Show Summary Details
Page of

(p. 49) Mentalizing in development 

(p. 49) Mentalizing in development
(p. 49) Mentalizing in development

Robbie Duschinsky

and Sarah Foster

Page of

PRINTED FROM OXFORD CLINICAL PSYCHOLOGY ONLINE ( © Oxford University Press, 2021. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Clinical Psychology Online for personal use (for details see Privacy Policy and Legal Notice).

date: 06 December 2021


Having situated Fonagy in biographical and institutional context in Chapters 1 and 2, we can now turn more directly to the development of the theory of mentalizing. This theory has undergone very substantial changes over time, and with these changes have come alterations in and additions to the meaning of the term ‘mentalization’. In their 2003 book, Psychoanalytic Theories, Fonagy and Target observed critically that it is quite characteristic of psychological theories to have a primary concept or two with a host of meanings. This concept then serves in part as a symbol of collective endeavour: ‘Many concepts referred to theoretically have multiple references, some pertaining to the developmental course, some to covert mental states and some to manifest presentation.’ Fonagy and Target highlighted that there are advantages to such a symbol, ‘particularly through enhancing a professional group identity by enabling members to believe that they share ideas, though ‘in the long term such fuzziness impedes progress, and scientific debate is degraded’.1 The concept of ‘mentalization’ has also been subject to such a process, as Fonagy and colleagues have themselves increasingly acknowledged over the past decade.2

The term ‘mentalization’ is an umbrella term that has encompassed diverse meanings and investments. It offers the feeling of coherence that facilitates community building, while also serving as a switch and relay between research, health policy, and clinical practice, as well as between different clinical specialisms.3 At times, the term has been yoked to specific definitions or operationalization in psychological constructs, though these explicit definitions have seen very significant changes over time (see Chapter 4). At other times, ‘mentalization’ has functioned as a buzzword, used to mean little more than desirable psychological functioning,4 or a process that is in some way psychological.5 The concept of mentalization (p. 50) appears as an inspiring ideal, and one with multiple possible uses. However, this picture has been achieved in part by punching holes in the concept’s ability to hold determinate meaning and facilitate clear communication across domains. This predicament has made discrepant research findings difficult to interpret, because researchers may well be measuring different phenomena but nonetheless calling them ‘mentalization’.6 A few of Fonagy’s collaborators have even argued that the term ‘mentalization’ should not be used with patients, given its untamed panoply of semantic and theoretical investments.7 However, this is not the majority view.

To unpack the different meanings the concept has gained over time, it is necessary to take a developmental approach. The focus of the chapter will be work by Fonagy and colleagues addressed to the field of developmental psychology in the 1990s, though later theory and research will be mentioned when there have been continuities. We will begin by considering the emergence of the concept of mentalization, the problems it was introduced to address, the theoretical perspective it encapsulated, and the clinical implications that stemmed from this perspective. We will then examine the development of the reflective functioning scale resulting from Fonagy’s work with Miriam and Howard Steele, a very important development for work on mentalizing, as well as for the broader fields of developmental and clinical psychology. The chapter will close with an analysis of some ambiguities and problems in the use of the concept of disorganized attachment by Fonagy and colleagues.

Origins of the mentalization concept

As discussed in Chapter 1, the idea of a ‘borderline’ profile of mental illness had emerged in the twentieth century and gained prominence from entering the DSM-III in 1980 as one of the personality disorders. In the 1980s and 1990s, the wider clinical community regarded borderline personality disorder (BPD) as untreatable, even as a growing proportion of patients in NHS psychotherapy services were identified as having this profile.8 However, clinicians at the Anna Freud Centre regarded these cases as requiring an adjustment of clinical (p. 51) technique to emphasize understanding and support. Looking back, Bateman and Fonagy observed in 2003 that the diagnostic category of BPD:

has evoked intense theorizing among psychoanalysts, behaviourists, psychiatrists, and others, and, perhaps because of its clinical difficulty and variability, that it represents a battlefield on which many of the controversies and schisms are played out.9

In the DSM-III, ‘borderline’ was considered to be a personality disorder because it signalled widespread difficulties an individual faced in understanding and regulating their ‘self’ in relation to others. In the 1980s, it was assumed that pervasive psychosocial symptoms had to be located in the ‘personality’ of the patient.10 Running somewhat contrary to this focus on individual personality was growing concern in the period with the socially situated self in developmental psychology, with the solidification of research concerned with social cognition. This represented important acknowledgement of social and developmental contributions to the mind of the individual (albeit with only somewhat greater acknowledgement of the role of culture).11 Within both British and American developmental psychology in the 1980s, this shift was fed by several currents.

One was attachment theory: in the third book of his Attachment and Loss trilogy, Bowlby had reinterpreted individual differences in attachment as reflecting variation in the filtering and processing of social information about the availability of attachment figures.12 Research in American developmental science had also started to explore empirically how children develop an understanding of mental phenomena, conceptualized as the child’s ‘theory of mind’.13 These perspectives helped scaffold empirical research by scholars such as Bretherton and Dunn on communication within families, which explored how children and parents spoke about feeling states and intentions.14 A further current was the growing maturity of empirical research on autism. Researchers such as Baron-Cohen had drawn on the idea of ‘theory of mind’ to suggest that autism reflected differences and/or deficits in the capacity of an individual to understand the mental states of other people.15 Research on social cognition was diverse, but had in common that it questioned the assumption that human experience of self and others is prewired, universal, and equally accessible. Instead, it was proposed that such experience depends upon socio-emotional, communicative, and neurodevelopmental supports, which vary substantially between individuals.

(p. 52) Reflecting on clinical cases such as Mr S (Chapter 1), in the late 1980s and early 1990s, Fonagy found a ‘point of contact’ between the developmental theory of research in social cognition and the profile of symptoms characterized as BPD.16 His proposal was that the difficulties with identity, mood, empathy, and social relationships associated with BPD could be conceptualized as expressions of a single underlying problem: disruption in the capacity to process and reconsider information about mental states. In therapy, Fonagy’s patients with BPD described childhoods characterized by a lack of parental empathy, consistency, and/or safety. Fonagy knew, of course, that the recollections of adults in therapy needed to be treated with caution. He also knew that children’s experiences of their parents may be inflected by their hopes and worries, as well as by the adult’s actual behaviour.17

Nonetheless, he extrapolated that experiences that would make it difficult for children to tolerate thinking about the mental states of their parents would hinder the development of social understanding more generally. The basic underpinning mechanism, Fonagy anticipated, was the classic psychoanalytic premise of the ‘pleasure principle’. This principle supposes that children will avoid displeasure wherever possible. Fonagy hypothesized that children will turn away from thinking about mental states if these were sources of unpleasure or outright pain.18 Reflecting on the case of Mr S., Fonagy proposed that the most direct cause would be when parents were ‘unloving and cruel’, because this would make ‘the contemplation of the contents of the mind of the object unbearable’.19 His clinical work and review of existing retrospective research on BPD led him to conclude that sexual and physical abuse would make an especially potent contribution.20 However, he also suggested that lack of parental empathy, consistency, and the provision of safety might contribute through various means. For instance, he argued that a ‘totally vacuous relationship’ might be expected to have the same effect as an abusive or hostile one, in making mental states a source of unpleasure.21

(p. 53) During the 1990s, Fonagy was not sure whether trauma was the cause of the inhibition in social cognition, or a correlate; and whether the inhibition could occur in the absence of discrete trauma, as in the case of a ‘vacuous’ relationship. Poor specification of the concept of trauma also meant that he sometimes lost track of these questions altogether, especially when trauma was itself defined as the absence of mentalization.22 Nonetheless, in general, in this period he held that trauma in the early child-parent relationship would especially predispose a turning away from attention to mental states throughout development.23 However, he accepted that other aversive experiences of caregiving could have the same effect:

Inhibition of mental processes normally arises in the context of intense trauma, but this does not necessarily involve parental disturbance. Such inhibition does, however, require an internal experience that is sufficiently intense to make a permanent distortion of a basic mental capacity. This was the situation in the case which follows. Joan had had a bone disease from the age of 5 weeks which persisted throughout the first two years of her life. She suffered from pain which spoiled all the normal, pleasurable, physical interactions with her mother; being cuddled, washed, dressed, and fed all hurt. In addition her mother had to help in painful and frightening medical procedures. This resulted in a deeply ingrained view of mother as attacker and of herself as hurt, damaged, and deprived. The mother in turn felt helpless and inadequate and severe distortions of object representations were inevitable. Because the early mother child interactions were painful rather than pleasurable, they gave rise to severe inhibition and stunting of mental processes.24

Influenced by the tendency of psychoanalytic theories to speculatively assign the origin of particular mental health problems to a developmental stage of early life, Fonagy gave particular prominence to the second year of life. He argued that children of this age would have the cognitive maturity to be capable of inhibiting attention to mental states, while still lacking more mature and differentiated strategies for responding to abusive, hostile, or ‘vacuous’ other minds.25

On the basis of this developmental perspective, Fonagy proposed that an inhibition of the capacity to represent mental states, perhaps initiated in toddlerhood, could account for the symptoms of BPD: disruptions in an individual’s sense of their identity, their identification and regulation of feelings, their capacity to empathize with others, and generally their capacity to engage in satisfying social relationships. To resolve these problems, Fonagy proposed that the capacity to represent mental states needed to be reignited: ‘Only the gradual (p. 54) elaboration of mental representations of thoughts and feelings will free them from the crushing grip that the concrete experience of reality imposes upon the human mind.’26

Where the mental states of self and other can be drawn upon as a source of clear, specific, relevant, and transferable information, this permits an individual to make gradual and constant adjustments to their expectations about social interactions. The blockage of this process was likewise regarded by Fonagy as the basis for the extreme stances of patients with BPD, as well as contributing to a sense of grievance with a world that will not align with their expectations.27 Fonagy held that blocks on thinking about mental states could also account for other symptoms of BPD. For example, the feelings of disintegration reported by Fonagy’s patients could be understood in this perspective as reflecting difficulties in integrating self-relevant personal and social information, which would usually stabilize a sense of self. The feelings of emptiness reported by the patients could be understood as reflecting blocks in what would otherwise be the vivid sensations from experiences of self and other minds, and the shallowness with which relationships are therefore experienced: ‘The loss of the underlying experience of shared consciousness makes the whole world appear flat, meaningless and isolating.’28 Without shared consciousness, there is a poverty of sources of value, given that values are inherently social.

Fonagy cited his debt to research on social cognition, and especially the work of Baron-Cohen and Dunn. These researchers had demonstrated that the capacity for social understanding was a developmental achievement. However, Fonagy felt that Baron-Cohen tended to depict the child as an independent cognitive agent, alone in attempting to know the world. This neglected the centrality of child–caregiver interaction and emotional relationship, as an important environment within which children learn what it means to recognize a feeling, thought, or intention within themselves or in others. These early relationships also shape children’s feelings about such recognition, including whether it will likely offer benefits or whether punishments will ensue.29 In contrast to Baron-Cohen, the work of Dunn was closer (p. 55) to Fonagy, because Dunn had suggested the importance of family interaction for scaffolding children’s understanding of the i) origin, ii) location, and iii) functioning of mental states. However, Dunn had also considered children’s iv) understanding of social and cultural conventions under the rubric of social cognition, and this more sociological set of meanings was not of such interest to Fonagy, at least in the 1990s (see Chapter 9).30 He was much more concerned with the individual’s capacity to attend specifically to their own mind and the mind of concrete others.

The term ‘mentalization’ was introduced to describe social cognition regarding the origin, location, and functioning of mental states. A desire on the part of Fonagy to circumscribe the object of concern was understandable—especially in retrospect, given that the concept of ‘mentalization’ has ballooned dramatically. However, the decision to cut understanding of social and cultural conventions from mentalizing would be incredibly important in circumscribing the scope of causal explanations and theorical concern, and for kicking into the long grass questions of context and culture that would haunt the paradigm over decades, as Fonagy and colleagues have recently acknowledged (see Chapter 9). Above all, the role of social and cultural conventions and processes in structuring the origin, location, and functioning of mental states—and who is attributed mental states at all—was ejected from view. In Lacanian terms, Fonagy’s decision to exclude concern for understanding rules and conventions from mentalizing limited the theory to concern with the felt experience of self and others (what Lacan terms ‘the imaginary’), neglecting the complex and conflicted cultural context that structures our felt experience of self and others (‘the symbolic’).31 There was no intrinsic reason why a concern with individual perceptions of minds could not have been integrated with attention to social and cultural conventions. However, at the time, it was not, with the result that culture had to be rediscovered in the 2010s (see Chapter 9).

