Show Summary Details
Page of

(p. 337) Alan Sroufe, Byron Egeland, and the Minnesota Longitudinal Study of Risk and Adaptation 

(p. 337) Alan Sroufe, Byron Egeland, and the Minnesota Longitudinal Study of Risk and Adaptation
Chapter:
(p. 337) Alan Sroufe, Byron Egeland, and the Minnesota Longitudinal Study of Risk and Adaptation
Author(s):

Robbie Duschinsky

DOI:
10.1093/med-psych/9780198842064.003.0004
Page of

PRINTED FROM OXFORD CLINICAL PSYCHOLOGY ONLINE (www.oxfordclinicalpsych.com). © Oxford University Press, 2020. 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: 27 September 2020

Biographical sketch

The Minnesota Longitudinal Study of Risk and Adaptation began in the early 1970s, initiated by Byron Egeland as a study of a large cohort of mothers living in poverty. In the 1970s and early 1980s, empirical attachment research had been primarily pursued by Ainsworth and her students. The research group led by Alan Sroufe and Egeland was important in providing a second pillar to hold up the paradigm from the 1970s to the present. The Strange Situation was conducted with the sample by Sroufe and his graduate students Everett Waters and Brian Vaughn. Early work by the research group documented the role of caregiving in shaping patterns of attachment in the Strange Situation, and also the capacity of infant attachment patterns to predict later social competence and mental health. Sroufe and Egeland created an ‘electric atmosphere’ in their research group, as they provided the first longitudinal evidence of the implications of attachment relationships.1 Students described their ‘imperturbable optimism’, ‘wisdom about human nature’, and ‘compassion’ as important qualities in the creation of the atmosphere, along with the sense of contributing to meaningful and cutting-edge developmental science.2 They were a great stabilising and integrative presence for the field of attachment research. Though Egeland and Sroufe have now retired, research with the Minnesota Longitudinal Study of Risk and Adaptation has continued.

Introduction

The Minnesota research group has been of critical importance to the establishment and development of attachment research. Whilst to the general public Bowlby remains the most visible and prominent representative of attachment research, within the American developmental science community this role has arguably been held by Sroufe and Egeland. Mary Ainsworth wrote to Sroufe in 1982, expressing her deep gratitude: ‘One of the very best things that ever happened to attachment research was your decision to participate in it!’3 In the introduction to their six-volume edited work on attachment theory, Slade and Holmes described the contribution of the Minnesota group as no less than a ‘revolution,’ one that ‘brought attachment theory into mainstream academic psychology’.4 In part as a consequence, attachment theory has at times been referred to as the Bowlby–Ainsworth–Sroufe (p. 338) tradition.5 The primary scientific contribution of the Minnesota group under the leadership of Sroufe and Egeland has been to show the developmental implications of early caregiving and attachment. This was of essential importance for consolidating Ainsworth’s findings and for showing why they mattered for development. Furthermore, on the basis of their empirical research, Sroufe, Egeland, and colleagues made important theoretical contributions to the conceptualisation of the role of attachment within development as a whole. Fonagy would describe Sroufe and the Minnesota group as having played a ‘seminal role in extending the scope of attachment theory from an account of the developmental emergence of a set of social expectations to a far broader conception of attachment as an organizer of physiological and brain regulation’.6

In the 1940s, Bowlby felt unable to report on his clinical observations of child abuse (Chapter 1). By the 1970s, however, the clinical community had come to acknowledge the potential for parents to abuse and neglect their children.7 There was growing policy interest in the role of abuse and neglect in contributing to later mental illness, aggressive behaviour, and social difficulties. There was also growing interest in the lives of single-parent families living in poverty, in the context of both changing family demography in America and the political problematisation of these families.8 The Minnesota Longitudinal Study of Risk and Adaptation began in the 1970s as the world’s first attempt to conduct a prospective study of child abuse. Until that point, all research on child abuse had been pursued retrospectively. This raised questions of validity, especially given the significant potential stigma for adults in revealing experiences of childhood abuse. Egeland, Amos Deinard, and Ellen Elkin wrote a proposal to the Centre of Maternal and Child Health to initiate a longitudinal cohort study. Funding was provided in 1975 for three years in the first instance. The researchers’ belief in the value of prospective research exploring abuse and neglect proved well founded: decades later, only half of the adults in their sample with prospectively documented histories of childhood abuse self-reported that they had experienced maltreatment.9

Shortly before the study began, Arnold Sameroff visited the department at Minnesota and shared a draft of his classic paper proposing an ‘ecological’ model in which human development is shaped by the interaction between risk factors and environmental supports.10 This model had a profound influence on the methodology of the study, which sought to identify and capture various risk and protective factors in line with Sameroff’s idea of development (p. 339) as multiply determined.11 Particular attention was given to risk and protective factors in the life of the child’s caregiver or caregivers. Egeland and colleagues recruited 267 women in the third trimester of pregnancy and living in poverty.12 However, in line with Sameroff’s ideas, Egeland and colleagues were intent on assessing the multiple adversities faced by the families, documenting the widespread prevalence of domestic violence, mental health problems, and drug and alcohol use. Forty-one percent of the mothers had not finished high-school education;13 86% of the pregnancies were unplanned, with a third of families having obtained no equipment necessary for the baby or made arrangements for somewhere for the baby to sleep;14 the mothers were often socially isolated, with very low average levels of social support;15 59% were single at the time of their baby’s birth, compared to a 13% national average in 1975;16 and only 13% of the fathers were living in the same home as their children by the time they had reached 18 months of age. Soon after the study began, Egeland contacted Sroufe, a colleague at Minnesota, and together they conducted the Ainsworth Strange Situation with the sample, with input from two graduate students—Everett Waters and Brian Vaughn.17 The Strange Situation was attractive as a procedure offering a window into the experiences of care and emotional development of the children in the sample. And the classifications for the procedure had the initial basis for construct validity through Ainsworth’s Baltimore study (Chapter 2). For his dissertation Waters also conducted a parallel study of 50 mother–infant dyads from Minnesota who were not living in poverty. This allowed the Minnesota group to draw some comparisons to explore the role of socioeconomic adversity in predicting outcomes of interest, such as distribution of classifications in the Ainsworth Strange Situation.18

Egeland, Sroufe, and colleagues in the Minnesota group were sensitive to the cultural and moral context of developmental research with a sample facing such adversities.19 One such factor was acrimonious debates at the time over whether abuse was in fact harmful to children’s development, or whether associations between abuse and with negative outcomes could better be attributed to poverty.20 One measure the Minnesota group took to respond to these controversies was to expend every effort to carefully measure both abuse and adversity (p. 340) in a variety of ways, building convergent validity piece by piece out of multiple sources of information. These included researcher observation, parent report, and professional assessments.21 Such an approach allowed the Minnesota group to distinguish the shared and non-shared contribution of abuse and other adversities to later outcomes. Shaffer, Huston, and Egeland later reported that 40% of the non-maltreated child participants in their sample met diagnostic criteria for at least one mental disorder in late adolescence. Reports from abuse from one source made an additional contribution to later mental health, with an additional 20% of participants diagnosed with at least one mental disorder. There was, however, an additional 32% where abuse was validated by more than one source.22

A sense of the serious ethical commitment of Egeland, Sroufe, and colleagues in commencing a longitudinal study with a large sample recruited on the basis of low income and adversity, and the value of making sense of parents’ experiences, comes through with special clarity in their early publications. In examining the antecedents of abusive caregiving, this did not imply that they ascribed blame to the caregivers.23 Egeland and colleagues had specifically chosen a sample in poverty because deprivation of resources was so predictable as an antecedent of abuse. Responsibility for the care of children did not reside solely with their parent or parents: ‘As a member of society one shares a responsibility with respect to the quality of care available to all children’, wrote Sroufe in 1985, and ‘if responsibility for the child’s well-being does not reside in his or her inborn variation, then it is ours’.24

At the time the longitudinal study was beginning, there was a great deal of scepticism regarding claims, such as those of Bowlby, that early experience has a lasting role in human development. In the 1960s, researchers such as Jerome Kagan and Howard Moss had found little continuity in children’s behaviour or cognitive responses over time.25 However, these researchers assumed that continuity meant that the same behaviour or cognitive responses would be seen. The Minnesota group did not assume mimetic (homotypic) continuity in the forms of cognition, behaviour, or emotion shown by children in the context of maturation.26 In this, they differed from Bowlby who tended to expect early adversities to be mirrored in later behaviour (Chapter 1). All development builds on itself, they believed, but straightforward continuity should not be expected. Even if the present is heavy with the past, this legacy is always altered by the need to respond to present circumstances.27 In this perspective, the likelihood and extent of behaviours like abusive parenting are influenced by the interaction of risk and protective factors over time. Continuity would lie in the prediction of (p. 341) later maladaptation by earlier maladaptation, even if these forms of maladaptation were conditioned by quite different developmental challenges. Such difficulties would be held in place not only by individual-level factors, but also by the insistent compounding of environmental risks and the paucity of forms of support over time, which provides poor soil for anything but problems to grow. The Minnesota group therefore focused their attention on factors that could contribute to change and stability, and the role of family life as a contributory to these processes.

A multigeneration longitudinal study was the necessary testing-ground for many of the ideas that Bowlby and Ainsworth had sketched about attachment. For Ainsworth, ‘longitudinal research deals with the very stuff of life and human development, with all its consequent complexities and difficulties’.28 The participant is sought and attended to in the round. In her time it was a general consensus among developmental researchers that ‘long-term longitudinal studies have proved unmanageable’, which was a good part of why she had limited herself to a study of the first year of infant–caregiver interaction.29 Yet the importance assigned to early attachment relationship could only be accepted if it predicted later outcomes anticipated by the theory, and the most valid way to do this was through a prospective study rather than retrospective recall, with all its ensuing biases.30 Sroufe, Egeland, and colleagues regarded the construct validation of infant attachment as demanding two tasks. One was already sufficiently available from the work of Bowlby and Ainsworth: ‘a rich, interwoven network of interconnected theoretical proposition’. However, secondly, ‘this structure must be secured through observation’.31 A set of observations linking infant attachment not only to caregiving behaviour, but also to relevant developmental sequelae provided a protective belt for the emergent paradigm.32

Ultimately, ‘the validity of the attachment concept, as with any construct, hinges on the network of empirical relations built up around it and its power in yielding a coherent picture of individual adaptation’.33

In line with Ainsworth’s approach, the Minnesota group adopted two priorities in judging their methodology, oriented by a holism that aimed to train attention on the ‘whole person’, (p. 342) not just cut-up aspects of the person’s behaviour.34 One was to seek opportunities for naturalistic observation, where possible, especially in the early years of the study. Whereas video technology was just becoming available to researchers in the late 1960s, for use by Ainsworth’s doctoral students but not in her home observations, the Minnesota group made extensive use of video observations as well as observer report in order to capture the behaviour of their families in detail, and in a way that permitted repeated coding and inter-rater reliability.35 Critics alleged that Ainsworth’s observational data by no means amounted to science. This criticism has been often repeated over the decades.36 By contrast, use of video technology and multiple observers helped the Minnesota group combine the naturalistic observation valued by Ainsworth with a strengthened apparatus of scientific reliability and replication.

