniorInvestigator Award (NF-SI-0514-10157). PF was in part supported by theNIHR Collaboration for Leadership in Applied Health Research and Care(CLAHRC) North Thames at Barts Health NHS Trust. The views expressedare those of the authors and not necessarily those of the NHS, the NIHRor the Department of Health.

Availability of data and materialsNot applicable.

Authors’ contributionsAll the authors contributed equally to the writing of this article. All authorsread and approved the final manuscript.

Competing interestsPF developed and currently trains for mentalization-based treatment; theproceeds of these trainings are paid to the Anna Freud National Centre forChildren and Families, of which he is Chief Executive.

Consent for publicationNot applicable.

Ethics approval and consent to participateNot applicable.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Author details1Research Department of Clinical, Educational and Health Psychology,University College London, London, UK. 2Faculty of Psychology andEducational Sciences, KU Leuven, Leuven, Belgium.


"This course was developed and edited from the open access article: What we have changed our minds about:

Part 1. Borderline personality disorder as a limitation of resilience, Fonagy et al., Borderline Personality

Disorder and Emotion Dysregulation (2017) 4:11 (DOI: 10.1186/s40479-017-0061-9), used under the Creative

Commons Attribution License.”

"This course was developed and edited from the open access article: What we have changed our minds about:

Part 2. Borderline personality disorder, epistemic trust and the developmental significance of social

communication, Fonagy et al., Borderline Personality Disorder and Emotion Dysregulation (2017) 4:9 (DOI:

10.1186/s40479-017-0062-8), used under the Creative Commons Attribution License.”

significant development changes in bipolar disorder...of both adult and child psychopathology and...

Documents