(p. 56) On his first use of the word in 1989, Fonagy described it as a contraction of the phrase ‘capacity to conceive of mental states’.32 The term presented itself in part because of an existing usage in Paris.33 Marty had been the president of the Parisian Psychoanalytic Society when Ivan Fónagy moved to the city and began training as a psychoanalyst. In 1972, Marty was co-founder of a clinic specializing in the treatment of ‘psychosomatic’ symptoms. The École Psychosomatique de Paris came to conceptualize these symptoms as expressing unprocessed mental states, which had been defensively blocked from achieving adequate and flexible representation.34 The capacity to form such representations was termed by Marty and colleagues as ‘mentalization’. Presumably conversations between Peter and his father contributed to an awareness of these developments even before they reached their full and widely known statement in Marty’s 1990 book, Mentalisation et Psychosomatique.35 (Perhaps influenced by Fonagy’s introduction of the concept, the term ‘mentalization’ would also enter the literature on social cognition in the 1990s to refer to the capacity to model the minds of others).36

Related ideas were also circulating in British psychoanalytic circles, associated especially with the work of Bion, who had described distortions of social and self-perception stemming from problems in the representation of thoughts and feelings.37 On Fonagy’s interpretation, Bion held that these distortions would be especially likely when an individual felt threatened, internally or externally. This helped Fonagy articulate the difference between mentalization and theory of mind, in a paper co-authored with the psychiatrist Anna Higgitt and published (p. 57) in 1990, the same year that Fonagy and Higgitt married. Baron-Cohen and colleagues studying autism had conceptualized this condition as a pervasive deficit in theory of mind. By contrast, Fonagy and Higgitt characterized mentalization as the capacity to continue generating models of mental states in the face of perceived internal or external challenges or conflict:

Failures of mentalisation tend to be limited and specific to situations characterised by intense psychic conflict, particularly of an interpersonal kind. The large number of poets and writers (e.g. Baudelaire, Carroll, Nerval, Rimbaud and Verhaeren) whose biographies illustrate that their psychic functioning was close to or at the borderline level, yet whose capacity for abstract thinking in relation to the most subtle of human emotions could not be doubted, testify to this fact.38

Whereas Baron-Cohen had conceptualized autism as a pervasive deficit in theory of mind, Fonagy and Higgitt characterized mentalization as, specifically, a capacity. It might not feature much of the time when not needed. Furthermore, like other capacities, we may sometimes experience lapses in our ability or desire to use it, perhaps especially under conditions of stress or anxiety. Nonetheless, Fonagy and Higgitt described the major potential value of having access to the capacity to identify and consider mental states for both psychological health and social relationships.

In a series of articles in the 1990s, Fonagy and Target elaborated implications of this perspective for clinical technique. They agreed with the observation of the borderline group at the Anna Freud Centre that psychoanalytic interpretations had the potential to backfire. They felt that they could pinpoint the reason for this: interpretations can only be helpful to patients who can think about the difference between i) their personal wishes and worries about their analyst, and ii) the analyst’s here-and-now reality as an ally in reflection. Fonagy and Target also argued, however, against overreliance on pragmatic advice to patients about concrete actions. They instead encouraged the use of ‘brief, accurate and simple statements of the analyst’s perception of the patient’s current mental state’.39 There are resonances here of Fonagy’s retrospective descriptions of his analysis with Anne Hurry. And, in the early 1990s, Fonagy echoed Hurry’s advice that clinicians should adopt a tone of empathy with their patients. However, as the 1990s went on, Fonagy came to disagree on this point: ‘I hate the word empathy because it carries the implication that you have to be nice.’40 Fonagy felt that therapists could adopt a variety of tones in helping the patient gain perspective on mental states—these could be kind without being ‘nice’. Hurry’s own use of benevolent humour had suggested that what was important was the capacity of the therapist to spur the patient to mentalize, rather than empathy itself.

Fonagy and Target speculated that the active ingredient across most successful therapeutic modalities was actually support for improved mentalization. When patients in psychoanalytic psychotherapy could make use of interpretations, this would improve mentalization directly by allowing the patient to reflect on and integrate different perspectives on potentially distressing mental states. Cognitive Behavioural Therapy (CBT) was likewise anticipated to (p. 58) contribute to mentalizing by helping patients elaborate and organize representations of their mental states, allowing them to be used appropriately even in the context of interpersonal or internal difficulties.41 For Fonagy and Target, enhanced mentalization would offer several benefits. It would support emotion regulation by interposing the capacity to reflect between the immediacy of felt experience and the chain of potential responses to these feelings. It would reduce anxiety by contributing to the orderliness of internal experience, and make this experience feel more manageable. And it would support the capacity of the patient to make use of social relationships to help them learn from experience, capitalize on potential opportunities for flourishing, and respond sensibly to adversities.

Reflective function

Interviewed in 2015, Fonagy was asked about the greatest regret of his career. He replied:

I spent the early years of my career a bit lost in the wilderness. It wasn’t until two PhD students focused my attention on attachment and Bowlby that I found the topic in psychology that I was really genuinely interested in.42

These students were Miriam and Howard Steele. The Steeles had been introduced to attachment theory by Larry Aber in New York.43 Attachment theory in the late 1980s was undergoing an exciting revolution. Main and colleagues in Berkeley had introduced the Adult Attachment Interview in 1985, as a measure of a speaker’s capacity to speak coherently about attachment-relevant memories. Main and colleagues found that some speakers were unable to maintain coherent discourse in the interview and would either close down the subject of attachment-relevant memories or become entangled in the memories and lose track of the interview questions. Another interruption of coherence could come from lapses in reasoning or discourse when speakers discussed experiences of bereavement or trauma, suggesting a disturbed state of mind regarding these events. The Adult Attachment Interview coding protocol is not concerned with assessing the speaker’s childhood. There are sub-scales for inferred experience, such as the extent to which the child was rejected or neglected by their parents. However, the coding system primarily focuses on examining individual differences in how childhood attachment relationships are discussed by the adult speaker.

(p. 59) Main and colleagues found that speakers lower in coherence on the Adult Attachment Interview were more likely to have relationships with their child classified as insecure in the Ainsworth Strange Situation.44 The Strange Situation is an observational procedure in which repeated separations and reunions of an infant from their familiar caregiver are inferred to activate the attachment system, prompting the infant to want to seek the availability of their caregiver. Individual differences in response to the activation of the attachment system are observed. ‘Secure attachment’ is coded when infants communicate their distress at the separation but can be comforted by reunion and able to then return to play. Ainsworth termed this pattern the use of the caregiver as a ‘secure base’ and ‘safe haven’, because it seemed to reflect trust in the caregiver’s availability. Some infants did not display their distress to the caregiver; Ainsworth found that these infants experienced rebuff of their proximity-seeking behaviours at home, and she termed the dyads ‘avoidantly attached’. Some infants showed frustration and were unable to accept comfort from the caregiver; Ainsworth found that these infants had caregivers who were not adept at responding accurately to their infants’ cues.45 Ainsworth labelled these dyads ‘ambivalent/resistantly attached’. A further classification for infant behaviour in the Strange Situation, introduced by Main and colleagues, was ‘disorganized’ attachment. Main and colleagues characterized disorganized attachment as representing the most concerning form of insecure behaviour.

It should be pointed out that the concept of disorganized attachment was introduced by Main and colleagues in quite a misleading way. The term ‘disorganization’ in ordinary language means something thrown into chaos. By contrast, they used the term in a technical sense, but without explaining this to the reader.46 Disorganized attachment was coded on the basis of infant displays of i) conflicted, ii) confused, or iii) apprehensive behaviour towards their caregiver, where these suggest conflict in expectations about the availability of the caregiver as a safe haven. When a child’s behaviour in the Strange Situation actually appears random, this is coded as ‘Cannot Classify’, rather than disorganization. The assignation of disorganized attachment actually requires the coder to observe determinate ‘patterning within the disturbance of coherence’, rather than unpredictability.47 Another source of confusion was that Main and Hesse proposed that disorganized attachment is caused by fear in relation to the caregiver. This suggests a unitary process, triggered by hostile caregiving. Certainly disorganized attachment was, soon after its introduction, identified as having increased prevalence in maltreatment samples.48 However, Main and Hesse found that the (p. 60) caregivers from dyads classified as showing disorganized attachment were especially likely to be unresolved for bereavement or trauma on the Adult Attachment Interview, among other antecedents.49 Such findings suggested that the Adult Attachment Interview could be a powerful tool for exploring factors that disrupt the development of secure child–caregiver attachment.

Miriam and Howard Steele brought the first article on the Adult Attachment Interview with them when they moved to London and shared this with Fonagy. Fonagy was especially excited by the fact that ‘both attachment theory and modern psychoanalysis have as their fundamental epistemic aim the description of the internal mechanisms responsible for the discrepancy between actual and psychic reality.’50 The Steeles and Fonagy attended training in the Adult Attachment Interview held in 1987 at the Tavistock Clinic, taught by Mary Main and Erik Hesse at the invitation of John Bowlby.51 Between 1987 and 1989, the Steeles conducted the Adult Attachment Interview with 100 expectant mothers and fathers in the third trimester. They also collected 96 infant–mother Strange Situation observations when the children were 12 months old, and 90 infant–father Strange Situation observations at 18 months. This was the basis for the University College London Parent–Child Project, a fascinating longitudinal study that would eventually follow the sample over the subsequent 17 years.

The first findings from the UCL Parent–Child Project were published by Fonagy, Steele, Moran, Steele, and Higgitt in 1991. The researchers reported powerful intergenerational continuities for mothers: around three-quarters of children whose mothers showed low coherence on the Adult Attachment Interview, conducted prenatally, displayed insecure patterns of attachment in the Strange Situation procedure. The same was true for around half of fathers. Eighty per cent of children whose mothers or fathers had high coherence in the Adult Attachment Interview displayed a secure pattern of attachment in the Strange Situation. The Strange Situation classifications were not predicted by self-report of the quality of the marital relationship or by attitudes towards caregiving by either parent.52 The coding system for the Adult Attachment Interview had a sub-scale for ‘meta-cognitive monitoring’, the extent to which a speaker seemed able to reflect on their own thoughts, feelings, and discourse. However, the scale was (and, in fact, still is) unfinished. The Steeles and Fonagy, examining the transcripts of coherent speakers with a clinical eye, ‘noticed that a defining feature of their narratives was the way they relied on language as a tool for giving meaning to experience, including (p. 61) the attribution of mental states (beliefs and desires).’53 These qualities seemed relevant across the Adult Attachment Interview, but especially in response to two questions that demanded examination of the motivations and influences guiding the behaviour of the speaker and their attachment figures: ‘When you think about your childhood experiences, do you think they have an influence on who you are today?’ and ‘When you think about your parents’ behaviour toward you when you were a child, why do you think they behaved the way they did?’54

The Steeles and Fonagy initially ‘referred to this phenomenon as evidence of the “internal observer” as we thought this term captured the sense in which adult speakers could observe how family life was when they were children, and distinguish this from the understanding they gained through later relationships across development.’55 However, the ‘internal observer’ concept was amended to the ‘reflective self’.56 This was a term from the psychoanalyst Christopher Bollas, used to encompass the fact that the attribution and reconsideration of mental states seemed to be more than perspective-taking on the past, and could include appraisal of different perspectives in the present or imagined future.57 The term was then extended again to ‘reflective function’, to encompass the capacity of the speaker to interpret the experiences of attachment figures with reference to mental states.

The Adult Attachment Interview had been modelled on a clinical interview, integrating two components. The earlier component was a psychoanalytic approach to clinical interviewing from the 1970s, in which psychological defences or confusions were deciphered by the clinician. The later component, under the influence of the American Psychiatric Association’s DSM-III, was use of clinical interviews from the 1980s to identify specifiable markers for categories of mental pathology within the particularities of what and how an individual reports their experiences.58 Reflective function can likewise be seen as a further integration of the Adult Attachment Interview with developments within psychoanalysis that turned attention to obstacles to thinking and understanding (see Chapter 1). Indeed, though not situated as such by Fonagy and colleagues at the time, it is very likely that an influence on the concept of ‘reflective function’ was Bion’s earlier concept of ‘alpha function’, to mean the process by which determinate thoughts and feelings are made out of experiences.59

(p. 62) Frequently, reflective function and mentalization were treated as synonyms by Fonagy and colleagues in the 1990s and 2000s. However, a few distinctions can be identified. Whereas the term ‘mentalization’ was used by Fonagy and colleagues to refer to the capacity of an individual to think about mental states, the term ‘reflective function’ appears to have been preferred, especially by the Steeles, to describe the capacity to think about mental states in the context of attachment relationships specifically.60 The Steeles and Fonagy developed a manual for coding reflective functioning and applied it to the transcripts. Though this was the development sample for the measure, and so strong findings might be expected, the results were still surprisingly clear. Reflective function in the Adult Attachment Interview not only predicted security in the Strange Situation better than coherence for mothers and fathers, but coherence had no association with security when reflective functioning was statistically controlled.61 The Steeles and Fonagy also checked how the reflective functioning measure compared with the inferred experience scales: again, they found that reflective functioning was a better predictor than the inferred experience scales, which had no association with the Strange Situation once reflective functioning was controlled. The researchers concluded:

The coherence of the parents’ perception of their past derives from their unhindered capacity to observe their own mental functioning, to have a plausible view of themselves and their objects as human beings, thinking, feeling, wishing, believing, wanting, and desiring. We assume, then, that coherence may be a measure of reflectiveness, and it is the latter attribute of the caregivers that has direct implications for their relationship to the infant … In our view, a child may be said to be secure in relation to a caregiver to the extent that, on the basis of his or her experience, he or she can make an assumption that his or her mental state will be appropriately reflected on and responded to accurately. The child’s confidence in this assumption will enable her or him to be more assured about the safety of the world of ideas and desires; she or he will feel secure in relation to her or his mental world.62

The unpublished manual conveys in more detail how the construct of reflective function was operationalized. It was by no means a quantitative count of occasions when mental state terms were used by speakers:

for a statement to be reflective, it has to imply an activity of reflection, rather than simply describing a mental state. Thus, the statement ‘I was angry’ would not be considered (p. 63) reflective, whereas the statement ‘At the beginning I felt really angry and then it changed to sadness’ does suggest that the feelings have been considered.63

Compared with the general construct of mentalization, the operationalization of reflective function is notable for a few subtle differences, perhaps shaped by the priorities of the Adult Attachment Interview. The coding of reflective function appeared to give special weight to the capacity of the speaker to identify change in their perspective over time. And, more than the idea of mentalization, reflective function was concerned with understanding behaviour and mental states as having bidirectional lines of influence, not solely the interpretation of mental states as the cause of behaviour.64

The manual advised that individual statements should be given a score on a 1–9 scale, as should the transcript as a whole. This required the coder to make a synthesis of the extent, intensity, and quality of the varied forms of reflective functioning evident across the transcript. Fonagy and colleagues reported that ‘when reflective-functioning is absent or ineffectual, quite distinct, readily classifiable, patterns of responses tend to emerge.’65 One pattern was rejection of reflective functioning, with the prompts in the interview to think about childhood experiences and the intentions of attachment figures responded to in a hostile way by the speaker. Another pattern of particular note was ‘hyperactive reflective function’, in which there is extensive discussion of mental states, but these are poorly integrated and do not have much bearing on actually answering the questions posed by the interviewer: ‘Mentalization spins like a car wheel which has lost contact with ground. Because so many of the connections between what is thought about thoughts or feelings and the original thoughts or feelings themselves have been lost, metarepresentational capacity works overtime but without real impact.’66 This kind of response would later be termed ‘hypermentalizing’ (see Chapter 5).