A second priority of the Minnesota group was to focus on broad-band competencies, rather than discrete measures.37 Egeland, Sroufe, and colleagues tended to create ‘composite’ variables out of their measures. For instance, attachment classifications and maternal sensitivity and intrusiveness were often, though not always, folded together into an ‘early caregiving composite’. The fact that sometimes they reported attachment classifications independently and sometimes as part of a composite, and sometimes the 12- and 18-month assessments independently and sometimes within the composite, led critics to accuse the Minnesota group of a lack of transparency. Lamb and colleagues, for instance, accused the Minnesota team of privately running the analysis for all the permutations and reporting those that had the best prediction.38 Certainly it is true that the choice of the use of the composite or attachment measures is generally left unjustified, which has at times proven an obstacle for attempts at replication or meta-analysis. However, on review of the publications of the Minnesota group, recourse to composite measures does not seem arbitrary. It occurs more often when the researchers were interested in broad-band or diffuse psychological processes. Additionally, for many of the investigations they wished to conduct, for instance in exploring the relative contribution of different adversities and of early care and attachment to later adaption, the Minnesota group had no precedents to follow. To an extent, they had to feel their way, trying out different approaches, hoping to hear the general trends in their data through composite measures, as well as to capture developmental processes anticipated (p. 343) to be diffuse and pervasive rather than localised. Many of the assessment measures originally developed in Minnesota were used in subsequent cohort studies.

The concern with observation and with assessing broad-band competencies led the Minnesota group to utilise the laboratory as a site for ‘critical situations’ (Chapter 2), where aspects of ordinary experience could be dramatised. They designed and set tasks that required the child or dyad to coordinate affect, cognition, and behaviour—the components of a behavioural system—to solve the problem. The laboratory-based challenge might differ in degree from those encountered within the rich, obscuring colour and camouflage of everyday life. Nonetheless it was anticipated to present a broad-band demand upon the resources, expectations, and capacities of the individual or dyad, and in this way offer a window into their history and present-day functioning. The paradigmatic example of such a challenge for the Minnesota group was the Ainsworth Strange Situation, which taxed children’s regulatory capacities and their trust in the availability of their caregiver, thereby offering some window into the dyad’s history of secure base and safe haven provision (Chapter 2).39 A later example was the ‘friendship interview’, a measure designed by the Minnesota group, which asked participants in sixth grade to describe their relationship with a close friend. Again, this was a broad-band demand relevant to the tasks characteristic of children’s developmental stage. Not all such measures were equally successful, however. The friendship interview had excellent associations with social competence in girls, and was predicted well by early infant–caregiver interactions. However, the assessment did not work as well with boys in the sample: ‘Minnesota males, and perhaps even more among our subpopulation of youth, are “men of few words”.’40

The Minnesota Longitudinal Study of Risk and Adaptation has now been running for over 40 years, and is rightly regarded as a masterpiece of developmental scientific research and the ‘gold standard in prospective studies of neglect and abuse’.41 Even decades later, the sample has retained 163 out of the original 267 participants, an astonishing achievement and a testament to the commitment of the participants and researchers. The Minnesota Longitudinal Study of Risk and Adaptation has provided a special opportunity to explore the lives of the grandchildren of the parents originally recruited to the sample, and to trace developmental processes across generations. This multigenerational research has revealed both substantial change and marked continuities in parenting practices between the generations, with stability calculated at r = .43, controlling for a variety of covariates. In a regression analysis with all the correlates of second-generation caregiving, the care that these adults had themselves received accounted for 12% of variance.42 This is far from the deterministic rhetoric sometimes deployed by Bowlby, especially in his populist writings (Chapter 1). However, a 12% difference in day-in-day-out parenting and care represents a very substantial difference in the family life of a young person. This finding exemplifies the contribution of the Minnesota study: considering a large range of risk and protective factors, following participants over (p. 344) long periods, pursuing multivariate analyses, and making powerful empirical contributions that in turn have implications for theory and child welfare practice.

In this, one area of particular contribution to the field of attachment research was in the conceptualisation and study of emotions. In order to situate the findings regarding attachment from the Minnesota study, this chapter will begin by outlining Sroufe’s general model of emotion. The selection of measures and the interpretation of findings by the Minnesota group were determined by a variety of factors which could vary from study to study. Nonetheless nearly all studies by the research group draw on Sroufe’s theory in the justification for their research design or in making sense of their results. It is therefore important to consider this contribution before describing the major empirical reports from the cohort study. And whilst Sroufe’s headline ideas are familiar and well cited, the development of his ideas, and their full depth, are less well known as they are often scattered across a variety of texts. As a result, many important aspects of Sroufe’s theory of human emotional life have been surprisingly untapped by later researchers.

Theory of affect

Emotional development

Bowlby identified three components to behavioural systems: behavioural components, cognitive components, and affective components (Chapter 1). In the 1960s and 1970s he gave special attention to behaviour. This left its mark in many ways on the theory, not least in the name ‘behavioural’ systems. Though already present from the 1960s in his concept of ‘internal working models’, from the 1980s Bowlby focused attention on the cognitive aspects of behavioural systems. By contrast, the affective component of behavioural systems was always acknowledged but never elaborated by Bowlby. In his private notes, he observed this tendency in his writings, and attributed it to a desire to backlight the importance of phenomena downplayed by psychoanalytic theory. However, the result was a hole in attachment theory where an account of affect, motivation, and development should have been. For studies on other topics, this hole might have been of less consequence. For conceptualising the results of a longitudinal study with a central focus on child maltreatment, it was unsustainable. Sroufe felt that ‘emotions and motivation are downplayed despite the fact that Bowlby’s observations led him to describe attachment as an “affective” bond’.43

The neglect of emotion in Bowlby’s work, in part an attempt to distance his approach from psychoanalysis, mirrored a trend in developmental psychology more generally. Though attention to emotion was on the rise, the main protagonist in the story of psychology in the late 1970s was cognition. Psychoanalytic theory retained a strong interest in emotion, but came into disrepute in American academic psychology departments, and there were few effective bridges between psychoanalytic theory and the empirical psychology of the period.44 Yet (p. 345) through the 1970s, across the social sciences, attention to emotional life, emotion regulation, and the social situatedness of emotions was growing, reflecting rising attention to these topics in wider society and politics.45 In a landmark book chapter from 1979, Sroufe argued that ‘the child grows not as a perceptive being, not as a cognitive being, but as a human being who experiences anxiety, joy, and anger, and who is connected to its world in an emotional way’.46 Together with Minnesota colleagues, Sroufe highlighted the flexibility of human behavioural systems, in contrast to reflexes, and the role of psychosocial challenges such as trust, autonomy, and control as providing the context of their progressive elaboration. Many behavioural systems include reflex or relatively inbuilt components, but become elaborated as they are fed by social and cultural experiences. For instance, the laugh response is predisposed for human infants by particular kinds of situation, such as incongruity in the context of mid–high arousal, but also by the availability of a caregiver who gives a positive appraisal to the incongruity.47 Sroufe narrated infant development in terms of the gradual differentiation of arousal into various, initially quite general affects, and from there into more refined emotions that are differentiated by their specificity and directedness.48

These refinements are based partly on the social support the child receives in identifying and modulating affects when they arise in different contexts. Sroufe ascribed an important role to the provision of a safe haven when a child is distressed for the modulation of affect. However, he also elaborated on an expanded role for exploration. For Bowlby, the exploratory system has the function of learning about new and/or complex phenomena. Sroufe added to this that play, especially in the context of a secure base and stable routines, helps the infant learn, over time, to remain regulated in the context of higher levels of arousal: ‘As caregiver and infant play, tension is escalated and de-escalated, to the edge of overstimulation and back again, commonly ending in the bursts of positive affect so rewarding for caregivers. Episode by episode, day by day, the infant’s own capacity to modulate (and tolerate) tension is developed, and a reservoir of shared positive affect is created.’49

This reservoir, which Sroufe largely identified with the attachment system, is nestled initially within the relationship. However, over time it becomes in part the property of the child, as a sense of confidence and fluency in accessing, modulating, and expressing feelings, as well as recouping or finding relief afterwards. This becomes a skill, one that can help make for successful family and peer interactions. Frustration and conflicts can be experienced (p. 346) without feeling unacceptable, dangerous, or spiralling into persistent activation or diffusion into other nearby emotions such as distress, fear, or despair:

Resolving conflicts is an important building block of the child’s emerging sense of competence at problem solving, and as Erikson argues, it also deepens the child’s trust in the caregiving relationship. Such prototypical conflict experiences within the security of the caregiving relationship can also represent a model for later close relationships, providing an abiding confidence that relationships may be sustained despite strife, which allows a person to risk conflict in relationships and, ultimately, to even see its value.50

However, as well as social scaffolding by caregivers, the refinements of emotional development of young childhood depend on cognitive maturation. Unlike Bowlby for whom emphasis on emotional processes was felt to risk underplaying the cognitive components of behavioural systems, Sroufe emphasised that the two are intertwined. Cognitive achievements are spurred and textured by affects. In turn, the registering and differentiation of emotions is scaffolded by cognitive development:

Cognitive factors underlie the unfolding of emotions: only with recognition is there pleasure and disappointment; only with some development of causality, object permanence, intentionality and meaning are there joy, anger, and fear; only with self-awareness is there shame. Also, distinctions among affects and their precursors call upon cognitive achievements—for example, fear, as reflected in more immediate distress upon a second exposure to a stranger, has been referred to as a categorical reaction, dependent upon assimilation to a negative scheme. Finally, the effects of sequence, setting, and other aspects of context on emotion are obviously mediated by cognition.51

Certain emotions, such as shame, pride, and guilt, are only possible once sufficient cognitive infrastructure is available, since, together with high arousal, they require identification of discrepancies between behaviour and the representation of our value from the perspective of others. Think, for instance, of the development over childhood of shame in what we have drawn with pencil and paper, or in our bodies, to the point that it is to an extent typical in adults. Sroufe observed that since toddlers are only just piecing together the representation of their value in the eyes of others, this impression may still be especially fragile. A consequence is that ‘the toddler is vulnerable to a global feeling of dissolution when being punished for a specific behaviour (especially if done harshly or in a degrading way)’.52 Whereas in the first year of life, anxiety in relationships is associated particularly with separation, Sroufe argued that over the second year, cognitive advances allow psychological separations, such as scolding, to be pierced by as much anxiety as physical separations.53

In his account of emotional development, Sroufe took particular interest in the basic smile reflex, highlighting the critical role of social interaction. He wrote beautifully of the social smile as the ‘crowning achievement’ of the first three months of life, initiating the use (p. 347) of positive affects within communication: ‘the infant smiles and coos at its own feet, at its toy giraffe, and especially at the caregiver. For the first time it laughs in response to vigorous stimulation. “Pleasure has become an excitatory phenomenon”, associated with high states of excitation (Escalona, 1968, p.159). With eyes sparking, caregiver and infant set out upon the task of establishing reciprocal exchanges.’54 The infant does not yet have a ‘representation’ of the relationship—that will only come later, along with language—but they do have ‘action schemes’ which allow for procedure-level memories and expectation of how the caregiver will respond, coordinated recognition, and sharing of differentiated emotions, and some basic forms of independent emotion regulation.55 Action schemes are encoded in a preverbal way that is qualitatively different to semantic memory, and that makes them more difficult to verbally analyse and re-evaluate.

With cognitive development, more complex and differentiated emotions are possible. Yet ‘The more mature emotions do not displace precursor emotions. The smooth execution of behaviour can still produce pleasure, unfathomability and vague threat can produce wariness, and diffuse frustration can lead to intense distress at any age.’56 Within the mature anger behavioural system of a one- or two-year-old remains the earlier more global, intense, and diffuse form, which becomes vocally evident in children’s tantrums. The same, Sroufe argued, was true of other behavioural systems like fear: abject fear loses specificity and directedness, as well as its differentiation from nearby emotions such as distress.57 The process of emotional maturation exemplified the broader commitment of Sroufe and the Minnesota group to the principle that all development is built on itself, with earlier forms remaining operative even if reforged by new opportunities, capabilities, and demands: ‘Development is a dynamic process wherein what evolves derives from what was there before in a logical way, while, at the same time, prior experience or prior adaptation has a fundamentally new meaning in the now more complex system.’58 Even if early experiences of support and care are supplanted by later adversity, they remain a resource that may be reactivated by a facilitative later context. Likewise, early vulnerability in the context of adversity can be reactivated, even many years later, when someone is faced with depriving or traumatic conditions.59

Within emotional maturation, Sroufe gave an important place to attachment processes. His stance was that ‘attachment is critical and has a central place in the hierarchy (p. 348) of development because of its primacy. The infant–caregiver attachment relationship is the core, around which all other experience is structured, whatever impact it may have.’60 However, in occupying this central place, Sroufe held that attachment processes are nonetheless dependent on an infrastructure of various factors that stretch beyond them. One is the dependence of the attachment system on cognitive maturation. He assigned a particular role to cognitive anticipation and expectation, which is reflected in the partly innate capacity for joyous greeting of attachment figures. He proposed that ‘bouncing, smiling, arms-raised gestures are differential to attachment figures, reflecting the positive value of the special scheme to which the attachment figures are immediately assimilated’.61 Cognitive anticipation is also reflected in the capacity to hold an angry mood, even with a caregiver, or to inhibit the activation of a behavioural system that will result in disapproval or punishment. The infant begins to include his or her own potential feelings within the field of awareness, influencing responses. For instance, the knowledge that punishment and sadness will ensue may help hold the inhibition of a behavioural system in place.

Sroufe proposed that the infant’s emerging sense of self is organised especially around ‘patterns of behavioural and affective regulation, which grant continuity to experience despite development and changes in context’.62 This is a good part of why attachment relationships are central to socioemotional development, and why the recognition, communication and modulation of emotion is central to the quality of attachment relationships. On the basis of this account, Sroufe offered rebuttal to Kagan’s strongly worded accusation that temperament was a sufficient explanation for apparent differences in the quality of attachment relationship.63 Sroufe argued that findings, such as those of Main and Weston (Chapter 3), showing barely any overlap between attachment classifications with different caregivers are important evidence here. They indicate that differences in attachment lie in the history of infant–caregiver interaction, and have no basis in differences in temperament.64 Over time, Behrens observed that ‘the attachment–temperament debate has subsided because most attachment researchers today incorporate some temperamental or biological assessment of children, recognizing the potential biological contribution in forming relationships’.65 Van IJzendoorn and Bakermans-Kranenburg later criticised Sroufe and Kagan as partially responsible for the ‘deadly war over dominance and territory’ between attachment and temperament researchers, sustaining a polemic that obscured the way that parenting, (p. 349) attachment, and temperament interrelate (Chapter 7).66 If, as Karen has suggested, Kagan intentionally helped to block access to key potential funders for attachment research through the 1980s, the war becomes quite intelligible.67 Yet with Kagan’s relevance for attachment research receding into the past, the opposition between attachment and temperament has lost some of its necessity. In an article from 2000, Sroufe and colleagues accepted that the majority of variance in internalising and anxiety symptoms in middle childhood is likely accounted for by temperamental and genetic factors.68 And in a 2005 article, Sroufe highlighted the interaction between infant temperament and caregiving sensitivity in predicting infant ambivalent/resistant attachment in the Strange Situation in the Minnesota Longitudinal Study of Risk and Adaptation.69 Nonetheless, with the exception of the ambivalent/resistant classification, a meta-analysis by Groh found little evidence for main effects of temperamental variation in the development of differences in infant attachment.70 Exciting examinations of potential interactions between temperament and caregiving experiences are ongoing, but on the basis of a scientific consensus that ascribes an irreducibly important place to caregiving.71

Felt security

Attention to the behavioural, cognitive, and emotional components of behavioural systems led Sroufe to a different position to Bowlby regarding the set-goal of the attachment system. In his writings for a psychoanalytic audience, Bowlby identified important emotional components related to the set-goal of the attachment system. He specified that the desire to achieve the set-goal would be felt as ‘anxiety’; when the caregiver is accessible if the child feels frightened, he or she will feel ‘secure’; when the caregiver is accessible after the child has felt frightened, he or she will feel ‘comforted’.72 Yet, in Attachment, Volume 1, Bowlby gave (p. 350) these subjective states primarily the role of concomitants to the set-goal of the attachment system, which he positioned as ‘proximity’ (Chapter 1). In the 1970s, however, there were growing concerns that the attachment paradigm was being hindered by a lack of precision in situating the subjective, felt aspects of the phenomena under discussion. The German ethologist and systems theorist Norbert Bischof, for example, observed that ‘authors dealing with attachment quite commonly talk about the mother as a source of infantile “security”. It should be noted, however, that this term does not clearly distinguish between an emotional state (feeling more or less secure) and an environmental fact.’73 Sroufe and Waters likewise argued that Bowlby’s neglect of emotion obscured, and even distorted, his account of the attachment behavioural system.

In ‘Attachment as an organizational construct’, published in 1977, Sroufe and Waters raised several objections to the sidekick role allocated to emotion.74 They noted that some infants, such as those in dyads classified B1 by Ainsworth, appeared to have had their attachment system terminated by merely the return of the caregiver and distance interaction, without physical proximity-seeking. The ambivalent/resistant infant also achieved proximity in the Strange Situation, and yet the attachment system remains activated. Sroufe and Waters argued that emotions not only are important parts of the motivational system of attachment, but also condition its terminating conditions. In contrast to Bowlby, who offered a behavioural set-goal for the attachment system, Sroufe and Waters broadened this out. They proposed that the attachment system would be terminated when infants ‘feel secure’. Proximity is one facilitating and sometimes a necessary contributor to this affective state. But it is the affective state that is the set-goal.75

Elsewhere, Sroufe proposed that ‘the infant secure in his or her attachment has experienced the caregiver as a reliable source of comforting, as responsive to his or her signals, and as available and sensitive. The infant has learned that stimulation in the context of the caregiver will generally not be overwhelming and that when arousal threatens to exceed the infant’s organizational capacity, the caregiver will intervene.’76 Indeed, Sroufe was one of the very few researchers who, following Ainsworth, highlighted that the capacity to feel secure in attachment relationships would be influenced by feelings of security or insecurity in other domains (Chapter 2). Furthermore, ‘the availability of cues associated with security (familiar procedures or objects, predictability, control) and a generalised expectancy concerning likely outcomes when tension is high’ were assumed to feed into felt security within attachment relationships,77 perhaps contributing to the small overlap in variance in attachment relationships with different caregivers.

Ainsworth did not appear to regard the broader emphasis on felt security as discrepant with the orthodox position represented by Bowlby. She initially reported reading the Sroufe and Waters paper with approval. However, over the years, she began to feel uncomfortable. Many second-generation attachment researchers began to rally to ‘felt security’, against what was increasingly regarded in the 1980s as the somewhat wooden quality to Bowlby’s account (p. 351) of attachment relationships in terms of activating and terminating conditions.78 Ainsworth’s former student Mark Cummings was even developing a new coding system for the Strange Situation focused on felt security, as an alternative to coding proximity-seeking and contact maintaining.79 Ainsworth brought the matter to Bowlby’s attention in 1987, and the collaborators reflected together.80 They were worried, justifiably as it turned out (Chapter 5), that the term ‘felt security’ left ambiguity regarding whether an individual would be consciously aware of whether or not they had a secure attachment. They were sure that individuals would not. However, they also registered that the set-goal of the attachment system, even in infancy, was not merely proximity. That early proposal of Bowlby’s became rather battered as time passed since the 1960s.