A further pattern identified by Fonagy and colleagues was ‘disavowal of reflective function’, the most common form of which were explanations of caregivers’ behaviour in ‘sociological’ rather than psychological terms, such as ‘people in their social class never expected to be affectionate with their children’. Fonagy had earlier excluded Dunn’s attention to social convention as part of mentalization. Taking this logic further, now the operationalization of reflective function actively treated sociological explanations as rather at the expense of reflection of the mental states of determinate figures. Mentalization in general, and reflective function in particular, were conceived of as ‘psychological processes underpinning the (p. 64) view of oneself and others as motivated by mental states’.67 Attention to processes above the level of mental states, such as sociological forces were therefore characterized as the absence of mentalizing (Chapter 9). Embodied processes below the level of mental states, such as endocrinology (e.g. appeal to the causal role of adrenalin or cortisol; consideration of post-partum hormone changes), are also ambiguous in this regard.68 The formulation offered by Fonagy and colleagues would seem to implicitly exclude them as legitimate objects of mentalization.69

The link between caregiver reflective function and infant attachment security in the UCL Parent–Child Project suggested to Fonagy and colleagues four revisions or amendments to important concepts in attachment theory in the late 1990s:

  • Sensitive care.

  • The cycle of abuse.

  • Internal working model.

  • Security.

First, Fonagy and colleagues proposed a new perspective on sensitive care. Ainsworth had proposed that the basis for secure infant attachment lay in caregiver ‘sensitivity’, a technical term she used for the capacity to ‘perceive and to interpret accurately the signals and communications implicit in her infant’s behavior, and given this understanding, to respond to them appropriately and promptly’.70 Though incredibly perspicacious, there are nonetheless several problems with this account. It assumes that infants offer a single signal at a time or that signals are ultimately compatible; that it is in the parent’s capacity to solve the problem; that there are no needs without, prior to, or contrary to signals; that needs are not incompatible; and it neglects the importance of distraction.71 Perhaps most importantly, not all signals pertain to attachment needs. An infant may cry because they are bored or confused, for example. So even exquisite sensitivity will only ever be at best a moderate predictor of (p. 65) attachment security.72 For their part, Fonagy and colleagues argued that caregiver sensitivity was largely a by-product of reflective function, and that it was the latter that was largely the basis for infant secure attachment. What was captured by Ainsworth’s sensitivity scale, Fonagy and colleagues proposed, was the extent to which the caregiver was attentive to the mental states of the child, could interpret these accurately, and could respond to them promptly and in an undistorted way.73 Later research has indeed supported the claim that some of the association between sensitivity and infant attachment can be accounted for in terms of parental reflective function.74 However, it would have been surprising if all of the association had been accounted for, because both sensitive care and attachment security can readily be seen in non-human animals with far less role for representation in parent–child interactions.75 There may be different paths to sensitivity through automatic and controlled forms of caregiver mentalizing (see Chapter 4)—but this was not a distinction Fonagy and colleagues had available in the late 1990s, nor one they have subsequently considered with respect to sensitivity.

A second alteration to attachment theory was regarding the ‘cycle of abuse’. This was a concept introduced by Egeland, Jacobvitz, and Sroufe in 1984, to describe findings from the Minnesota longitudinal study that about a third of caregivers who reported that they had been abused went on to abuse their own child: a much higher proportion than matched controls, even if overall a minority of cases.76 Egeland and colleagues reported that the abused mothers who did not go on to abuse their child were distinguished by one or more of three experiences. First, some had emotional support from a non-abusive adult during their childhood. Second, some had participated in therapy. Third, some had a stable and emotionally supportive relationship with their adult partner. Fonagy and colleagues argued that what these experiences had in common was that they could be expected to increase a caregiver’s reflective functioning, offering possibilities for interrupting responses that might otherwise predispose abusive or neglecting behaviour towards their child.77

(p. 66) A third change to attachment theory in the late 1990s was regarding the concept of ‘internal working model’. This concept had been introduced by Bowlby to characterize the expectations held by a child about the availability of their caregiver. Fonagy held that the concept of ‘internal working model’ was too vague, given that it encompassed memories of previous interactions with attachment figures, expectations about interactions with attachment figures, ramifications of these memories and expectations for a child’s sense of self, and ramifications of these memories and expectations for the child’s sense of their caregiver.78 When attachment researchers proposed that internal working models were the mechanism through which early childhood experiences contributed to later behaviour, Fonagy regarded this as a badly underspecified claim. He and his father had been discussing the ideas of the philosopher Hegel, who had emphasized the integral role of the other in the self’s recognition of itself.79 And from György Gergely, Fonagy took the idea that self-recognition would be facilitated when the caregiver’s behaviour provided cues to the child that their mental states and intentions were being acknowledged. Gergely argued that an especially potent form of recognition a child could receive is ‘marked mirroring’. This is when a caregiver provides enough of a mirror to the child’s behaviour to show that the intention and/or mental state implicated in the behaviour has been acknowledged, alongside some slight transformation or elaboration to also communicate a sense of acknowledgement. So, for instance, a young child’s cry may be met initially with a sad mouth and tonal echo from the parent, signalling that the emotion has been received, but this response is then transformed by a half-smile and a rising-pitch intonation, as if to convey that the sadness is localized, resolvable, and will be okay soon.80 Marked mirroring utilizes the interplay of primary and secondary meanings highlighted by Ivan Fónagy as integral to human communication (see Chapter 1), offering the primary meaning of acceptance and the secondary meaning of containment.

(p. 67) Drawing on these ideas, Fonagy and colleagues proposed that patterns of attachment should be regarded as a correlate of the child’s early mentalizing capacities, achieved through recognizing themselves in the minds of their caregiver. He felt that attachment researchers had reified the infant classifications and failed to truly develop proposals about ‘the mechanisms or psychic processes that may underlie such behavioral clusters’.81 He speculated that secure attachment reflected a child’s experiences of being treated by their caregiver as having intentions and/or mental states. Avoidant and ambivalent attachment reflected obstacles in the caregiver’s recognition of their need for a secure base or a safe haven, and in turn could be anticipated to contribute to deficits in reflective function: “The avoidant child shuns the mental state of the other, while the resistant child focuses on its own state of distress to the exclusion of intersubjectivity”.82

Yet despite such qualitative differences, Fonagy and colleagues argued that the underlying mechanisms were likely best characterized in terms of a dimension. Ultimately, for Fonagy and colleagues, the importance of individual differences in attachment was that they contributed to or hindered the child’s capacity for mentalization. It was this, not attachment itself nor the representations of caregivers that stem from them, that they felt contributed to a child’s later socio-emotional development. Fonagy and Target argued that ‘the influence of attachment security on later development has nothing to do with representations of early relationships, and a futile search for this link has distracted attachment researchers’.83 The transition in the stance of Fonagy and colleagues can be seen in comparing the developmental model in a 1998 diagram with one 10 years later. Attachment security still figures in the second diagram, but it is not treated as a necessary path to child mentalising, nor necessary for a child’s emotion regulation or later mental health. (p. 68)

Figure 3.1 Developmental model (‘paraphrased from Hegel’).

Figure 3.1 Developmental model (‘paraphrased from Hegel’).

Source: Reproduced from Peter Fonagy, ‘Prevention, the Appropriate Target of Infant Psychotherapy’, Infant Mental Health Journal, 19(2): 124–150, Figure 4, DOI:<124::AID-IMHJ4>3.0.CO;2-O Copyright © 1998 Michigan Association for Infant Mental Health.

Figure 3.2 Developmental model.

Figure 3.2 Developmental model.

Source: Reproduced from Carla Sharp and Peter Fonagy, ‘The Parent’s Capacity to Treat the Child as a Psychological Agent: Constructs, Measures and Implications for Developmental Psychopathology’, Social Development, 17(3): 737–754, DOI: Copyright © Blackwell Publishing Ltd. 2008.

Finally, a fourth revision to attachment theory proposed by Fonagy and colleagues addressed the concept of security and its clinical implications. Bowlby and several other attachment theorists had argued that the therapist offered a ‘secure base’ to the patient, from which they could venture to explore difficult thoughts and feelings. This process activated the patient’s expectations about attachment relationships, and at the same time provided an opportunity to reflect on them. In this way, the therapy could contribute to revision in internal working models and a potential increase in the security of these models. Fonagy and colleagues agreed that the therapeutic relationship activated the patient’s expectations about attachment relationships. However, he did not believe that the benefits to the patient of therapy stemmed from an increase in attachment security.

Instead, Fonagy praised Patricia Crittenden for the proposal that, when the attachment system is activated by perceived external or internal sources of threat, both children and adults would be predisposed to fall back on automatic processing.84 Drawing ideas from Linda Mayes, Fonagy and Bateman redescribed this process as a tendency among patients with BPD to fall into an implicit, non-mentalizing form of processing when the attachment system was activated.85 As the attachment bond between therapist and client intensifies, the quality of mentalizing of patients, especially those with BPD, was anticipated to drop off. The therapeutic task, then, was to help patients retain their capacity to mentalize even in the context of attachment relationships. It was anticipated that successes in the mentalization of attachment relationships, because of their foundational position for psychological development, would be especially likely to be extrapolated to other areas of life:

(p. 69)

For reasons that may have evolutionary significance, the intense activation of the networks underpinning attachment feelings and experiences also appears to inhibit the intensity of cognitive and emotional scrutiny over mental contents. This state of affairs creates a unique opportunity for the psychotherapist. By balancing the activation of attachment against the presentation of negative mental contents, they are able to present new stimuli (mental contents) to the patient without evoking mental resistance against the incorporation of new ways of experiencing the world into existing cognitive–emotional schemata.86

Fonagy and colleagues proposed that a therapeutic focus on mentalization would be of benefit to patients with most mental health problems. Putting matters stridently, Fonagy and Target stated that, ‘in our view, change can happen solely through the revival of reflective function.’87 However, they argued that a focus on mentalization in the context of attachment relationships would especially benefit patients with BPD, because they regarded this condition as primarily reflecting deficits in mentalizing in the context of attachment relationships. If difficulties with mentalizing could be drawn as a spectrum, BPD was conceptualized as the extreme end of this spectrum. However, at times, they also described the pathway to other conditions—such as dissociative disorders—in exactly the same terms as the pathway to BPD, in terms of a turning away from conceiving of the mind of the caregiver resulting in mentalization deficits.88 By the early 2000s, such diffuse appeal to the same explanatory mechanism and psychological constructs was starting to raise serious questions about the exact meaning of mentalization and about the specificity of the developmental model underpinning it. However, this model was anchored in place, at least for a time, by the manner in which Fonagy and colleagues made use of the concept of disorganized attachment.

Disorganization and mentalization

The concept of disorganized attachment was introduced by Mary Main and Judith Solomon, and technically refers to the display of conflicted, confused, or apprehensive behaviour by young children in the Strange Situation procedure. In writing with the Steeles in the early 1990s, reporting empirical findings from studies using the Strange Situation, this was the meaning of the term in Fonagy’s work.89 Fonagy and Target were enthusiastic, stating that (p. 70) ‘the most promising area of attachment research is undoubtedly the study of disorganized/disoriented attachment behaviour’.90 However, in the context of this enthusiasm, the term ‘disorganized attachment’ took on a further life through the 1990s and 2000 in the writings of Fonagy and his immediate collaborators. This usage was cut free from the actual observable characteristics of disorganized attachment. The most important factor was that Main, Solomon, and colleagues had used the term ‘disorganized’ to refer to their observations, but had intended a technical meaning that they did too little to clarify for the reader: observations of conflict, confusion, and apprehension, from which can be inferred conflict of expectations about the availability of the caregiver as a safe haven. Seeing the confusion their early writings have caused for subsequent researchers, Main, Solomon, and colleagues have accepted a portion of the responsibility for subsequent hazy or spurious uses of the category.91

Fonagy’s group, however, were especially vulnerable to this wider problem. After Howard and Miriam Steele left for New York, none of Fonagy’s circle of collaborators were reliable coders of disorganized attachment, allowing their theorizing to drift free from the operationalization of the construct.92 Use of the concept of disorganized attachment seems to have been implicitly shaped by the narrative need for an antagonist within stories about the development of mentalizing.93 This was likely unintentional. However, unmoored to the construct’s operationalization, the result has been an account of disorganization that slips into metaphor, or—since the metaphorical status is not signalled—into ‘pretend mode’ (see Chapter 5).

To take one example: in a chapter from 2010, Fonagy, Luyten, Bateman, Gergely, Strathearn, Target, and Allison depict disorganized attachment as ‘undirected/bizarre behaviour’ and ‘trying to escape the situation even in the caregiver’s presence.’94 However (p. 71) ‘undirected/bizarre behaviour’ is a partial description of one of the seven indices of disorganized attachment (‘undirected/misdirected behaviour’). Trying to escape the situation is actually coded as avoidant attachment (‘5b’ in the avoidance scale protocols: baby ‘tries to go past her out the door’), and is not necessarily treated by coders as disorganization.95 Fonagy and colleagues then say, ‘in these individuals, the attachment system may be quite readily triggered, and they may appear to be constantly preoccupied with attachment relationships’.96 This is ambivalent/resistant, not disorganized attachment. Some infants in dyads classified as disorganized show preoccupation with attachment relationships. Others combine conflicted, confused, or apprehensive behaviours with avoidance of the caregiver. Others combine these behaviours with otherwise secure use of the caregiver as a secure base and safe haven.