Ultimately, Bowlby and Ainsworth came to the conclusion that the set-goal of the caregiver is the feeling of the caregiver’s availability when called upon, which includes the possibility of proximity as needed:

I am in agreement with you … I have for my part referred to ‘availability’ as the set-goal in the case of children and adults. It requires belief in the effectiveness of communication & belief that the attach figure will respond if called upon & that physical proximity is readily attainable when desired.81

The set-goal of the attachment system in infancy was revised by Ainsworth to be caregiver ‘availability’, assumed to incorporate behavioural, cognitive, and emotional components. In this, access to proximity was maintained as one of the proximal goals of the attachment system at high levels of activation, but not the ultimate set-goal regardless of level of activation. Unfortunately there was only one book chapter in which Ainsworth explicitly flagged the shift in position, and it was framed as a rejection rather than partial acceptance of Sroufe and Waters’ position.82 Furthermore, most readers assumed that when Bowlby and Ainsworth were referring to availability, they meant exclusively older children.83 Attachment researchers still today have generally not integrated the different positions, and the set-goal of the attachment system remains variously defined as ‘proximity’, ‘availability’, ‘accessibility’, and ‘felt security’.84 Ainsworth’s student Roger Kobak, recognising this problem, roundly (p. 352) criticised everyone. Bowlby and Ainsworth were criticised for recognising so late in the day that caregiver availability, even in infancy, was broader than proximity. And Sroufe and Waters were criticised for underspecifying the role of physical proximity, and for defining felt security too loosely. Not least, Kobak was concerned that a whole host of factors could contribute to feeling secure, not all of which would be based in attachment processes.85

Kobak reserved particular criticism, however, for the way that Sroufe and Waters’ proposal contributed to the split in attachment theory between the developmental tradition and the social psychology tradition (Chapter 5). Kobak regarded the split as based in part on a lack of clarity about how the attachment system actually works in adulthood, caused in part by Sroufe and Waters’s overencompassing and subjective notion of felt security.86 More recently, Kobak has moderated his criticisms, likely in the context of increased rapprochement between the developmental and social psychological traditions of attachment research. He has allowed that the shift from a focus on proximity to availability helped support a shift in ‘the focus of attachment research from the study of young children’s separations from caregivers to investigations of the quality of emotional communication in maintaining attachment bonds’, a shift that he regards as overall a positive development for the field.87

Social affects

In the 1980s, Sroufe’s theoretical reflections on attachment and emotional development were elaborated by the Minnesota group’s empirical research. An important early study was ‘The role of affect in social competence’, published in 1984 by Sroufe, Fox, and Pancake, reporting a mixed methods study of the experiences of children in the sample in preschool.88 Two preschool classes (15 children and 25 children) were constituted out of children in the sample who had received the same attachment classification at 12 and 18 months, stratified to include children from each of the Ainsworth classifications. The preschool lasted for 20 weeks. (p. 353) Based on detailed ethnographic observations, Sroufe and colleagues were struck by the role of positive emotion, which seemed to serve as the major currency of social interaction.

The children in the sample who found it difficult to socialise seemed often those who were unable to resource positive emotions to serve this role of currency.89 From childhood onwards, Sroufe and colleagues theorised, something has to be shared in order to make or sustain a relationship. Emotion offers potent forms of sharing, and indeed of not sharing.90 Relative access to shared positive affect therefore contributes to the creation and stabilisation of social hierarchies, which in turn channel the expression and acceptance of emotion between the preschool students. In the two different classes, Sroufe and colleagues asked the teachers to rank the students in order of social competence. They found a strong association between competence ranking and the frequency of affectively positive social interaction (ρ‎ = .70, p < .002). Notably, however, the frequency of social interaction with a negative affective tone made no additional contribution, once shared positive emotions were taken into account. This suggests that the emotional tone of peer interactions was not simply a product of social competence, but that positive social interaction was linked, likely bidirectionally, to social success.91 The researchers also documented that positive affect in itself was not critical for success with peers at the school, but shared positive social interaction specifically: ‘the sharing of good feelings and the rewarding of other’s overtures with positive affect’.92

Sroufe worked with his partner June Fleeson (later June Sroufe) in reflecting further on the social quality of affects. They made two key claims. A first was that if the relationship depends on sharing, then all partners in a relationship have to learn one another’s roles in order to understand their own. As such, children learn all the ‘parts’ of a relationship with an attachment figure, not just their own role. There is no sense of determinism in Sroufe and Fleeson’s reflection. Individuals may develop thoughts, feelings, or behaviours in relationships that develop out of earlier roles only in the most non-linear way.93 However, early roles may press themselves forward at times, particularly when an adult is exhausted, bored, depleted, and/or poorly resourced. Across our lives, Sroufe and Fleeson argued, when we respond in ways that are intuitive and unreflective, this may be especially coloured by our history, and especially our experiences of attachment and of trauma. Bowlby had argued that children who have experienced abuse may find it hard to find and sustain the role of adequate caregiver because their representations in adulthood mirror those of their childhood. By contrast, Sroufe and Fleeson proposed that children who have experienced abuse not only have the abusive role available, but also may have relative difficulties accessing procedural components of sensitive caregiving—for instance, gentle, non-intrusive touch and the accurate recognition and modulation of affects. Children who have experienced abuse may also have comparatively less access, even into adulthood, to social relationships that help them sustain the caregiving role. In fact, in some cases, they may only have access to (p. 354) relationships with individuals who have some investment in maintaining earlier patterns of power and identity. So efforts these children undertake as adults to avoid the abusive role may still leave them disadvantaged in attempting to parent well, especially when contextual demands elicit intuitive and unreflective responses.

A second key claim made by Sroufe and Fleeson was that the organisation of life within the family and the psychological coherence and emotional life of its members influence one another. If the integration of behaviour, thought, and emotion within the family over time is fragile or punctuated by crises, the members of the family will find emotion regulation difficult. Likewise, if a member of the family has difficulty organising behaviour, thinking, and emotions in coherent ways, this makes integration at the level of the family a greater challenge.94 At both the individual and the family level, the integration of emotion with thoughts and feelings permits the modulation and direction of emotional expression. In turn this allows emotional experiences to inform and guide action but without threatening to overwhelm or disrupt its effectiveness.95 Families regularly face external stresses and developmental change—some especially. This means that the organisation of family life is never a fully stable achievement, and for some families is under constant bombardment and active attempts at repair. Nonetheless, when integrative processes are linking behaviour, thinking, and emotions well, experiences shared by families, even difficult ones, can help inform the actions of its members in ways that help them respond to the present; likewise, experiences by individual members contribute to action at a family level.

By contrast, when family members find integration behaviour, thinking, and affect confusing or difficult, there will often be patterns to the fallout and response. Two broad possible ‘strategies for adaptation’ were highlighted by Sroufe and Fleeson, corresponding to the ‘two primary patterns of anxious attachment’. A first is for the family to treat a wide array of situations as potentially combustable or distressing, and to seek continual emotional contact with one another around these troubles. A second is for the family to cut themselves off from interaction or communication when this might raise potentially troubling or distressing experiences for individuals.96 Families adopting the latter strategy, who experience one another as unreachable or threatening under certain conditions, may outlaw those conditions rather than feel confident in their capacity to modulate and be informed by them. For instance, any expression of anger may be completely forbidden, with tense situations evoking ‘ritualist (repetitive) patterns of avoidance’.97 Sroufe and Fleeson observed that it is quite usual for one or even both of the strategies to be deployed by families handling the fallout from disruption or challenges, before a new organisation has been re-established. What signals a pathway towards problems for the family and its members, and may predispose or contribute to mental health problems, is if such strategies become rigid to the point that they block opportunities for repair, learning, or development.

From her doctoral research, a study of family relationships, Fleeson reported supportive findings.98 Children were likely to display a tight and rigid control over their emotions when (p. 355) raised in families where anger was disguised or distorted and in which members of the family did not feel able to communicate safely. Conversely, children were likely to display dysregulation of emotion under two conditions, which could occur separately or together. One was when integration was hindered by intrusive sharing of emotions by the parents. The other was when the roles in the family were distorted, with the opposite-sex parent seeking a closer relationship with the child than with his or her partner. In either case, intimacy occurs primarily to the parent’s pacing, without modulation for the child’s needs. Both situations were theorised by Fleeson to contribute to a feeling of lack of control for the child. The child’s attachment relationship with the parent means that insulation against intrusive parental intimacy is a complex task at best. It may, in fact, be ‘essentially impossible’ for many younger children and early adolescents.99 At that age, especially, our parents’ concerns can be all but inescapable.

Intrusive intimacy

Pursuing the implications of intrusive intimacy with the parent as a distortion of the attachment relationship, Sroufe and Ward filmed mothers’ behaviour during a toy cleanup task in the laboratory with the Minnesota sample when the children were toddlers. The researchers measured the extent to which the mothers showed sensual physical contact, sexual teasing, or requests for affection in the course of trying to persuade their toddler to put away the toys. Of the 173 participants who attended the toddler call-back, 16 showed one or more of these behaviours.

Sroufe and Ward distinguished forms of boundary dissolution from warmth or affection on a number of grounds: the behaviours were directed almost exclusively by the mothers towards their male toddlers; they were associated with more, not less, physical punishment and threats of punishment; they were not associated with measures of cooperative, encouraging, supportive, or affectionate behaviour; and they tended to interrupt the child’s behaviour putting away the toys rather than facilitate this goal.100 In fact, half of the punishing behaviours shown in the entire sample were displayed in these 16 dyads; this suggests a lack of monitoring of the environment by the mothers, as they knew they were being filmed.101 Mothers who showed seductive behaviour towards their son did not display the same behaviour towards their daughter; instead they showed derision.102 Eight of the 16 mothers were included in a subsample of 36 participants given a full family history interview, and assessed independently for indicators of abuse, including incest. Seven of the eight had histories suggesting that they had been required to supply emotional or sexual intimacy to their father or stepfather during childhood in ways that were ‘suggestive of an incestuous-type (p. 356) relationship’.103 In a follow-up at 42 months, nearly half of the women reported experiences of intrafamilial sexual abuse (compared to 8% of the total sample).

The behaviours shown by adults displaying intimately intrusive behaviours should not be regarded as imitations of their own caregivers, as anticipated by a simple learning approach to intergenerational transmission of abuse—as for instance held by Bowlby (Chapter 1). Instead, Sroufe and colleagues argued that children learn both roles within attachment relationships, and that aspects of both can appear within adult behaviour:

These mothers do not simply show deferred imitation of their parents’ previous behaviour. For example, those who were exploited by their fathers do not literally do what their fathers did to them. Rather, they engage in the culturally specified adult female form of cross-gender child exploitation. They have internalised a relationship and not simply a set of behaviours.104

For this reason, the fact that some of the mothers were doing well with their children in other regards was not a surprise to the Minnesota group. The problem in some instances was quite localised: the attachment system may inform the caregiving system, and they will share some components, but they are distinct. The adult caregiving system is influenced in powerful ways, for instance, by contemporaneous social support. Therefore a parent’s behaviour towards his or her child would not necessarily be expected to mirror the parent’s past. For instance, if other relationships are available to provide a secure base and safe haven, children would be less likely to be treated as objects of the attachment system and elicit the sexuality and derision that had been woven into it by these mothers’ childhood experiences. In support of the protective role of contemporaneous support, the Minnesota group found that only 9% of the mothers who engaged in boundary dissolution behaviours had a stable relationship with a partner, compared to 25% of the rest of the sample.105

The construct of boundary-dissolving behaviours was later found to have substantial predictive value. There was also substantial stability over time. In observations made at age 13 of dyads who had earlier shown seductive behaviours, boundary-dissolving behaviour was displayed by both adolescents and their parent.106 It had become an organisational quality of the relationship involving both members of the dyad. A decade later, the Minnesota group reported a strong association between early boundary-dissolving behaviours by mothers and their child’s number of sexual partners in adolescence.107 Boundary dissolution in the age 13 assessment was also found to predict conduct and attention problems at age 16, (p. 357) over and above behavioural problems shown at age 13 and other measures of family functioning.108 Such predictive validity offered support for the construct of boundary dissolution, though the measure has not subsequently seen use in later research by developmental scientists. One reason may have been that the idea of mothers behaving seductively towards their toddlers was simply too controversial and unthinkable an idea to be taken up by researchers. A broader trend may have been the decline of reference to family systems theory within developmental science over the subsequent decades. Furthermore, ideas of ‘child abuse and exploitation’ and the ‘parentification of children’ were increasingly salient in clinical discourses during the 1980s and 1990s: these covered some of the same ground as Sroufe and colleagues’ work on boundary dissolution, but in both cases trained attention more on older children and on dyadic interaction.109 Nonetheless, the fact that Sroufe’s approach to thinking about affects and relationships contributed to the novel finding of the importance of boundary dissolution offered demonstration of its value. And the finding also highlighted the importance of caregiving experienced within attachment relationships for children’s development.110 Exploration of the antecedents and sequelae of early care was central to the concerns of the Minnesota group, with particular interest in measures of individual differences in infant attachment.

Correlates of attachment

Antecedents of attachment

Sroufe, Waters, and Vaughn conducted the Strange Situation with the sample of 267 mothers when their infants were 12 and 18 months. Sixty percent of the sample had stable attachment classifications over the six-month period. Of the 267 mothers, there were 31 cases of infant maltreatment, confirmed scrupulously through multiple methods.111 Of these 31 mother–infant dyads, at 12 months, 62% were classified as insecurely attached, compared to 45% in the total Minnesota sample and 34% of the combined middle-class samples reported in Ainsworth and colleagues’ Patterns of Attachment.112

(p. 358) In the total Minnesota sample, 21% of the dyads were classified as ambivalent/resistant. This was substantially higher than the 13% seen by Ainsworth.113 The Minnesota group found that there was a powerful overlap between ambivalent/resistant attachment in the Strange Situation and the nurse’s report on the regulatory capacities of the child as a newborn (r = .46), including their startle response (r = .20).114 However, these findings, suggestive of a role for temperament, were reported in an undertone by the Minnesota group in their publications of the 1980s and 1990s, and were rarely mentioned in their review or theoretical works. (The field would have to wait for a recent meta-analysis by Groh and colleagues for the resistance–temperament link to resurface for discussion.)115 Instead, the Minnesota group focused on the portion of the variance in ambivalent/resistant attachment that appeared to be environmental. Ainsworth had situated the broad notion of ‘inconsistent care’ as the origin of ambivalent/resistant attachment, though her more specific theory seemed to be that these children had caregivers who struggled to hold them in mind (Chapter 2). Egeland and Sroufe argued that inconsistent care may have been made more likely by the poverty and other adversities faced by their sample.116 This interpretation was buttressed by their finding that half the cases of neglect were in dyads receiving a classification of ambivalent/resistant attachment. And a qualitative examination of the cases indicated a high level of drug use, which contributed to leaving the infant unattended for long stretches.117

A puzzle in the findings from this study was that 12 of the 31 abused infants were classified as securely attached, using Ainsworth’s coding protocols. They displayed protest on separation, proximity-seeking, and some contact-maintenance on reunion, but could be comforted and were able to return to play. Yet they also displayed other behaviours that seemed to interrupt or contradict the secure pattern described by Ainsworth. Informally, the Minnesota group called these 12 cases ‘unhealthy Bs’: though they met the Ainsworth criteria for Group B, their behaviour in the Strange Situation did not seem secure.118 In watching the tapes closely, Egeland and Sroufe identified behaviour that was ‘neither avoidant nor resistant’, but which nonetheless suggested insecurity, for instance responses to the caregiver that were ‘apathetic or disorganized’.119 Rather than disregarding these cases as (p. 359) ‘noise’ produced by imprecision in the Strange Situation procedure, they suspected that the cases represented an attachment-relevant process of some significance. However, because Sroufe was coding the 18-month Strange Situations, he needed to be ‘blind’ as to which recordings were those of maltreated children. He therefore went rigorously out of his way to avoid speculating about the aetiology of this behaviour until data analysis was complete. Instead, Sroufe gave copies of these recordings to Main to examine (Chapter 3).

The Minnesota group were interested in the antecedents of the Ainsworth patterns of attachment, to see whether their findings aligned with those of the Baltimore study. Confirming Ainsworth’s conclusions, Egeland and Farber reported that caregivers in secure dyads were more likely to display sensitivity during feeding episodes with their infants at three and six months and free play at six months. Though the association between sensitivity and security in the Minnesota study was much less strong than in Ainsworth’s data, it was in the expected direction. The Minnesota group identified other variables that might suggest a more favourable and settled caregiving environment for infants in securely attached dyads. Mothers who lived with their partner for the first 12 months of the child’s life were more likely to be part of a secure dyad.120 It was also noted that mothers who reported fewer current adversities or stressors were more likely to have a child with a secure attachment relationship, a finding that ‘shows the foolishness of blaming parents for developmental problems of the child. As life stress is reduced, quality of care improves, as does the quality of the child’s attachment relationship.’121

Egeland and Farber examined nurses’ ratings of the mothers’ interest in their newborn baby. They found that infants in dyads later classified as avoidant had mothers who were reported to show less interest. Three months after the birth, caregivers in ambivalent/resistant dyads also scored lower on the Maternal Attitude Scale. This was a self-report measure with three scales: appropriate vs inappropriate control of child’s aggression, encouragement vs discouragement of reciprocity, and acceptance vs denial of emotional complexity in child care. Furthermore, they found that infants in ambivalent/resistant dyads scored lower on scales of motor and mental development at nine months. The researchers suspected that if the infants had been easier to care for, or had caregivers who accepted the complexity of child care and could adapt to their child’s needs, many of these dyads would have developed secure attachment relationships.122 They found no difference between groups in the amount of warmth demonstrated to the infants by their caregivers. However, the mothers of infants who changed from secure to anxious over the six months were rated by observers as showing less delight in their baby than other caregivers at six months. Self-report measures of maternal personality, such as aggression and impulsivity, had little relationship with caregiving behaviours or Strange Situation classifications.123

One potential form of adverse caregiving that especially interested the Minnesota group was emotional neglect. This was a relatively new concept. An early advocate of the category of emotional neglect had been John Bowlby. Though he tended to ignore the role of emotion (p. 360) as a cause or process in psychological life in his theorising, he was highly attentive to child emotion as an index of problems with the availability of adequate care. Already in 1971, Bowlby was arguing to the Home Office select committee on the adoption of children that there is an ‘urgent need’ for ‘a really adequate research study of emotional neglect’.124 The use of emotional life in the identification of abuse and neglect was a matter of contention in the early 1980s, as it seemed to involve an even stronger value judgement by social workers than physical or sexual abuse or physical neglect.125 Yet emotional abuse and neglect swiftly grew into the primary basis for referrals to child protection services.126 In 1985, Sroufe reported on the antecedents of emotional neglect/psychological unavailability in the Minnesota sample.127 Though the infants who received this form of care received typical scores for mental and motor development at ten days and three months, by six months their development was impeded, and they departed more and more from the rest of the cohort with each further assessment of their development. At 12 months, 42% of the infant–caregiver dyads were classified as avoidant, and at 18 months 86% received this classification.128 Much later, at age 17, Egeland reported that the large majority met diagnostic criteria for a psychiatric disorder: 53% met criteria for an anxiety disorder compared to 34% of the rest of the sample; 53% met criteria for a conduct disorder compared to 19% in the wider sample; and 25% met criteria for PTSD compared to 12% in the cohort as a whole.129

In the 1990s, after the introduction of the disorganised/disoriented attachment classification by Main and colleagues (Chapter 3), Elizabeth Carlson recoded the Minnesota Strange Situation recordings for disorganised attachment and examined caregiving antecedents. In her analyses she used the highest rating at either 12 or 18 months to represent a score for disorganisation. Examining antecedents of disorganised attachment, she found that the best predictor was a composite variable comprised of intrusive and insensitive caregiving (r = .38), and that various forms of neglect and maltreatment made a modest additional contribution (overall abuse/neglect r = .29; physical abuse r = .20; emotional neglect/psychological unavailability r = .23). She found no association between disorganised attachment and maternal mental health or drug and/or alcohol abuse.130 This last finding, however, differs from later studies. A meta-analysis by Cyr and colleagues found that parental drug and/ (p. 361) or alcohol abuse did predict disorganised infant attachment, especially when compounded by other risk factors.131 Likewise, some aspects of maternal mental health has been found to be relevant to the prediction of infant insecure and insecure-disorganised attachment, such as PTSD.132 (On the relationship between caregiver depression and infant attachment see Chapter 3.)