Over time, Fonagy has come to increasingly acknowledge that the relationship between disorganized attachment and mentalizing is weak and probably indirect (see Chapter 7).97 However, through the 1990s, 2000s, and early 2010s, reference to disorganized attachment in causal accounts of disturbed development by Fonagy and colleagues was pervasive,98 and one of the least meticulous aspects of their developmental model. Rather than the observable behaviour, or known correlates, the characterization of disorganized attachment seemed shaped more by connotations of the word ‘disorganized’ itself, which in ordinary language means to be thrown into chaos. This was facilitated, not just in the work of Fonagy and colleagues but also in much other speculation about disorganized attachment, by Mary Main’s introduction of the term with a technical meaning, but without adequate definition or explanation.99 In a chapter from 2007, Fonagy and Higgitt acknowledged this problem, observing that ‘disorganised attachment is currently poorly described in the literature,’ implying that this has hampered their own use of the concept.100

Examination of appeals to disorganized attachment in their writings reveals that Fonagy and colleagues have offered diverse non-overlapping descriptions of the nature of (p. 72) (p. 73) disorganized attachment (see Tables 3.1 and 3.2). Each of these five descriptions is an extrapolation from some among the behaviours used to code disorganized attachment. The most common description of disorganized attachment as a whole, especially in writings with Luyten, is of a combination of avoidant and resistant attachment.101 In fact, only a small fraction of child–caregiver dyads who receive a disorganized attachment classification show both avoidance and resistance.102 Another common description is of disorganized (p. 74) attachment as the absence of any logic to interpersonal interactions in the context of emotional dysregulation. Again, there are some infants in the disorganized category who show pervasive disturbances of affective control. However, these are a vast minority.103 Children classified as disorganized under the Main and Solomon protocols do not show a coherent avoidant or resistant strategy; but this does not mean that their behaviour lacks strategy or (p. 75) any form of regulation.104 Part of the problem is that it is not entirely clear what is meant by ‘absence of coherent affect regulation’ or ‘absence of any strategy’. But, in general terms, they are at best very partial characterizations of disorganized attachment, and rather misleading. It would appear that Fonagy and colleagues are not primarily talking about the same thing that developmental attachment researchers have operationalized under the rubric of ‘disorganized attachment’. This is not to minimize the importance of the process they are discussing. However, it remains insufficiently clear what this process is, and use of the term ‘disorganized attachment’ is misleading.

Table 3.1 Descriptions of disorganized attachment in Fonagy and colleagues




A combination of avoidant and resistant attachment behaviour

Luyten, P., van Houdenhove, B., Lemma, A., Target, M., and Fonagy, P. (2012). ‘A Mentalization-Based Approach to the Understanding and Treatment of Functional Somatic Disorders’. Psychoanalytic Psychotherapy, 26(2): 121–140: ‘When attachment figures have been insufficiently available, abusing and/or nonresponsive, attachment deactivating or hyperactivating strategies, or a combination of both, as is typically observed in disorganized attachment, become a habitual response to stress’ (p. 126).


The breakdown of combinations of avoidant and resistant attachment behaviour into a diffuse state

Luyten, P., Fonagy, P., Lemma, A., and Target, M. (2012). ‘Depression’, in A. W. Bateman and P. Fonagy (eds.), Handbook of Mentalizing in Mental Health Practice (pp. 385–418). Washington, DC: American Psychiatric Association: ‘Patients with dysfunctions of the attachment system may show a sequence, moving from deactivating strategies to hyperactivating strategies if the former fail, or may show marked oscillations between the two strategies … depressed individuals with these features often show many borderline features … At higher levels of personality functioning, these patients may show a pattern of fearful-avoidant attachment, and thus a typical approach-avoidance conflict with regard to attachment relationships. This conflict is typically expressed either in very brief and barren attachment relationships, as in patients with schizoid and schizotypal features, or in hypermentalising accounts revealing both their desire for and their fear of relationships, as in patients with hysterical features. It is our impression that in individuals at lower levels of personality functioning, disorganised attachment is related to identity diffusion, while individuals at higher levels of personality functioning seem to be characterised by marked rigidity. Hence, whereas the former group may be particularly responsive to mentalisation interventions, the latter may show high levels of pseudomentalising, which is not only difficult to distinguish from genuine mentalisation, but also perhaps more difficult to treat’ (pp. 413–414).


A hyperactivating strategy similar to anxious attachment, though often with ‘frantic attempts to downregulate’

Fonagy, P. (1998). ‘Prevention, the Appropriate Target of Infant Psychotherapy’. Infant Mental Health Journal, 19(2): 124–150: ‘Anxious attachment or disorganization of attachment develops (Erickson, Sroufe, and Egeland, 1985; Greenberg et al., 1993) to the mother as part of a strategy to avoid being blocked from access to her’ (p. 135).

Bateman, A. W. and Fonagy, P. (2016). Mentalization-Based Treatment for Personality Disorders: A Practical Guide (2nd edn), Oxford: Oxford University Press: ‘Disorganised—Incoherent: hyperresponsive, but often frantic attempts to downregulate’ (p. 123).


The absence of coherent affect regulation

Fonagy, P. and Higgitt, A. (2000). ‘Early Influences on Development and Social Inequalities: An Attachment Theory Perspective’, in A. R. Tarlov and R. F. S. Peter (eds), The Society and Population Health Reader. Vol. II: A State and Community Perspective (pp. 104–130), New York: New Press: ‘Disorganised or unresolved patterns are held to be indicative of an absence of coherent affect-regulation strategies’ (p. 109).


The absence of any strategy to interpersonal relationships–for instance, the absence of help-seeking in the context of need

Stein, H., Koontz, A. D., Fonagy, P., Allen, J. G., Fultz, J., Brethour, J. R., … and Evans, R. B. (2002). ‘Adult attachment: What are the underlying dimensions?’. Psychology and Psychotherapy: Theory, Research and Practice, 75(1): 77–91: ‘With more insecurity comes an increased need for a strategy to manage attachment relationships. At the extreme insecure end of the dimension, adaptive strategies again diminish, from dismissing or preoccupied strategies to no coherent strategy. Thus, there are probably three important positions, conflated until now: low insecurity with no need for special strategy; moderate to high insecurity with specific strategies for sustaining relationships; and extreme insecurity with a disorganized attachment system that precludes strategy’ (p. 88).

Bevington, D., Fuggle, P., and Fonagy, P. (2015). ‘Applying Attachment Theory to Effective Practice with Hard-To-Reach Youth: The AMBIT Approach’, Attachment & Human Development, 17(2): 157–174: ‘The core assumption of AMBIT is that those considered “hard to reach” are hard to reach for reasons: their avoidance of help is frequently active and intentional, rooted as it may be in profound disorganizations within their attachment systems’ (p. 160).

Table 3.2 Descriptions of the proximal mechanism of disorganized attachment in Fonagy and colleagues




Principally driven by the child’s constitutional characteristics

Fonagy, P., Bateman, A. W., and Luyten, P. (2012). ‘Introduction and Overview’, in Handbook of Mentalizing in Mental Health Practice (pp. 3–42). Washington, DC: American Psychiatric Publishing: ‘Abnormal hippocampal and hyperactive amygdala functioning in children who are at risk for later developing BPD may cause highly anxious and emotionally labile infant behaviour and thus an infant who is not able to benefit from the regulating qualities of the attachment relationship. This constellation might well create disorganised attachment relationships, principally driven by the child’s constitutional characteristics’ (p. 8).



Fonagy, P. (1999). ‘The Transgenerational Transmission of Holocaust Trauma’, Attachment & Human Development, 1(1): 92–114: ‘The behaviour of the infant in this situation gives clear indication of multiple, incoherently integrated structures, highly reminiscent of dissociative adult patients’ (p. 97).


Fear of the caregiver

Fonagy, P. (2003). ‘The Violence in our Schools: What can A Psychoanalytically Informed Approach Contribute?’ Journal of Applied Psychoanalytic Studies, 5(2): 223–238: ‘Neither disorganization of the attachment system nor violence are static. The attachment system becomes disorganized in the face of external and internal demands to deal with safety issues. It is the persistence of fear alongside the activation of the attachment system that accounts for the picture of disorganization. This means that assessment must always be focused on the interplay of the situation and the person. It is neither the situation nor the person that represents the risk, but rather the interaction of the two’(p. 227).


Disruption of the ‘self’

Fonagy, P., Gergely, G., Jurist, E. L., and Target, M. (2002). Affect Regulation, Mentalization, and the Development of the Self. London: Karnac Books: ‘The disorganisation of self disorganises attachment relationships by creating a constant need for this projective identification’ (p. 12).


A preoccupation with ensuring perfect contingency from caregivers

Fonagy, P., Gergely, G., Jurist, E. L., and Target, M. (2002). Affect Regulation, Mentalization, and the Development of the Self. London: Karnac Books: ‘An early association between disorganised infant attachment and an abnormal preference for perfect contingencies’ (p. 250).

As well as offering descriptions of disorganized attachment at a behavioural level, Fonagy and colleagues variously offer five different characterizations of the proximal mechanism of disorganized attachment: constitutional factors; dissociation; fear of the caregiver; disruption of the ‘self’; and a preoccupation with perfect contingencies in attachment relationships. It is absolutely possible that the forms of behaviour observed by Main and Solomon in the Strange Situation are prompted by these diverse mechanisms. It may be that all these pathways lead to a single phenomenon of ‘disorganized attachment’; or they may be different pathways to different forms of confusion, conflict, or apprehension in attachment relationships. The greater problems, however, are that Fonagy and colleagues do not reconcile these different accounts theoretically, and some are lacking much in the way of evidence.105 For instance, in 2002, Fonagy and colleagues write that ‘The disorganization of self disorganizes attachment relationships’; in 2004, they write ‘disorganized attachment is rooted in a disorganized self’.106 Of course, it is quite possible for there to be a bidirectional relationship. However, the two claims are not reconciled. A hazy notion of disorganized attachment, combined with a hazy concept of ‘self’ (see Chapter 6), appears to have hindered the precision of causal claims, and the coherence of the theory built from them. In their writings, Fonagy and colleagues also offer various characterizations of what disorganized attachment entails at a cognitive level. These include the ‘absence of stable representation of interactions and therefore predictability’, internalization of ‘confusing and hostile parenting’,107 and failure to develop emotional awareness.108 Again, however, these accounts are not reconciled.

(p. 76) Furthermore, this picture seems to have been based in part on the connotations of the word ‘disorganized’. The primary trainers in coding disorganized attachment, Alan Sroufe and Elizabeth Carlson, have criticized the depiction of disorganized attachment as lack of predictability, which they regard as an erroneous extrapolation from the fact that some, not all, children in this category show confused behaviours.109 , 110 The idea that disorganized attachment entails confusion of representations about attachment figures may also have partly have stemmed from Main’s original Berkeley study, where some children showing disorganized attachment in infancy showed themes of conflict, confusion, and/or fear in story-stem narratives about attachment relationships at age 6.111 However, this is not necessarily the same as the absence of stable representations about interactions. Nor does it imply that the children lack predictability in their behaviour. In fact, in their classic study, Main and Cassidy found that infants from dyads classified as disorganized showed highly concerted, goal-directed behaviour on reunion at age 6 with their caregivers, either controlling and solicitous or controlling and punitive.112

As we will see in the subsequent chapters, many of the themes gestured to by disorganized attachment—fear, constitutional factors, dissociation, disruption of the ‘self’—have played a significant role in the developing conceptual system of Fonagy and colleagues. Yet, recognition of the distinct role of these factors has been hindered by the absorptive construct of disorganized attachment. The concept of disorganized attachment, as a container, has tended to devour the elements that it ostensibly contains. This problem has been supported by the lack of clarity in attachment research in general, and in the work of Fonagy and colleagues in particular, regarding what exactly constitutes an attachment relationship as opposed to other kinds of relationship. As articulation of role of fear, constitutional factors, dissociation, disruptions to the ‘self’, and other more specific factors has improved in the work of Fonagy and colleagues over the decades, appeal to disorganized attachment by Fonagy and colleagues has been dropping away in recent years. These conceptual improvements stem especially from self-criticism and important theoretical innovations in the 2000s, prompting revisions in the model used by Fonagy and colleagues in conceptualizing the development of mentalizing and non-mentalizing. These revisions will the focus of the next chapter.


1 Fonagy, P. and Target, M. (2003). Psychoanalytic Theories: Perspectives from Developmental Psychopathology, London: Whurr Publications, p. 19.

2 E.g. Fonagy. P. and Allison, E. (2012). ‘What is Mentalization? The Concept and its Foundations in Developmental Research’, in N. Midgley and I. Vrouva (eds), Minding the Child: Mentalization-Based Interventions with Children, Young People and their Families, pp. 11–34, Hove: Routledge, p. 32. This criticism of the concept was first made in a sustained way by Choi-Kain, L. W. and Gunderson, J. G. (2008). ‘Mentalization: Ontogeny, Assessment, and Application in the Treatment of Borderline Personality Disorder’. American Journal of Psychiatry, 165(9): 1127–1135: ‘Quickly adapted into psychiatric vernacular, the term mentalization has been used with variable meanings. The broad territory of the concept paradoxically contributes to its familiarity as well as to its ambiguity’ (pp. 1127–1128). More recently, see Tipple, R. (2017). ‘Thinking versus Mentalization’. Art Therapy Online, 8(2).