Early social competence

Sroufe, Egeland, and colleagues did not regard secure attachment as any guarantee of positive later outcomes, but as a promotive or protective factor. Nonetheless, the Minnesota Longitudinal Study of Risk and Adaptation demonstrated a number of important and valuable findings regarding the later positive correlates of infant secure attachment. In a follow-up when the children were two years old, the dyads were invited to the laboratory; 190 families attended, of the 267 who started the study. Mother–child dyads were observed as they completed a series of tool-based problem-solving tasks together, several of which were more demanding than most two year olds would find comfortable. The reason for this was that assessments of behaviour or cognition alone would risk being confounded with other variables like IQ. Completion of a problem-solving task with a caregiver was assumed to represent a developmentally appropriate challenge to a toddler, requiring the integration of affect, cognition, and behaviour in order to stay on task and coordinate together effectively. This meant that the task could serve as an assessment of the ability of the children to mobilise and coordinate personal and social resources to realise the opportunities and potentials of the environment.133

This capacity was termed by Waters and Sroufe ‘social competence’. The term was drawn from scholars such as Diana Baumrind, for whom social competence signified the ‘social responsibility, independence, achievement orientation, and vigor’ required of citizens of American capitalism.134 Waters and Sroufe appreciated the metaphor of individual possibilities as shaped by the resources available to them. However, they departed from Baumrind and other individualising competency discourses in emphasising that qualities such as social responsibility and independence are not best thought of as individual properties, and certainly not in children. Rather, they should be regarded as transactional effects of individual–environment interactions, shaped by experiences of early care. The importance of early care for Waters and Sroufe lay in the fact that it shapes both the inner and external resources available to the child, and also influences how these resources are interpreted and used. In line with this emphasis on early experiences of care, the Minnesota group found that, according to the parent-report, children from dyads who had previously been classified as insecure attached showed more behaviour problems at home. Those caregivers from previously (p. 362) insecure dyads who reported that their children showed fewer behavioural problems were observed to be respectful of their autonomy, offered more support, provided clearer structure, and were less hostile during the problem-solving tasks. These were also mothers who reported greater access to family support and friendship networks.135 This suggested that the capacities of the caregiver were likewise shaped integrally by the availability of supportive provisions, and should not be regarded merely as an individual quality.

In later assessments at preschool age, the Minnesota group found that an interaction between supportive maternal care and secure attachment predicted stronger capacity for managing and delaying behaviour and wishes in order to flexibly achieve an overarching goal (‘executive function’).136 In the two preschool classes constituted from children from the sample, there were also distinct sequelae of infant attachment. Children who so infuriated the teacher that they were sent to the corner were, on every occasion, those who had been classified as having an avoidant attachment relationship with their caregiver in infancy.137 Sroufe and colleagues found that the children who were identified by observers—unaware of the attachment classification—as unable to engage effectively in independent activities were those who had earlier been in dyads classified as insecurely attached. The children from ambivalent/resistant dyads were observed to exhibit chronic, low-level contact seeking and attention seeking. By contrast, the children who had earlier been classified as part of avoidant dyads did not seek contact from the teacher, except when the intimacy of the contact would be reduced, such as in large-group time.138 The preschoolers from insecure dyads were in general less socially competent than their peers, and found making and retaining friends more difficult.139

In the preschool classroom, the children were assigned into pairs and observed for a series of free-play interactions. The researchers found that only children with histories of avoidant attachment with their caregiver were seen to attempt to bully or victimise the other children. When children ended up in a sustained role of bullied or victimised, these were always children who had earlier received an insecure classification with their caregiver in the Strange Situation.140 Sroufe and colleagues were interested in the fact that children with experience of an avoidant attachment relationship could assume either role, victimiser or victim. They interpreted this as evidence that, in receiving rejection or rebuff from their caregiver in response to their attachment behaviour, the infant in an avoidantly attached dyad is learning both how to reject and how to be rejected by others. Later researchers confirmed the link between infant avoidant attachment relationships and later aggressive behaviour, with meta-analytic research reporting an association of d = .58 for observation-based studies.141 Sroufe (p. 363) and colleagues put forward a hypothesis to account for the association between avoidant attachment relationships and later aggressive behaviour.142 They argued that the attachment relationship serves as a non-linear prototype for later relationships, with avoidant attachment based on a history of unavailability, disappointment, and frustration. Viewing others as hostile and remote, children with avoidant attachment relationships may respond with aggression fed by their frustration, leading in turn to further rejection,143 and selective association with aggressive peers.144

Curiously, three decades later, this remains one of the only hypotheses proposed for the association between avoidant attachment and aggressive behaviour, and no test of the hypothesis has ever been specifically conducted.145 Writing with Michael Rutter, Sroufe later acknowledged that the exact mechanism leading to conduct problems remains unclear: ‘Avoidant attachment in infancy is associated with later conduct problems, but avoidant attachment is associated with low self-esteem, negative attributional biases, and rejection by peers and teachers, all of which may predispose to conduct problems. We are only beginning to understand how these chain effects operate.’146 One factor that may have inhibited the generation and testing of hypotheses in this area has been the growth of findings that suggest that disorganised infant attachment is a somewhat better predictor of later externalising behaviours than avoidant attachment.147

In their study of preschool behaviour, the Minnesota team additionally found differences in the play of the children, in pairs, and also in other contexts. Preschoolers who had been classified as secure had narratives to the stories and games they made that routinely assumed a positive resolution to dangers, sickness, or other serious problems. If someone was hurt, they were taken to hospital and made better. This follows what Waters and Waters later called the ‘secure base script’ (Chapter 2). Children from avoidantly attached dyads were less likely to create such secure base stories. In fact, ‘what is more noteworthy is the almost complete (p. 364) absence of fantasy play concerning people. Such fantasies dominate the play of almost all preschool children and were well represented in the play of those with secure histories in our sample. These data reveal sharp contrasts in the working models of the two groups—one world is richly peopled, the other is not.’148 This finding was partially supported by later research with the sample. As discussed in Chapter 3, when participants were asked to complete a family drawing, the drawings of children from avoidantly attached dyads were not found to be less populated than those of other children. However, adult figures were assessed by coders blind to the infant attachment classifications as showing more emotional distance between figures, more tension or anger, and figures were more likely to have their arms rigidly held at their sides rather than offering contact (e.g. holding hands).149

When the children in the sample were aged either five or eight, thirty families who also had another child no more than three years younger were called back to the lab. The parent was asked to leave and the siblings were given a set of games to complete together. For instance, in one game the siblings had to build a house out of blocks; in another the siblings had to work together to write using a stylus. The Minnesota group found that in sibling dyads where the elder sibling took on the role of guide in an effective way, this elder sibling was more likely to have had a history of secure attachment as assessed in earlier Strange Situations (r = .37).150 Furthermore, the extent to which the elder sibling took the role of guide, and the extent to which they treated their sibling with sensitivity, independently predicted observer and teacher assessments of their social competence at school, over and above attachment relationships. There was no association between attachment classifications and measures of how much fun the siblings seemed to be having together or their cooperativeness. As such, the Strange Situation seemed to predict, not a positive relationship in general, but specifically secure base provision by older siblings to their younger brother or sister. This aligned with Sroufe’s idea that children learn both roles in their relationships with attachment figures. This quality of the sibling relationship, even accounting for parent–child attachment, was found in turn to predict the elder sibling’s peer competence at school. Sroufe and colleagues observed that such findings offer evidence for the potential importance of the attachment components of the relationships with siblings in the development of peer competence.151

Overall, the Minnesota Longitudinal Study of Risk and Adaptation was able to offer ample evidence that early attachment experiences are associated with early social competence. This provided suggestive evidence towards a transactional model of individual–environment interactions, in which experiences in relationships with attachment components contribute to forms of social competence, perhaps especially through the availability of secure base scripts. Slade and Holmes observed that the preschool findings at Minnesota were of particular importance for establishing attachment research as a paradigm within developmental psychology.152 Sroufe, Egeland, and colleagues situated their findings as demonstrating the predictive validity of the Ainsworth Strange Situation, for instance in the finding (p. 365) of an association between avoidant attachment and later conduct problems. Indeed, this association has gone on to be well replicated. In retrospect, however, it is interesting that in their analyses the Minnesota group did not quite treat the Strange Situation as the priviledged measure of early attachment that it would become subsequently.153 For instance, the Minnesota researchers would at times composite the Strange Situation with, or study its interactions with, other measures of early care. Sroufe and Egeland appear to have felt that ‘attachment’ itself cannot be directly measured, even by the Strange Situation, but must be inferred from the network of correlations that link early care to a child’s later behaviour in theoretically expectable ways (Chapter 2). Having established correlations in early childhood, a next step in appraising the influence of early attachment relationships would be to see whether associations could still be found as the cohort matured into late childhood and adolescence.

Adolescence

When the sample reached 11 years old, in 1986, 48 children were recruited to attend a summer camp lasting four weeks. At this point, research on adolescence within academic psychology was on the cusp on becoming established: empirical study of adolescence barely existed in the late 1970s, and yet would be a central concern of developmental psychology by the early 1990s. The Society for Research on Adolescence was founded in the winter of 1984, and its first meeting was held in in Madison, Wisconsin, in 1986.154 Adolescence was an object of increasing public and policy concern. However, more proximally, the attention given to the topic in developmental psychology can also be situated in the context of the maturing of a number of longitudinal cohort studies established in the 1970s. The Minnesota Longitudinal Study of Risk and Adaptation was a few years ahead of this general curve.

In the mid- to late 1980s, a central concern in the research literature was with how to achieve the psychological measurement of aspects of adolescent social development in an ecologically valid way.155 The use of three summer camps with members of the cohort study was an ingenious solution. Most of the children were drawn from the subsample who had been studied in the preschool and who were already therefore stratified by attachment classification. Each of the three camps had 16 children. The programme of activities included group games, singing, swimming, art and craft projects, and sports. Weekly day trips were taken to recreation parks, and there was also one overnight camp-out.156 During the camp, the children were observed, and in the last week each child was interviewed twice about his or her experiences of the camp and relationships with other campers.

Sroufe and colleagues examined the maintenance of appropriate gender boundaries at the summer camp as one particular index of social competence at 11 years of age. The construct of ‘gender boundary violations’ encompassed two rather different kinds of (p. 366) behaviour. One was hovering near opposite-sex groups or joining in their activities. The other was forms of sexual harassment or cross-gender aggression. No report was made on whether these behaviours were correlated. These two kinds of behaviours were aggregated by the researchers, though the way that the scale was constructed gave more weight to sexual harassment or cross-gender aggression than cross-gender interaction in general.157 Elicker, Englund, and Sroufe found that there were marked negative associations between gender boundary violations and peers’ ratings of campers social skill (r = –.33).158 Children with a history of ambivalent/resistant relationships engaged in gender boundary violations much more frequently. They also sat next to opposite-gender children during large-group circle times more often. Gender boundary violations were predicted by earlier intergenerational boundary dissolution in early childhood.159 Gender boundary violation would turn out to be an important variable in the Minnesota study, with strong predictive value for later outcomes. For instance, gender boundary violations at age 11 had a negative association with contemporary (r = .58) and later (r = .31) competence with peers in social situations,160 lower satisfaction in adolescent dating relationships (r = .47),161 and a dramatic association with age of commencement of sexual activity for girls (r = .75).162

During the summer camps, the 11-year-olds with histories of avoidant attachment relationships were indistinguishable from those with secure relationships when the tasks and challenges of the camp related to physical tasks or the inanimate environment. They also appeared confident in the first day of the camp, with less of the ordinary clumsy vulnerability and reserve of the other early adolescents. However, difficulties in social situations became more apparent over time.163 For example, they were less effective when given tasks that required inferences about the thoughts and feelings of others, and they displayed a higher level of negative bias in their evaluations of their own group when asked to engage (p. 367) in groupwork.164 Kagan argued that the unruffled behaviour of infants in avoidant dyads in the Strange Situation did not represent any kind of problem. He suggested that it may just reflect infants with a bold temperament, or infants who had been socialised by their parents to healthy independence.165 Elicker, Englund, and Sroufe argued that the sequelae of infant avoidant attachment showed that neither interpretation was especially plausible. Instead, the behaviours shown at the summer camp by participants who had been in avoidant attachment relationships as infants could better be regarded as an effect of their experience of less caregiver availability as a safe haven to help develop successful emotion regulation and social competence.166

However, even if attachment does not begin as a trait, the Minnesota group concluded that individual differences related to attachment may become stable with development. When the children were recruited back for a summer camp reunion weekend at age 15, assessments of peer interaction were conducted. Whereas there was little stability in individual behaviours, global assessments of peer competence were very stable between the summer camp and the reunion.167 A curious finding reported by Sroufe, Egeland, and Carlson was that peer competence at the reunion was predicted well by Strange Situation classifications and preschool behaviour a decade earlier, with little additional prediction accounted for by observations of social skills in the intervening years. The researchers were led to conclude that ‘by age 5, much of the variance in adolescent social competence can be accounted for. Even though numerous capacities are only nascent or not at all apparent in the preschool period (e.g. coordinated group activities, intimacy).’168 Soufe and colleagues highlighted that since preschool social competence was itself predicted by infant Strange Situation classifications, there seemed to be both a direct and an indirect effect of attachment on peer competence through childhood and adolescence. The researchers attributed this especially to the role that secure base and safe haven availability have for the modulation of emotion. Such capacity for modulation can contribute to skills in handling closeness in the face of potential peer conflicts and to tolerating the stimulation of group activities.169 The emphasis on early attachment was also supported by another finding from the reunion weekend, reported by Englund, Levy, Hyson, and Sroufe in 2000. Within the group task at the reunion, the groups were asked to elect a spokesperson to report back to the wider group. The long-term implications of early experiences within attachment relationships were suggested by the fact that ‘out of a total of 16 adolescents named spokesperson, 3 were (p. 368) classified as insecurely attached in infancy and 13 were classified as securely attached in infancy’.170

W. Andrew (Andy) Collins was the Director of the Minnesota Institute of Child Development from 1982. One of his particular research interests was the psychology of teenagers, and when the sample of the Minnesota Longitudinal Study of Risk and Adaptation reached this age he became heavily involved with the cohort study. At age 16 years, participants were invited back to the laboratory and participated in interviews about their dating and social lives, considered by the researchers to be key developmental challenges of mid-adolescence. Participants who had been in dyads classified as ambivalent/resistant in the Strange Situation were less likely than other participants to have dated, a finding Collins and Sroufe attributed to the relative emotional and social immaturity of this group. Participants who had been in dyads classified as securely attached in infancy were more likely to have had a romantic relationship that lasted at least three months.171

When the sample were 20–21 years old, 78 participants who were in a romantic relationship were invited to the laboratory along with their partner. Each member filled out a battery of self-report assessments of their relationship, and were also observed completing a collaborative task and discussing a touchy subject for their relationship. First publications from analysis of these data began to appear in 2001.172 Study of the observations of the callback at 20–21 years was supported by the appointment of Jeff Simpson as a new professor at Minnesota in 2004. Simpson’s background as a social psychologist in the self-report tradition of attachment research meant that he was well equipped for the study of adult romantic relationships.173 The Minnesota group reported that infant attachment as assessed by the Strange Situation predicted several aspects of adult relationship functioning, including a robust association between infant disorganised attachment and the couple’s hostility with one another (r = .42).174 They found that secure infant attachment assessed by the Strange Situation was also associated with twice the rate of effective resolution of conflict among the young adults regarding the touchy subject.175

Other variables were also found to be very important. One concurrent factor was gender. Associations between Strange Situation classifications in infancy and the behaviour of the couple were generally stronger for men than for women. The researchers observed that male participants tended to set the tone in the collaborative task, even when female participants (p. 369) had been confident and forthright in individual interviews.176 Other antecedent variables were also highlighted.177 Of particular importance was the finding by Simpson, Collins, Trans, and Haydon that the link between Strange Situation classifications and interaction with the romantic partner was partially mediated by the quality of peer relationships at school age.178 Though their findings qualified the emphasis on continuity of Bowlby’s early and populist writings, the researchers regarded their findings as confirmation of the proposal of developmental pathways from Bowlby’s late theoretical writings (Chapter 1).179 Aspects of early care remained relevant to some aspects of adult functioning, even after decades.180 The factors proposed by the Minnesota group were of particular importance for these developmental pathways, such as emotion regulation and effective social relationships, and were also those that would suggest links to mental health. Following the centrality of mental health to Bowlby’s conceptualisation of attachment and its implications, the Minnesota group gave this topic special attention.

Mental and physical health

When child participants in the cohort study reached 17 years of age, they were brought back for a detailed psychiatric interview. Other assessments were conducted at the time, including a holistic interview with the young people about their school, work, living situation, and dating life—and a parallel interview about participants with their mother.181 If, as Slade and Holmes argued, the preschool study at Minnesota provided important support for attachment research as a paradigm in the 1990s, a parallel role was played by the findings from the psychiatric interview at age 17 in the 2000s, reported by Carlson.182 In general, Kobak and Bosmans observed that infant attachment has ultimately been disappointing as a predictor (p. 370) of adult mental health, with ‘lack of specificity and relatively small effect sizes’.183 However, Carlson’s report helped to support the idea that early attachment as measured by the Strange Situation does play a distinct and important role, even if this role is of less magnitude than Bowlby tended to suggest. In line with the general preference at Minnesota for composite measures, Carlson treated the psychiatric interview as a measure of overall mental health, coded on a seven-point scale. She did not differentiate by kind of mental health problem. She found that infant avoidant attachment, infant disorganised attachment, middle childhood behaviour problems, and poor observed child–parent boundaries at age 13 each made an independent contribution to total mental health problems at age 17, together accounting for a third of the variance.

In the past few years, meta-analytic research has led several senior figures in the developmental tradition of attachment research to query the conclusion drawn by Carlson, and taken by her readers, that Strange Situation classifications are associated with general mental health problems. Meta-analytic work indicated that infant avoidant and disorganised attachment in the Strange Situation tended to be much more associated with externalising problems, and less with internalising problems such as anxiety or depression.184 In fact, Carlson was not actually intending to make the specific claim that avoidant and disorganised attachment would be generally associated with all mental ill health185—though technically this is what she reported and what readers naturally understood. Instead, her report was based on the Minnesota creed of using composite measures, especially for pioneering work, with the expectation that later researchers would pursue differentiation.186 Today, the question of whether the Strange Situation predicts primarily externalising problems remains an area of ongoing discussion among researchers, not least since several phenomena of clinical concern (e.g. PTSD, ADHD) are poorly captured by the division between externalising and internalising mental health issues.187

As well as studying prospective links between the Strange Situation and general mental health at age 17, building on the ideas of Main and colleagues (Chapter 3) Carlson conducted a hypothesis-driven analysis of associations with dissociative symptoms. She found that infant avoidant attachment, infant disorganised attachment, and middle childhood behaviour problems made independent contributions, and together accounted for 17% of the symptoms. Infant disorganised attachment and child abuse also made independent contributions to later self-injurious behaviour and other symptoms of borderline personality disorder. The association between the Strange Situation classifications and later borderline symptoms was (p. 371) mediated by responses to the family drawing task at age eight and interview measures of representations of the self and key relationships (including the ‘friendship interview’) in early adolescence.188

Another paper by the Minnesota group reported findings that infant ambivalent/resistant attachment was distinctively associated with anxiety problems in adolescence (r = .26), even controlling for maternal anxiety and child temperament, but had no other links to mental illness.189 This was important in providing a source of prospective validity for the category. Following Main (Chapter 3), Sroufe proposed that the infant ambivalent/resistant attachment pattern entails a strategy of hypervigilance and hyperattentiveness regarding the availability of the caregiver: ‘Such a stance may be adaptive in insuring contact with the caregiver when there is a genuine threat, but a price is paid for such chronic wariness and vigilance.’190 However, attempts to replicate the finding of a link between infant ambivalent/resistant attachment and later anxiety problems have failed, at least when observational and interview measures have been used.191

Another finding that ran contrary to expectations was a report by the Minnesota group using a composite of Strange Situation classifications and measures of early caregiving. The researchers examined the association of this composite with depression at age 17, anticipating based on available theory that insecure attachment and early care would both be robust predictors of later depression.192 Yet Duggal, Carlson, Sroufe, and Egeland reported in 2001 that in the Minnesota study the composite accounted for only 3% of variance, controlling for maternal depression. Furthermore, it lost significance entirely in a regression once account was taken of the amount of support from family and friends the mother received in the child’s first five years. By contrast, prospectively documented child abuse remained significant in predicting depression: together, maternal social support and documented child abuse accounted for 19% of variance in adolescent depression.193 However, intriguingly, (p. 372) more recently when the sample were followed up to adulthood, symptoms of depression were more likely to remit among those participants who had been in a secure attachment relationship as infants.194 Early attachment relationships may play less of a role in predisposing or triggering depression than in stabilising symptoms.