3 On the crafting of an idea and ideal of a knowable object of study as a guarantee to sustain coalitions of practice, identification, and research, see Wiegman, R. (2012). Object Lessons, Durham, NC: Duke University Press.

4 On buzzwords, see Bensaude Vincent, B. (2014). ‘The Politics of Buzzwords at the Interface of Technoscience, Market and Society: The Case of “Public Engagement in Science”’. Public Understanding of Science, 23(3): 238–253.

5 Slade, among others, has observed that the term ‘mentalization’ seems to operate with narrow and broader meanings: Slade, A. (2005). ‘Parental Reflective Functioning: An Introduction’. Attachment & Human Development, 7(3): 269–281. To take one example: Fonagy and Adshead suggest that ‘classical behaviour therapy involves mentalisation. When patients are taught new responses to fear-inducing stimuli, changes occur in the ventral prefrontal cortex and amygdala, which are involved in fear reactions.’ The capaciousness of the term ‘mentalization’ is visible here, entailing only that mental states are identified and implicated. This makes mentalization relevant to every psychological modality and specialism. Fonagy, P. and Adshead, G. (2012). ‘How Mentalisation Changes the Mind’. Advances in Psychiatric Treatment, 18(5): 353–362, p. 35. But surely to claim that classical conditioning, effective with rats and artificial intelligences, implies mentalization is difficult to reconcile with Bateman and Fonagy’s majestic claim that ‘mentalising is a uniquely human capacity—it can be seen as what defines humanity’:Bateman, A. W. and Fonagy, P. (2016). Mentalization-Based Treatment for Personality Disorders: A Practical Guide (2nd edn), Oxford: Oxford University Press, p. 4. This statement might be interpreted as reflecting Fonagy’s earlier claim that ‘No animal, not even the most intelligent of non-human primates, can discern the difference between the act of a conspecific due to serendipity and one rooted in intention, wish, belief or desire.’ Fonagy, P. (2012). ‘Mentalization and Attachment: The Implication for Community Based Therapies’. Paper presented at the Community of Communities 10th Annual Forum, March. London. Accessed at:

6 E.g. Ballespí, S., Vives, J., Debbané, M., Sharp, C., and Barrantes-Vidal, N. (2018). Beyond Diagnosis: Mentalization and Mental Health from a Transdiagnostic Point of View in Adolescents from Non-Clinical Population’. Psychiatry Research, 270: 755–763.

7 Reported in Keaveny, E., Midgley, N., Asen, E., Bevington, D., Fearon, P., Fonagy, P., … Wood, S. 2012). ‘Minding the Family Mind: The Development and Evaluation of Mentalization Based Treatment for Families at the Anna Freud Centre in London’, in N. Midgley and I. Vrouva (eds), Minding the Child: Mentalization-Based Interventions with Children, Young People and their Families, London: Routledge, pp. 98–112, p. 102.

8 For later discussion of this attitude, see Chiesa, M., Fonagy, P., Bateman, A. W., and Mace, C. (2009). ‘Psychiatric Morbidity and Treatment Pathway Outcomes of Patients Presenting to Specialist NHS Psychodynamic Psychotherapy Services: Results from a Multi‐Centre Study’. Psychology and Psychotherapy: Theory, Research and Practice, 82(1): 83–98.

9 Bateman, A. W., and Fonagy, P. (2003). ‘The Development of an Attachment-Based Treatment Program for Borderline Personality Disorder’. Bulletin of the Menninger Clinic, 67: 187–211, p. 188

10 Cf. Fonagy, P. (2007). ‘Personality Disorder’. Journal of Mental Health, 16(1): 1–4: ‘We suspect now that it is a considerable oversimplification to assume that a pre-existing characterological basis to psychopathology can provide a breeding ground for psychiatric disorder.’ (1) Pickersgill, M. (2013). ‘How Personality Became Treatable: The Mutual Constitution of Clinical Knowledge and Mental Health Law’. Social Studies of Science, 43(1): 30–53.

11 Jovanovic, G. Allolio-Näcke, L., and Ratner, C. (eds) (2018). The Challenges of Cultural Psychology: Historical Legacies and Future Responsibilities, London: Routledge.

12 Bowlby, J. (1980). Loss, New York: Basic Books; Bretherton, I. (1987). ‘New Perspectives on Attachment Relations: Security, Communication and Internal Working Models’, in J. Osofsky (ed.), Handbook of Infant Development, New York: John Wiley, pp. 1016–1100.

13 Wellman, H. M. and Estes, D. (1986). ‘Early Understanding of Mental Entities: A Reexamination of Childhood Realism’. Child Development, 57: 910–923. See also Flavell, J. H. (2004). ‘Theory-of-Mind Development: Retrospect and Prospect’. Merrill-Palmer Quarterly, 50(3): 274–290.

14 Bretherton, I., McNew, S., and Beeghley-Smith, M. (1981). ‘Early Person Knowledge as Expressed in Gestural and Verbal Communication: When do Infants Acquire a “Theory Of Mind”?’, in M. E. Lamb and L. R. Sherrod (eds), Infant Social Cognition, Hillsdale, NJ: Erlbaum; Dunn, J., Bretherton, I., and Munn, P. (1987). ‘Conversations about Feeling States between Mothers and their Young Children’. Developmental Psychology, 23(1): 132–139.

15 Baron-Cohen, S., Leslie, A. M., and Frith, U. (1985). ‘Does the Autistic Child have a “Theory of Mind”?’ Cognition, 21(1): 37–46.

16 Fonagy, P. (1999). ‘Interview with Peter Fonagy’, in S. M. Stein and J. Stein (eds), Psychotherapy in Practice: A Life in the Mind, Oxford: Butterworth Heinemann, pp. 77–98, p. 86. This point of contact was further facilitated by a symposium held to facilitate dialogue between developmental scientists and clinicians at the Anna Freud Centre. Moran, G. S. (1990). ‘Report on the One-Day Symposium between European Developmental Scientists and Anna Freud Centre Staff’. Bulletin of the Anna Freud Centre, 13(3): 167–168.

17 Cf. Fonagy, P. (2011). ‘Discussion of Juan Pablo Jimenez’s Paper, “A Fundamental Dilemma of Psychoanalytic Technique. Reflections on the Analysis of a Perverse Paranoid Patient”’, in J. P. Jimenez and R. Moguillansky (eds), Clinical and Theoretical Aspects of Perversion, London: Karnac Books, pp. 63–76: ‘Of course, the malevolent thoughts of the adult may not be real, but may be anticipated by the child, giving rise to a phantasy that may be the product of neurotic compromise, as when the Oedipal child imagined parental retribution’ (pp. 64–65).

18 Fonagy, P. (1991). Thinking about Thinking: Some Clinical and Theoretical Considerations in the Treatment of a Borderline Patient’. The International Journal of Psychoanalysis, 72: 639–656: ‘In the case of borderline functioning, I believe the deficit to be self-imposed and partial. It is brought about by a defensive disavowal of the mental existence (in terms of psychic functioning) of the object. Such disavowal is undertaken in the face of an anticipation of unbearable psychic pain’ (p. 651).

19 Ibid. 650.

20 Fonagy, P. (1989). ‘On Tolerating Mental States: Theory of Mind in Borderline Patients’. Bulletin of the Anna Freud Centre, 12: 91–115, pp. 107–108. In fact, later research would show that physical abuse was less predictive of personality disorder than other forms of maltreatment. Cirasola, A., Hillman, S., Fonagy, P., and Chiesa, M. (2017). ‘Mapping the Road from Childhood Adversity to Personality Disorder: The Role of Unresolved States of Mind’. Personality and Mental Health, 11(2): 77–90, Table 2.

21 Fonagy, P. (1996). ‘The Significance of the Development of Metacognitive Control over Mental Representations in Parenting and Infant Development’. Journal of Clinical Psychoanalysis, 5(1): 67–86: ‘In cases of an abusive, hostile, or simply totally vacuous relationship with the caregiver, the infant may deliberately turn away from the mentalizing object because the contemplation of the object’s mind is overwhelming as it harbors frankly hostile intentions toward the infant’s self’ (p. 84). For criticism of the necessity of such speculative attributions by the infant to the caregiver, see Fuchs, T. and De Jaegher, H. (2009). ‘Enactive Intersubjectivity: Participatory Sense-Making and Mutual Incorporation’. Phenomenology and the Cognitive Sciences, 8(4): 465–486.

22 E.g. Fonagy, P. (2003). ‘Clinical Implications of Attachment and Mentalization: Efforts to Preserve the Mind in Contemporary Treatment. Epilogue’. Bulletin of the Menninger Clinic, 67: 271–280: ‘Trauma may be defined subjectively as an experience that is intolerable to entertain: unmentalizable’ (p. 272).

23 The status of trauma in Fonagy’s early thinking as probably one sufficient but not necessary condition for problems with the capacity to conceptualize mental states is already visible in Fonagy, ‘On Tolerating Mental States’. This position would later be qualified: Fonagy, P., Bateman, A. W., and Luyten, P. (2012). ‘Introduction and Overview’, in Handbook of Mentalizing in Mental Health Practice, Washington, DC: American Psychiatric Publishing, pp. 3–42: ‘General characteristics of family function rather than maltreatment per se generate vulnerability to loss of mentalisation under stress’ (pp. 15–16); ‘the mentalisation-based treatment model does not attribute a central role to trauma … In our view the effect of trauma is most likely to be felt as part of a more general failure to consider the child’s perspective’ (p. 16).

24 Fonagy, P., Edgcumbe, R., Target, M., and Miller, J. (1999). Contemporary Psychodynamic Child Therapy: Theory and Technique, London: The Anna Freud Centre and University College London, pp. 46–47.

25 Fonagy, P. (1991). ‘Thinking about Thinking: Some Clinical and Theoretical Considerations in the Treatment of a Borderline Patient’. The International Journal of Psychoanalysis, 72: 639–656, p. 654.

26 Fonagy, P. (1995). ‘Playing with Reality: The Development of Psychic Reality and its Malfunction in Borderline Personalities’. The International Journal of Psychoanalysis, 76(1): 39–44, p. 43.

27 Fonagy, P. (1993). ‘Psychoanalytic and Empirical Approaches to Developmental Psychopathology: An Object-Relations Perspective’. Journal of the American Psychoanalytic Association, 41: 245–260: the capacity to think about mental states ‘equips the individual with a sort of ballast, a self-righting capacity where working models may become the object of review and change. Such gradual and constant adjustments give rise to an internal world where the behavior of objects can be experienced as predictable and stable, where the need for splitting of incoherent mental representations of the other is minimized and new experiences can be readily integrated with past internal representations’ (p. 256).

28 Fonagy, P. and Target, M. (2007). ‘Playing with Reality: IV. A Theory of External Reality Rooted in Intersubjectivity’. The International Journal of Psychoanalysis, 88: 917–937, p. 921. See also Fonagy, P. (2000). ‘Attachment and Borderline Personality Disorder’. Journal of the American Psychoanalytic Association, 48(4): 1129–1146: ‘The emptiness is a direct consequence of the absence of secondary representations of self states, certainly at the conscious level, and of the shallowness with which other people and relationships are experienced … Emptiness (and in extreme cases a sense of dissociation) is the best description such individuals can give of the absence of meaning that the failure of mentalization creates’ (p. 1140). Fonagy and colleagues may now regard this emptiness as expressing aspects of the primary unconscious—see Chapter 6. This would seem to be the position of Grotstein, J. S. (1984). ‘A Proposed Revision of the Psychoanalytic Concept of Primitive Mental States, Part II: The Borderline Syndrome-Section 2: The Phenomenology of the Borderline Syndrome’. Contemporary Psychoanalysis, 20(1): 77–119; Grotstein, J. S. (1990). ‘Nothingness, Meaninglessness, Chaos, and the “Black Hole” I—The Importance of Nothingness, Meaninglessness, and Chaos in Psychoanalysis’. Contemporary Psychoanalysis, 26: 257–290. However, unlike Fonagy and colleagues, Grotstein differentiates between emptiness as unintegration—the uncommitted potential for meaning-making—and emptiness as disintegration. The distinction is between a container with nothing yet in it and a broken container in which meaning has drained away.

29 Implicit in Fonagy’s early remarks on Baron-Cohen, this was brought out explicitly a few years later in Target, M. and Fonagy, P. (1996). ‘Playing with Reality: II. The Development of Psychic Reality from a Theoretical Perspective’. The International Journal of Psychoanalysis, 77: 459–479: ‘Our position is at odds with the majority of developmental psychologists (e.g. Leslie, 1987; Baron-Cohen, 1995). Current psychological theories stress mainly cognitive precursors of theory of mind. The child is seen as an isolated processor of information, engaged in the construction of a theory of mind from first principles, accumulating representations of the world through observation. From a psychoanalytic viewpoint this is a barren picture that ignores the central role of the child’s emotional relationship with the parents or other caregivers in fostering the capacity to understand interactions in terms of mental states . . . We may assume, as do most cognitivists working in this area, that the development of a theory of mind is canalized (Waddington, 1966) or prepared, but the canal is dug not by biology, but by infant-parent interaction’ (p. 461).

30 Fonagy, P. (1989). ‘On Tolerating Mental States: Theory of Mind in Borderline Patients’. Bulletin of the Anna Freud Centre, 12: 91–115: ‘Dunn (1988) considers four aspects of this development: 1) understanding others’ feelings, 2) understanding others’ intentions, 3) understanding rules and conventions and 4) understanding that others have minds’ (p. 91).

Components of social cognition for Dunn (1988)

Components of mentalizing for Fonagy (1989)

Understanding the origin of mental states

Understanding the origin of mental states

Understanding the location of mental states

Understanding the location of mental states

Understanding the functioning of mental states

Understanding the functioning of mental states

Understanding of social and cultural rules and conventions

31 Lacan, J. ([1953–1954] 1988). Freud’s Papers on Technique, trans. John Forrester, NY: Norton. Among other consequences, Fonagy’s evacuation of the symbolic rather depoliticized the concept of mentalizing. So, for instance, when Fonagy conceptualized conduct problems as caused by individual non-mentalizing (see Chapter 6), he did so without adequate attention to the role of inequalities in shaping conventions for the attribution of mental states and the role of institutionalized non-mentalizing. There was no intrinsic reason why a concern with individual perceptions of minds could not have been integrated with attention to social and cultural conventions. However, at the time, it was not, with the result that this concern has had to be rediscovered in the 2010s (see Chapter 9).

32 Fonagy, P. (1989). ‘On Tolerating Mental States: Theory of Mind in Borderline Patients’. Bulletin of the Anna Freud Centre, 12: 91–115, p. 97.

33 Fonagy, P. and Bateman, A. (2009). ‘A Brief History of Mentalisation-Based Treatment and its Roots in Psychoanalytic Theory and Practice’, in Brownescombe Heller, M. and Pollet, S. (eds), The Work of Psychoanalysts in the Public Health Sector, London: Routledge, pp. 156–176: ‘The first time we used the term “mentalization” was in 1989 (Fonagy 1989), influenced by the Ecole Psychosomatique de Paris, but we used the term as operationalized by developmental researchers investigating theory of mind (Leslie, 1987)’ (pp. 157–158). In fact, the term ‘mentalisation’ does not appear in Leslie: Leslie, A. M. (1987). ‘Pretense and Representation: The Origins of “theory of Mind”’. Psychological Review, 94(4): 412–426.

34 Luquet, P. (1981). ‘Le Changement dans la Mentalisation’. Revue Française de Psychoanalyse, 45: 1023–1028.

35 Marty, P. (1990). Mentalisation et Psychosomatique: Les Empecheurs de Penser en Rond, Paris. In recent years, Fonagy and colleagues have returned to the topic of somatic disorders, bringing to the topic their developed theory of mentalization—distinct from the École Psychosomatique de Paris. See e.g. Ballespí, S., Vives, J., Alonso, N., Sharp, C., Ramírez, M. S., Fonagy, P., and Barrantes-Vidal, N. (2019). ‘To Know or Not to Know? Mentalization as Protection from Somatic Complaints’. PloS One, 14(5): e0215308; Luyten, P., De Meulemeester, C., and Fonagy, P. (2019). ‘Psychodynamic Therapy in Patients with Somatic Symptom Disorder’, in David Kealy and John S. Ogrodniczuk (eds), Contemporary Psychodynamic Psychotherapy: Evolving Clinical Practice, New York: Academic Press, pp. 191–206.

36 Morton, J. and Frith, U. (1995). ‘Causal Modeling: A Structural Approach to Developmental Psychopathology’, in D. Cicchetti and D. J. Cohen (eds), Manual of Developmental Psychopathology, New York: Wiley, pp. 357–390: ‘The development of theory of mind (Premack and Woodruff, 1978) or “mentalising”—our ability to predict and explain the behaviour of other humans in terms of their mental states. Our ability to mentalise is revealed in our use and understanding of such words as believe, know, wish, desire, intend and pretend … Mentalising is primarily unconscious or implicit. It is a property of our cognitive apparatus that comes into action when triggered by particular stimuli, and it “makes sense” of other people’s and our own behaviour fully automatically’ (p. 363).

37 The common element between Marty and Bion is likely Freud, S. ([1912] 2001). ‘Formulations on the Two Principles of Mental Functioning’, in The Standard Edition of the Complete Psychological Works of Sigmund Freud, Volume 12 (1911–1913): The Case of Schreber, Papers on Technique and Other Works, pp. 213–226, London: Vintage. To an extent, these concerns had been dealt with by Kleinian psychoanalysis under the rubric of ‘symbolization’. However, as Fonagy observed, the concept was insufficiently precise for fine work: Fonagy, P. (1991). ‘Thinking about Thinking: Some Clinical and Theoretical Considerations in the Treatment of a Borderline Patient’. The International Journal of Psychoanalysis, 72: 639–656: ‘The achievement of a representation of mental events, whether conscious or unconscious, is frequently referred to in the psychoanalytic literature in the context of the capacity for symbolization (Segal, 1957); (McDougall, 1974); (Edgcumbe, 1984). The term symbolization is over-burdened with meanings, particularly in psychoanalysis. It is certainly not possible to restrict it to the notion of the secondary representation of mental states’ (p. 641).

38 Fonagy, P. and Higgitt, A. (1990). ‘A Developmental Perspective on Borderline Personality Disorder’. Revue Internationale de Psychopathologie, 1: 125–159, p. 140.

39 Fonagy, P. and Target, M. (1995). ‘Understanding the Violent Patient: The Use of the Body and the Role of the Father’. The International Journal of Psychoanalysis, 76: 487–501, p. 498.

40 Fonagy, P. (1999). ‘Interview with Peter Fonagy’, in S. M. Stein and J. Stein (eds), Psychotherapy in Practice: A Life in the Mind (pp. 77–98). Oxford: Butterworth Heinemann, p. 91.

41 See Goodman, G., Midgley, N., and Schneider, C. (2016). ‘Expert Clinicians’ Prototypes of an Ideal Child Treatment in Psychodynamic and Cognitive-Behavioral Therapy: Is Mentalization seen as a Common Process Factor?’ Psychotherapy Research, 26(5): 590–601. Recent work has found greater convergence in how CBT and psychoanalytic therapists respond to disengaged patients than how they respond to engaged patients. In terms of Fonagy and Hepworth’s hypothesis that mentalizing is the active ingredient in both cases, it might be supposed that the modalities have different strategies for sustaining mentalizing—for instance, use of transference or homework—but both used similar approaches for bringing mentalizing online. e.g. ‘they actively structured the sessions, asked many questions and tried to bring up material from previous sessions in an attempt to invite the young person to talk’. The researchers note that this strategy for responding to disengagement is mentioned in the treatment manuals of neither approach. Calderon, A., Schneider, C., Target, M., and Midgley, N. (2019). ‘“Interaction Structures” between Depressed Adolescents and their Therapists in Short-Term Psychoanalytic Psychotherapy and Cognitive Behavioural Therapy’. Clinical Child Psychology and Psychiatry, 24(3): 446–461, p. 457.

42 Fonagy, P. (2015). ‘I Would Like to Abolish Silo Working.’ The Psychologist, 28: 948. Accessed at:

43 Steele, H. and Steele, M. (2005). ‘Understanding and Resolving Emotional Conflict: The London Parent–Child Project’, in K. E. Grossmann, K. Grossmann, and E. Waters (eds), Attachment from Infancy to Adulthood: The Major Longitudinal Studies, New York: Guilford Press, pp. 137–164.

44 Main, M., Kaplan, N., and Cassidy, J. (1985). ‘Security in Infancy, Childhood, and Adulthood: A Move to the Level of Representation’. Monographs of the Society for Research in Child Development, 50(1–2): 66–104.

45 Ainsworth. M. D. S. (1983). ‘Patterns of Infant–Mother Attachment as Related to Maternal Care: Their Early History and their Contribution to Continuity’, in D. Magnusson and V. L. Allen (eds), Human Development: An Interactional Perspective, New York: Academic Press, pp. 35–55.

46 The challenge for psychological theory was evoked well by Bion in his Brazil lectures: ‘If we invent words, nobody will understand what those words mean. If we do not, their sensuous history is evoked. As with a “dead” metaphor carelessly used, its ghost begins to walk.’ Bion, W. R. ([1973] 1990). Brazilian Lectures, London: Karnac Books, p. 34. And, in any case, ‘attempts to contrive a term devoid of a distorting penumbra of associations is often defeated by the speed with which such a meaningless term collects a meaning’. Bion, W. R. (1962) Elements of Psychoanalysis, London: Karnac Books, p. 88. As a thought experiment, if Main and Solomon had set out their observations using a technical vocabulary rather than the evocative language of ‘disorganisation’ and ‘fear’, it seems likely that it would have had only a fraction of the take-up outside attachment research. At the same time, a price was paid in simplifying the concepts beyond all recognition. Duschinsky, R. (2018). ‘Disorganization, Fear and Attachment: Working towards Clarification’. Infant Mental Health Journal, 39(1): 17–29.

47 Reijman, S., Foster, S., and Duschinsky, R. (2018). ‘The Infant Disorganised Attachment Classification: “Patterning within the Disturbance of Coherence”’. Social Science & Medicine, 200: 52–58.

48 Carlson, V., Cicchetti, D., Barnett, D., and Braunwald, K. (1989). ‘Finding Order in Disorganization: Lessons from Research on Maltreated Infants’ Attachments to their Caregivers’, in D. Cicchetti and V. Carlson (eds), Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect, New York: Cambridge University Press, pp. 494–528.

49 Main, M. and Hesse, E. (1990). ‘Parents’ Unresolved Traumatic Experiences are Related to Infant Disorganized Attachment Status: Is Frightened and/or Frightening Parental Behavior The Linking Mechanism?’, in M. T. Greenberg, D. Cicchetti, and E. M. Cummings (eds), Attachment in the 638 Child Development Preschool Years, Chicago: University of Chicago Press, pp. 161–182. On the antecedents of disorganized attachment, see also Cyr, C., Euser, E. M., Bakermans-Kranenburg, M. J., and Van IJzendoorn, M. H. (2010). ‘Attachment Security and Disorganization in Maltreating and High-Risk Families: A Series of Meta-Analyses’. Development and Psychopathology, 22(1): 87–108; Granqvist, P., Sroufe, L. A., Dozier, M., Hesse, E., Steele, M., van IJzendoorn, M., … and Duschinsky, R. (2017). ‘Disorganized Attachment in Infancy: A Review of the Phenomenon and its Implications for Clinicians and Policy-Makers’. Attachment & Human Development, 19(6): 534–558.

50 Fonagy, P. (1999). ‘Points of Contact and Divergence between Psychoanalytic and Attachment Theories: Is Psychoanalytic Theory Truly Different?’ Psychoanalytic Inquiry, 19(4): 448–480, p. 452.

51 Steele, H. and Steele, M. (2005). ‘Understanding and Resolving Emotional Conflict: The London Parent–Child Project’, in K. E. Grossmann, K. Grossmann and E. Waters (eds), Attachment from Infancy to Adulthood: The Major Longitudinal Studies, New York: Guilford Press, pp. 137–164.

52 Fonagy, P., Steele, M., Moran, G. S., Steele, H., and Higgitt, A. C. (1991). ‘Measuring the Ghost in the Nursery: A Summary of the Main Findings of the Anna Freud Centre/University College London Parent–Child Study’. Bulletin of the Anna Freud Centre, 14: 115–131.

53 Steele, H. and Steele, M. (2005). ‘Understanding and Resolving Emotional Conflict: The London Parent–Child Project’, in K. E. Grossmann, K. Grossmann, and E. Waters (eds), Attachment from Infancy to Adulthood: The Major Longitudinal Studies, New York: Guilford Press, pp. 137–164, p. 157.

54 Steele, H. and Steele, M. (2008). ‘On the Origins of Reflective Functioning’, in F. Busch (ed.), Mentalization: Theoretical Considerations, Re-Search Findings, and Clinical Implications, New York: Taylor and Francis, pp. 133–158.

55 Steele, H. and Steele, M. (2005). ‘Understanding and Resolving Emotional Conflict: The London Parent–Child Project’, in K. E. Grossmann, K. Grossmann, and E. Waters (eds), Attachment from Infancy to Adulthood: The Major Longitudinal Studies, New York: Guilford Press, pp. 137–164, p. 157.

56 As in the subtitle of Fonagy, P., Steele, M., Steele, H., Moran, G. S., and Higgitt, A. C. (1991). ‘The Capacity for Understanding Mental States: The Reflective Self in Parent and Child and its Significance for Security of Attachment’. Infant Mental Health Journal, 12(3): 201–218.

57 This usage was influenced by Bollas, C. (1990). ‘The Origins of the Therapeutic Alliance’. Paper presented at the English Speaking Conference of the British Psychoanalytical Society, October, London. Later published as Bollas, C. (1998). ‘Origins of the Therapeutic Alliance’. Scandinavian Psychoanalytic Review, 21(1): 24–36.

58 Duschinsky, R. (2020). Cornerstones of Attachment Research, Oxford: Oxford University Press.

59 Fonagy and Target would acknowledge alignment between the idea of reflective function and that of alpha function, though not any relation of influence e.g. Fonagy, P., Target, M., Steele, H., and Steele, M. (1998). Reflective-Functioning Manual, Version 5, London: University College London/Anna Freud Centre, p. 4; Fonagy, P. and Target, M. (2003). ‘Evolution of the Interpersonal Interpretive Function: Clues for Effective Preventive Intervention in Early Childhood’, in S. W. Coates, J. L. Rosenthal, and D. S. Schechter (eds), September 11: Trauma and Human Bonds, Hillsdale, NJ: Analytic Press, pp. 99–113, p. 102. The developmental model of reflective functioning likewise has marked similarities to the model proposed by Bion. Though since Fonagy and Target do not acknowledge influence, it is difficult to know whether this can inferred. See e.g. Bion, W.R. ([1973] 1990). Brazilian Lectures, London: Karnac Books: ‘If the baby has a mother who is able to tolerate being frightened that the child is dying, or that it is mad, or stupid, then the baby seems to feel better for having such a mother. If she cannot tolerate it, neither can the baby who consequently seems unable to grow mentally. If it is going to grow it has to do so in a peculiar way; this, in turn, has the effect of making it grow a certain shape. Later, someone will say … “borderline case” ’ (p. 132).

60 Sharp, C. and Fonagy, P. (2008). ‘The Parent’s Capacity to Treat the Child as a Psychological Agent: Constructs, Measures and Implications for Developmental Psychopathology’. Social Development, 17(3): 737–754: ‘What exactly does mentalizing mean within this context? … When this capacity is operationalized within the context of attachment relationships, Fonagy refers to it as “reflective functioning” (RF) … The term is used to describe the parent’s capacity to reflect upon his/her own or the child’s internal mental experience within the context of attachment’ (p. 740).

61 Fonagy, P., Steele, M., Steele, H., Moran, G. S., and Higgitt, A. C. (1991). ‘The Capacity for Understanding Mental States: The Reflective Self in Parent and Child and its Significance for Security of Attachment.’ Infant Mental Health Journal, 12(3): 201–218.

62 Ibid. 215.

63 Fonagy, P., Target, M., Steele, H., and Steele, M. (1998). Reflective-Functioning Manual, Version 5, London: University College London/Anna Freud Centre.

64 This point seems to be explicitly affirmed in Murphy, A., Steele, M., and Steele, H. (2013). ‘From Out of Sight, Out of Mind to In Sight and In Mind: Enhancing Reflective Capacities in a Group Attachment-Based Intervention’, in Bettmann Joanna Ellen and Friedman Donna Demetri (eds), Attachment-Based Clinical Work with Children and Adolescents, NY: Springer, pp. 237–257: ‘We define reflective functioning as (1) an awareness of the nature of mental states in the self and others; (2) the mutual influences at work between mental states and behaviour; (3) the necessity of a developmental perspective; and (4) the need to be sensitive to the current context’ (pp. 238–239). Discrepancy between the characterizations of reflective function and mentalization, despite claims by Fonagy and colleagues that they relate to the same construct, have also been noted by Eagle, M. (2013). Attachment and Psychoanalysis, New York: Guilford Press, p. 184. For Eagle, mentalization requires treating thoughts and feelings as intentions; reflective function addresses thoughts and feelings without the assumption of intentionality. Eagle also sees mentalization as implicit compared with reflective function as an explicit process. In our view, both contrasts are important, and there are some uses of the concepts that fit this contrast. However, at other times, reflective function is treated as implicit and assuming intentionality, and mentalization is treated as explicit and concerned with thoughts and feelings in general. Eagle’s characterization therefore does not hold, though he is certainly correct to identify these two axes of underlying ambiguity.

65 Fonagy, P., Target, M., Steele, H., and Steele, M. (1998). Reflective-Functioning Manual, Version 5, London: University College London/Anna Freud Centre.

67 Fonagy, P. (1996). ‘The Significance of the Development of Metacognitive Control over Mental Representations in Parenting and Infant Development’. Journal of Clinical Psychoanalysis, 5(1): 67–86, p. 74.

68 Cf. Cassidy, J., Jones, J. D., and Shaver, P. R. (2013). ‘Contributions of Attachment Theory and Research: A Framework for Future Research, Translation, and Policy’. Development and Psychopathology, 25(4.2): 1415–1434: ‘Consideration of linkages between representational and nonrepresentational processes must include the possibility that causality flows in both directions: physiological stress responses can presumably prompt a person to engage in higher level cognitive processes to understand, justify, or eliminate the stressor’ (p. 1419).

69 The exception would be where conventions are established to infer mental states from physiology or endocrinology. E.g. McHugh, B., Dawson, N., Scrafton, A., and Asen, E. (2010). ‘Hearts on Their Sleeves’: The Use of Systemic Biofeedback in School Settings’. Journal of Family Therapy, 32(1): 58–72; Fotopoulou, A. and Tsakiris, M. (2017). ‘Mentalizing Homeostasis: The Social Origins of Interoceptive Inference’. Neuropsychoanalysis, 19(1): 3–28. The distinction between the physical and the mental in the case of gender has been described by Fonagy as ‘wrong’ and ‘inappropriate’. However, he has not yet been clear about whether attention to embodied processes below the level of mental states would constitute mentalizing. See Fonagy’s remarks in Steidinger, S. (2018) ‘Trans-Actions: An Exploration of Gender Dysphoria’. Accessed at:

70 Ainsworth, M. (1969). ‘Scale 1: Sensitivity vs. Insensitivity to the Baby’s Signals’. Accessed at: Some but not all aspects of the sensitivity scale and construct were described in Stayton, D. J., Hogan, R., and Ainsworth, M. (1971). ‘Infant Obedience and Maternal Behavior: The Origins of Socialization Reconsidered’. Child Development, 42(4): 1057–1069, pp. 1060–1061.

71 Some of these concerns are discussed variously in Kondo-Ikemura, K. (2001). ‘Insufficient Evidence’. American Psychologist, 56(10): 825; Keller, H. and Otto, H. (2009). ‘The Cultural Socialization of Emotion Regulation During Infancy’. Journal of Cross-Cultural Psychology, 40(6): 996–1011; Shai, D. and Belsky, J. (2017). ‘Parental Embodied Mentalizing: How the Nonverbal Dance between Parents and Infants Predicts Children’s Socio-Emotional Functioning’. Attachment & Human Development, 19(2): 191–219.

72 Woodhouse, S. S., Scott, J. R., Hepworth, A. D., and Cassidy, J. (2020). ‘Secure Base Provision: A New Approach to Examining Links Between Maternal Caregiving and Infant Attachment’. Child Development, 91(1): e249–e265 .

73 Fonagy, P., Steele, H., Steele, M. and Holder, J. (1997). ‘Attachment and Theory of Mind: Overlapping Constructs?’ Association for Child Psychology and Psychiatry Occasional Papers, 14: 31–40: for Ainsworth, ‘the caregiver’s perception of the child as an intentional being lies at the root of sensitive caregiving’ (p. 36).

74 Evaluating the criticisms of Ainsworth by Fonagy and colleagues, Zeegers and colleagues conducted a meta-analysis to investigate the relative contributions of sensitivity and mentalization/reflective function to infant attachment classifications. They found that together the two predictors accounted for 12% of variance in attachment classifications. After controlling for sensitivity, the relationship between parental mentalization and infant–caregiver security was r = .24. And, after controlling for mentalization, the relationship between parental sensitivity and infant–caregiver security was r = .19. Sensitivity also partially mediated the association between mentalization and infant–caregiver security (r = .07). Such findings suggest that Fonagy and others were right to argue for the importance of mentalization/reflective function, but that sensitivity is not reducible to reflective function. Zeegers, M. A., Colonnesi, C., Stams, G. J. J., and Meins, E. (2017). ‘Mind Matters: A Meta-Analysis on Parental Mentalization and Sensitivity as Predictors of Infant–Parent Attachment’. Psychological Bulletin, 143(12): 1245–1272.

75 Bowlby, K. (1991). ‘Ethological Light on Psychoanalytical Problems’, in P. Bateson (ed.), The Development and Integration of Behaviour: Essays in Honour of Robert Hinde, Cambridge: Cambridge University Press, pp. 301–313; Stevenson-Hinde, J. (2005). ‘The Interplay between Attachment, Temperament and Maternal Style: A Madingly Perspective’, in K. E. Grossmann, K. Grossmann and E. Waters (eds.), Attachment from Infancy to Adulthood: The Major Longitudinal Studies, New York: Guilford Press, pp. 198–222.

76 Egeland, B., Jacobvitz, D., and Sroufe, L. (1988). ‘Breaking the Cycle of Abuse’. Child Development, 59(4): 1080–1088.

77 Offering some support for this conclusion, researchers would find a large correlation between caregivers’ reflective functioning in the Adult Attachment Interview and the absence of frightening, dissociative, anomalous, withdrawing behaviours or severe disruptions to parent–child communication during filmed observation. Fonagy, P. and Target, M. (2005). ‘Bridging the Transmission Gap: An End to an Important Mystery of Attachment Research?’ Attachment & Human Development, 7(3): 333–343.

78 Fonagy, P. (2001). ‘The Human Genome and the Representational World: The Role of Early Mother–Infant Interaction in Creating an Interpersonal Interpretive Mechanism’. Bulletin of the Menninger Clinic, 65(3): 427–448, p. 436.

79 Fónagy, I. and Fonagy, P. (1995). ‘Communication with Pretend Actions in Language, Literature and Psychoanalysis’. Psychoanalysis and Contemporary Thought, 18(3): 363–418. In the later formulation of Bateman, A. W. and Fonagy, P. (2004). Psychotherapy for Borderline Personality Disorders: Mentalization-Based Treatment, Oxford: Oxford University Press: ‘We have to assume a dialectic model of self-development (Hegel 1807) where the child’s capacity to create a coherent image of mind is critically dependent on an experience of being clearly perceived as a mind by the attachment figure’ (p. 64).

80 Gergely, G. and Watson, J. (1996). ‘The Social Biofeedback Model of Parental Affect Mirroring. The Development of Emotional Self-Awareness and Self-Control in Infancy’. The International Journal of Psychoanalysis, 77: 1–31. A difference between Gergely and Fonagy in this regard is that the latter quite frequently translated the idea of marked mirroring into the Kleinian vocabulary of the caregiver offering ‘containment’ for the child’s affect. Kleinians, especially Bion, had certainly offered aligned ideas over the previous decades, though usually phrased in the language of projective identification. For instance, Bion wrote that ‘An understanding mother is able to experience the feeling of dread, that this baby was striving to deal with by projective identification, and yet retain a balanced outlook. This patient had had to deal with a mother who could not tolerate experiencing such feelings and reacted either by denying them ingress, or alternatively by becoming a prey to the anxiety which resulted from introjection of the infant’s feelings … a well-balanced mother can accept these and respond therapeutically: that is to say in a manner that makes the infant feel it is receiving its frightened personality back again but in a form that it can tolerate’,Bion, W. R. (1962). ‘The Psycho-Analytic Study of Thinking’. The International Journal of Psychoanalysis, 43: 306–310, p. 308. One subsequent criticism that Fonagy and colleagues have raised regarding the concept of marked mirroring is that it underplays the irreducibly kinaesthetic aspects of the process: Shai, D. and Fonagy, P. (2014). ‘Beyond Words: Parental Embodied Mentalizing and the Parent Infant Dance’, in M. Mikulincer and P. R. Shaver (eds), Mechanisms of Social Connections: From Brain to Group, Washington, DC: American Psychological Association, pp. 185–203. More than marked mirroring, the revised account in Shai and Fonagy emphasizes repair of interactive missteps as the paradigmatic form in which parental reflective function contributes to child attachment security, felt coherence, and socio-emotional development. See also Shai, D. and Meins, E. (2018). ‘Parental Embodied Mentalizing and its Relation to Mind‐Mindedness, Sensitivity, and Attachment Security’. Infancy, 23(6): 857–872.

81 Fonagy, P. (1999). ‘Points of Contact and Divergence between Psychoanalytic and Attachment Theories: Is Psychoanalytic Theory Truly Different?’ Psychoanalytic Inquiry, 19(4): 448–480, p. 469.

82 Ibid. 460.

83 Fonagy, P. and Target, M. (2002). ‘Early Intervention and the Development of Self-Regulation’. Psychoanalytic Inquiry, 22(3): 307–335, p. 328.

84 Fonagy, P. (1998). ‘Moments of Change in Psychoanalytic Theory: Discussion of a New Theory of Psychic Change’. Infant Mental Health Journal, 19(3): 346–353: ‘At the heart of this new theory is the notion of implicit or procedural memory borrowed from cognitive science (Schacter, 1992). Bob Clyman (1991) should be credited with bringing this idea to the attention of psychoanalysts, and Crittenden (1990) with integrating the idea with attachment theory’ (p. 348).

85 Fonagy, P. and Bateman, A. W. (2006). ‘Mechanisms of Change in Mentalization‐Based Treatment of BPD’. Journal of Clinical Psychology, 62(4): 411–430. See also Mayes, L. C. (2006). ‘Arousal Regulation, Emotional Flexibility, Medial Amygdala Function, and the Impact of Early Experience: Comments on the Paper of Lewis et al’. Annals of the New York Academy of Sciences, 1094: 178–192.

86 Fonagy, P. and Adshead, G. (2012). ‘How Mentalisation Changes the Mind’. Advances in Psychiatric Treatment, 18(5): 353–362, p. 359.

87 Fonagy, P. and Target, M. (2005). ‘Some Reflections on the Therapeutic Action of Psychoanalytic Therapy’, in J. Auerbach, K. Levy, and C. E. Shaffer (eds), Relatedness, Self-definition and Mental Representation: Essays in Honor of Sidney J. Blatt, New York: Taylor & Francis, pp. 191–212, p. 200, italics added.

88 E.g. Fonagy, P. ([1997] 2002). ‘Multiple Voices versus Meta-Cognition: An Attachment Theory Perspective’, in V. Sinason (ed.), Attachment, Trauma and Multiplicity: Working with Dissociative Identity Disorder, London: Brunner-Routledge, pp. 71–85. The in-but-mostly-out status of dissociation within Fonagy’s explanatory model would later be a point of acknowledged self-criticism e.g. Ensink, K., Bégin, M., Normandin, N., Godbout, N., and Fonagy, P. (2017). ‘Mentalization and Dissociation in the Context of Trauma: Implications for Child Psychopathology’. Journal of Trauma & Dissociation, 18(1): 11–30: ‘The finding that dissociation is central in processes associated with externalizing and sexualizing behaviors, partly by itself and partly with RF, nuanced our predictions based on the model proposed by Bateman and Fonagy (2008) in which mentalizing is considered to be the key mental process that has a social regulation role … dissociation was not considered in Bateman and Fonagy’s conceptual model’ (p. 24).

89 However, already in these writings, non-apprehensive conflict was downplayed in favour of an emphasis on fear and confusion. See e.g. Fonagy, P., Steele, M., Steele, H., Moran, G. S., and Higgitt, A. C. (1991). ‘The Capacity for Understanding Mental States: The Reflective Self in Parent and Child and its Significance for Security of Attachment’. Infant Mental Health Journal, 12(3): 201–218: ‘A small group of infants show a fourth pattern of response: one of confusion and disorganization (D)’ (p. 206).

90 Fonagy, P. and Target, M. (2003). Psychoanalytic Theories: Perspectives from Developmental Psychopathology, London: Whurr Publications, p. 244.

91 Duschinsky, R. and Solomon, J. (2017). ‘Infant Disorganized Attachment: Clarifying Levels of Analysis’. Clinical Child Psychology and Psychiatry, 22(4): 524–538; Reijman, S., Foster, S., and Duschinsky, R. (2018). ‘The Infant Disorganised Attachment Classification: “Patterning within the Disturbance of Coherence”’. Social Science & Medicine, 200: 52–58. On the ‘concept creep’ of ideas like disorganized attachment that signify psychological harm, see also Haslam, N. (2016). ‘Concept Creep: Psychology’s Expanding Concepts of Harm and Pathology’. Psychological Inquiry, 27(1): 1–17.

92 Target, M., Mayes, L., and Bach, S. (2000). ‘Panel 4: The Pathology of the Self: The Fragmented Self, Disorders of the Self, and the Dissolution of the Self’. Journal of Infant, Child and Adolescent Psychotherapy, 1(3): 63–72. Hepworth remarked that ‘it is interesting to note Breuer and Freud’s (1895) comment in Studies on Hysteria: “It is easy to fall into a habit of thought which assumes that every substantive has substance behind it. We find as time goes on, that we have actually formed an idea which has lost its metaphorical nature, and which we can manipulate easily, as though it were real” (pp. 227–228)’ (p. 63). See also Reijman, S., Foster, S., and Duschinsky, R. (2018). ‘The Infant Disorganised Attachment Classification: “Patterning within the Disturbance of Coherence”’. Social Science & Medicine, 200: 52–58.

93 E.g. Fonagy, P., Gergely, G., Jurist, E. L., and Target, M. (2002). Affect Regulation, Mentalization, and the Development of the Self, London: Karnac Books: ‘At the extreme end of the safety-to-fearfulness dimension, there can be no strategy because the attachment system is not there to sustain a consistent set of defences. In these cases the interpretive mechanism that sustains social relations functions so poorly that the capacity to arrive at representations of the motivational or epistemic mind states of the other, independent of those of the self, is profoundly compromised. This is attachment disorganisation or, rather, the absence of the mental function that sustains attachment. Thus we conceive of attachment disorganisation as lying at the opposite end of the scale to attachment security and as an indicator of the regular failure of the interpersonal interpretive mechanism’ (pp. 135–136). Fonagy has subsequently acknowledged that the 2002 book was ‘patchy’. One of the most powerful contributing problems is the multiple and conflicting ways that the different chapters use the concepts of disorganization and mentalizing. BBC Radio 4 (2020). ‘Peter Fonagy on a Revolution in Mental Health Care’. The Life Scientific Podcast, 28 January. Accessed at:

94 Fonagy, P., Luyten, P., Bateman, A. W., Gergely, G., Strathearn, L., Target, M., and Allison, E. (2010). ‘Attachment and Personality Pathology’, in J. Clarkin, P. Fonagy, and G. Gabbard (eds), Psychodynamic Psychotherapy for Personality Disorders: A Clinical Handbook, Arlington, VA: American Psychiatric Publishing, Inc., pp. 37–88, 40–41.

95 Ainsworth, M., Blehar, M., Waters, E., and Wall, S. ([1978] 2015). Patterns of Attachment: A Psychological Study of the Strange Situation, Bristol: Psychology Press, p. 347.

96 Fonagy, P., Luyten, P., Bateman, A. W., Gergely, G., Strathearn, L., Target, M., and Allison, E. (2010). ‘Attachment and Personality Pathology’, in J. Clarkin, P. Fonagy, and G. Gabbard (eds), Psychodynamic Psychotherapy for Personality Disorders: A Clinical Handbook, Arlington, VA: American Psychiatric Publishing, Inc., pp. 37–88, 40–41.

97 Fonagy, P., Luyten, P., Allison, E., and Campbell, C. (2017). ‘What We have Changed our Minds about: Part 1. Borderline Personality Disorder as a Limitation of Resilience’. Borderline Personality Disorder and Emotion Dysregulation, 4: 11: ‘We have previously placed considerable weight on the nature of attachment disorganization in our accounts of BPD based on the mentalizing model. We maintain that the role of attachment is highly significant in the developmental origins of PD. However, we argue that its role might perhaps be best understood as only one (albeit very important) form of content learned from the social environment. This is congruent with recent work suggesting that the relationship between infant attachment status and later outcomes is more complicated than that suggested by early attachment studies. Other findings have suggested limited evidence for linking childrearing environments to later outcomes.’

98 Fonagy was also a strong advocate for the category to professional groups, above all clinicians and teachers: e.g. speaking to the House of Commons Education Committee (2016) ‘Mental Health and Well-Being of Looked After Children’, HC 481’, Wednesday, 13 January. Accessed at: ‘Here is agreed risk and also an opportunity for teachers to benefit from understanding problems that arise in relation to disrupted attachment and the disorganisation of attachment. In the NICE [National Institute for Health and Care Excellence] guidance development group that I chaired, one of the recommendations that we identified was in relation to training teachers to be more aware of the attachment issues.’

99 Duschinsky, R. (2015). ‘The Emergence of the Disorganised/Disoriented (D) Attachment Classification, 1979–1982’. History of Psychology, 18(1): 32–46.

100 Fonagy, P. and Higgitt, A. (2007). ‘The Early Social and Emotional Determinants of Inequalities in Health’, in G. Baruch, P. Fonagy, and D. Robins (eds), Reaching the Hard to Reach: Evidence-Based Funding Priorities for Intervention and Research, Chichester: John Wiley & Sons, pp. 3–34, p. 13.

101 This claim seems to be based on a conflation of disorganized attachment as an assessment for the Strange Situation with the co-occurrence of anxiety and avoidance in self-report measures of attachment. In the early 1990s, Shaver and colleagues claimed that the co-occurrence of attachment anxiety and avoidance is analogous to Main and Solomon’s disorganized attachment classification. This was based on a proposal by Bartholomew, who called this co-occurrence ‘fearful attachment’. However, examination of Barthlomew’s 1989 unpublished doctoral thesis indicates an important qualification: Bartholomew regarded fearful attachment not as a general analogue for the disorganized attachment classification, but as an analogue for approach–avoidance conflict specifically. Jacobvitz and colleagues criticized the assumption by Shaver and others that disorganized attachment is the same as co-occurrence of attachment anxiety and avoidance. Brennan, K. A., Shaver, P. R., and Tobey, A. E. (1991). ‘Attachment Styles, Gender, and Parental Problem Drinking’. Journal of Social and Personal Relationships, 8: 451–466; Bartholomew, K. (1989). Attachment Styles in Young Adults: Implications for Self-Concept and Interpersonal Functioning, unpublished doctoral dissertation, Stanford University; Jacobvitz, D., Curran, M., and Moller, N. (2002). ‘Measurement of Adult Attachment: The Place of Self-Report and Interview Methodologies’. Attachment & Human Development, 4(2): 207–215, p. 209. There is also no prospective or concurrent association between measures of disorganized attachment and co-occurrence of anxiety and avoidance in self-report measures of attachment. Roisman, G. I., Holland, A., Fortuna, K., Fraley, R. C., Clausell, E., and Clarke, A. (2007). ‘The Adult Attachment Interview and Self-Reports of Attachment Style: An Empirical Rapprochement’. Journal of Personality and Social Psychology, 92(4): 678–697; Fraley, R. C., Roisman, G. I., Booth-LaForce, C., Owen, M. T., and Holland, A. S. (2013). ‘Interpersonal and Genetic Origins of Adult Attachment Styles: A Longitudinal Study from Infancy to Early Adulthood’. Journal of Personality and Social Psychology, 104(5): 817–838, web-based supplement C.

102 Some infants show co-occurrence of avoidance and resistant attachment, and on this basis receive a disorganized classification. But this is hardly the predominant phenotype. A rare form of behaviour under Index I of the Main and Solomon coding protocols is hyperarousal followed by relatively abrupt downregulation. However, this is by no means a behaviour representative of the category as a whole. Most child–caregiver dyads who receive a disorganized classification do so on the basis of other forms of behaviour in the Main and Solomon indices, such as other kinds of approach/avoidance conflict, stereotypies or confusion, or apprehension without any apparent conflict. The predominance of different forms is anticipated to depend somewhat on the kinds of adversities faced by the sample. Solomon, J., Duschinsky, R., Bakkum, L., and Schuengel, C. (2017). ‘Toward an Architecture of Attachment Disorganization: John Bowlby’s Published and Unpublished Reflections’. Clinical Child Psychology and Psychiatry, 22(4): 539–560.

103 Lyons-Ruth and colleagues have criticized the characterization of disorganized attachment as the absence of any strategy to interpersonal relationships, observing that a majority of infants who receive the classification nonetheless show proximity-seeking and contact maintenance with their caregiver when distressed, in the manner of the secure category. Lyons-Ruth, K., Bureau, J. F., Easterbrooks, M. A., Obsuth, I., Hennighausen, K., and Vulliez-Coady, L. (2013). ‘Parsing the Construct of Maternal Insensitivity: Distinct Longitudinal Pathways Associated with Early Maternal Withdrawal’. Attachment & Human Development, 15(5–6): 562–582.

104 For instance, a toddler who is solicitous and caring towards their parent when distressed is not showing an avoidant or resistant strategy, and would likely be placed in the disorganized classification, despite the fact that their behaviour is strategic and their affects may be coherent in their expression (Chapter 7). Mary Main and Erik Hesse, personal communication, August 2019: ‘Development allows humans to override a behavioural system in other ways than the two conditional strategies, producing a wider variety of potential strategies than those available to infants. These might well not be conditional strategies in the technical sense of being a behavioural repertoire made available by human evolutionary history for solving problems of survival and reproduction. They could be described as “strategic” in the non-technical sense—but it depends on how the word is being used.’

105 For instance: the claim that all disorganized attachment represents a preoccupation with perfect contingencies in attachment relationships, emphasized especially by Fonagy and colleagues in the early 2000s, seems to have been downplayed over the past decade. One reason is that the evidence base for the claim has not expanded beyond Gergely’s original studies. The primary subsequent source of supportive evidence comes from the observation of poor caregiver–infant contingency among infants who would later receive a disorganized classification in the Strange Situation by Beebe, B., Jaffe, J., Markese, S., Buck, K., Chen, H., Cohen, P., . . . and Feldstein, S. (2010). ‘The Origins of 12-Month Attachment: A Microanalysis of 4-Month Mother–Infant Interaction’. Attachment & Human Development, 12(1–2): 3–141. However, there would be a variety of other ways of interpreting this finding, besides the idea that all disorganization reflects preoccupation with perfect contingencies. See e.g. Bernier, A. and Meins, E. (2008). ‘A Threshold Approach to Understanding the Origins of Attachment Disorganization’. Developmental Psychology, 44(4): 969–982; van IJzendoorn, M. H. and Bakermans-Kranenburg, M. J. (2019). ‘Bridges across the Intergenerational Transmission of Attachment Gap’. Current Opinion in Psychology, 25: 31–36.

106 Bateman, A. W. and Fonagy, P. (2004). Psychotherapy for Borderline Personality Disorders: Mentalization Based Treatment, Oxford: Oxford University Press, p. 90.

107 Fonagy, P. (1999). ‘Points of Contact and Divergence between Psychoanalytic and Attachment Theories: Is Psychoanalytic Theory Truly Different?’ Psychoanalytic Inquiry, 19(4): 448–480, p. 467.

108 Fonagy, P. and Target, M. (2006). ‘The Mentalization-Focused Approach to Self Pathology’. Journal of Personality Disorders, 20: 544–576: ‘Disorganised infants, even if interpersonally perceptive, fail to integrate this emotional awareness with their self-organisation’ (p. 565).

109 Padrón, E., Carlson, E., and Sroufe, A. (2014). ‘Frightened Versus Not Frightened Disorganized Infant Attachment: Newborn Characteristics and Maternal Caregiving’. American Journal of Orthopsychiatry, 84(2): 201–208.

110 See also Duschinsky, R. (2018). ‘Disorganization, fear and attachment: Working towards clarification’. Infant Mental Health Journal, 39(1): 17–29.

111 Kaplan, N. (1987). Individual Differences in 6-Years Olds’ Thoughts about Separation: Predicted from Attachment to Mother at Age 1. Unpublished doctoral dissertation, Berkeley, CA: Department of Psychology, University of California.

112 Main, M. and Cassidy, J. (1988). ‘Categories of Response to Reunion with the Parent at Age 6’. Developmental Psychology, 24(3): 415–426. It should also be noted that subsequent findings since Main have found much less stability than the original study between disorganized attachment in the Strange Situation and later representational measures (r =. 26). Pinquart, M., Feußner, C., and Ahnert, L. (2013). ‘Meta-Analytic Evidence for Stability in Attachments from Infancy to Early Adulthood’. Attachment & Human Development, 15(2): 189–218. An important reason for this difference was that Main and colleagues developed their story-stem coding system quasi-inductively on the sample, and then examined its retrodiction of the Strange Situation. Associations between the Strange Situation and the story-stem narratives at age 6 was therefore not an independent finding but a methodological principle for the elaboration of the 6-year coding system. It should not be assumed that disorganized attachment can be treated as meaning the absence of stable representation of interactions and therefore predictability.