Somewhat better specificity and greater effect sizes were seen with respect to responses to trauma. Enlow, Egeland, and colleagues reported an analysis of PTSD symptoms in data from the adolescent psychiatric interview. Rates of PTSD were doubled for dyads where mothers were assessed as showing emotional neglect/psychological unavailability.195 There was no association between Strange Situation classifications and later traumatic experiences per se.196 However, for those participants with insecure attachment relationships who did experience later trauma, there was a dose-response relationship with PTSD symptoms: 12% of infants in dyads classified as secure at both 12 and 18 months in the Strange Situation developed PTSD following trauma exposure; 28% received the diagnosis when assessed as belonging to an insecure dyad in either the 12- or 18-month Strange Situation; and 46% had PTSD in adolescence following trauma exposure if they had an insecure attachment relationship at both 12 and 18 months.197 Disorganised infant attachment appeared to make no additional contribution beyond insecurity.

Sroufe, Carlson, Levy, and Egeland also reported findings regarding the respective contribution of Strange Situation classifications and middle childhood behaviour problems to later mental health. Children with a history of secure attachment relationships who showed few conduct problems at school had the lowest scores for mental illness. Children with a history of insecure attachment relationships who showed significant conduct problems at school had the very highest scores for mental illness, and the lowest scores for their education and social relationships. The researchers reported, however, that children with a mixed history (secure attachment and later behaviour problems; or insecure attachment and later positive functioning) had equivalent scores. They were likewise functioning at an equivalent level in their education and social relationships. Sroufe and colleagues concluded that children who were struggling in middle childhood but who had a secure attachment relationship in infancy ‘drew upon a more positive foundation’, which accounted for their better outcomes than their peers with insecure infant attachment relationships.198

In educational research in the 1990s, adolescent conduct problems and poor achievement were identified as cardinal predictors of young people dropping out of school. The Minnesota Longitudinal Study of Risk and Adaptation was unusual in being able to examine (p. 373) this question prospectively, including variables from early childhood. A regression analysis indicated that the child’s gender, the degree of chaos in the early home environment, and a composite of infant attachment and caregiver sensitivity could predict 77% of children who had dropped out of school by age 16. When these variables were included, more proximal variables such as truancy, disciplinary problems, and failing grades in high school were no longer significant. The same was true for peer rejection, conduct problems, and low school achievement. As such, Jimerson, Egeland, Sroufe, and Carlson concluded that ‘many established “predictors” may be better conceptualized as markers of presence on the pathway’, leading from early caregiving to dropping out of school. The reason for this, they argued, is that ‘success in school calls upon numerous capacities for self-regulation that begin to be formed in the early years’.199 Raby and colleagues, in fact, later documented that early maternal sensitivity predicted academic achievement by adulthood at least as well as it predicted social competence, controlling for covariates such as maternal education and socioeconomic status.200 The picture is complicated, unsurprisingly, by the fact that experiences of education have a bidirectional relationship with development. Englund, Egeland, and Collins elsewhere documented that developmental outcomes could be altered, for some children, by their experience of the school itself.201 As we saw in Chapter 2, Ainsworth anticipated that feelings of security in different domains would influence each other. And Bowlby specifically identified that a school may serve attachment functions for a young person.202

Though social psychologists had pursued research on the topic for over a decade,203 in the 2010s the relationship between attachment and physical health outcomes was heralded as a ‘new frontier’ for developmental attachment research.204 Puig, Englund, Simpson, and Collins examined the links between early attachment and later medically unexplained or ‘non-specific’ symptoms at age 32, controlling for a variety of covariates including adult socioeconomic status, life stress, and social support. Individuals who had been part of ambivalent/resistant dyads in infancy as assessed in the Strange Situation were three times more likely to report symptoms such as dizzy spells, headaches, skin trouble, back ache, and indigestion than those adults who had been part of secure dyads. They were also six times more likely to report physical health conditions. Individuals who had been part of avoidantly attached dyads in infancy were three times more likely than those from secure dyads to report inflammation-related illnesses such as heart disease, high blood pressure, and asthma. Participants who were part of dyads classified as secure at both 12 months and 18 months reported the fewest physical illnesses in adulthood compared to those who were part of dyads classified as insecure at one or both assessments.

(p. 374) The researchers proposed that caregiver provision of a secure base and safe haven in the context of stable routines offers an external source of biological, as well as psychological, regulation. Insensitive care may heighten stress reactivity across the lifespan, inscribed on the body in the form of ‘non-specific’ symptoms like headaches, skin trouble, and back ache, where physical and psychological stress together may contribute to symptom development and maintenance over time. Conversely, sensitive care may reduce stress reactivity across the lifespan, leading to fewer non-specific symptoms and fewer symptoms relating to the human inflammatory response, which is activated by alarming or threatening experiences.205 This hypothesis has been confirmed by other findings from the Minnesota group showing that assessments of maternal sensitivity blocked the association between stressful life events and physical ill health, and also predicted a lower body-mass index and fewer physical symptoms for participants with fewer stressful life events.206 Also relevant to this hypothesis were findings that child neglect contributed to adult physical health according to both self-report measures of physical symptoms and biomarkers of cardiometabolic risk.207

The study of attachment and health has been an exciting area of research in the social psychological tradition over the past decade.208 It seems likely that investigation of the implications of attachment for physical health is likely to be a growth area over the coming years in the developmental tradition, pursuing the pioneering and striking findings from Minnesota.209 This will be facilitated by the collection of genetic data from the cohort. It will also be facilitated by the fact that a number of new large cohort studies were initiated in the 2000s in Europe with a rich array of measures of physical health. For example, the Generation R cohort study in the Netherlands includes Strange Situation procedures with 900 infants, a number of observational measures of parent and child behaviour, and a vast array of assessments of physical health over the subsequent decade, as well as genetic information on the sample.210 The prospect of new integrations between attachment research and biological and medical research using the Minnesota Longitudinal Study of Risk and Adaptation and other cohort studies seems not only possible, but likely.

Another area where future reports are to be expected is on three-generational processes. The Minnesota group recently followed up the grandchildren of the original women recruited to the sample. Strange Situation procedures were conducted with these grandchildren. Raby, Steele, Carlson, and Sroufe found little evidence of intergenerational continuities in attachment security. One pattern in the data was that mothers who had themselves had insecure attachment relationships in the 1970s but who formed secure attachment relationships with their own infant had higher-quality concurrent support from family and friends. Though (p. 375) there was little continuity between generations in security or insecurity in general, there was specifically substantial continuity in disorganised infant attachment: half of the mothers who had been part of dyads that received a classification of disorganised attachment in the 1970s had children who received the same classification. This was in contrast to only 20% of dyads where the mother had no history of an infant disorganised attachment classification.211

Dissociation

For Sroufe, an individual’s experience of ‘self’ was best conceptualised as an emergent organisation, constituted through the routine integration of cognition, emotion, and behaviour. Within this organisation, the whirring clockwork that sustains ordinary adaptation is assembled out of both historical and contemporary machinery. Earlier components are incorporated into later behavioural systems and may remain accessible. This produces the strange and uneven inner space of thoughts and feelings characteristic of human experience, as well as contributing to the way that our own behaviour has the capacity to surprise or disconcert us. Sroufe’s emphasis, however, was above all on the coherence of the assemblage of components, an expression of his methodological principle of holism.212 He regarded this constituted organisation as experientially stabilised by memory processes, which register continuities in an individual’s qualities and capacities across contexts and relationships.213 Nonetheless, memory, expectations, and images of the value of the self in the eyes of others can become a facet of self-monitoring and a guide for behaviour, filtering the meaning of later social experiences. Experiences in relationships, in turn, contribute to the further stabilisation or reorganisation of the inner organisation: ‘the self was forged within vital relationships; within such relationships, it continues to evolve’.214

The identification of selfhood with organisation over the course of development contributed to the special attention given by Sroufe and colleagues to dissociative symptoms. In Defences that Follow Loss, Bowlby had situated dissociation as a form of defensive exclusion, a psychological process that blocked episodic information about the perception of events in time and space from reaching and activating a behavioural system (Chapter 1). It was a ‘particular sort of segregating process’, and in this had elements in common with attachment avoidance, even if Bowlby believed that the two were distinguishable.215 However, Bowlby did not publish his speculations on dissociation. Though he had Robertson’s qualitative observations, he lacked the data to press further. Writing in 1996, Egeland and Susman-Stillman (p. 376) acknowledged that, in the wake of Bowlby’s threadbare published comments, ‘there is some confusion over the relation of attachment and dissociation’.216

The Minnesota group included questions relevant to child dissociative symptoms in twelve different assessments made by mothers and teachers, the first in kindergarten and the last at 16 years. At age 19, the 168 participants remaining in the study provided self-report of their own experiences of dissociative symptoms on the Carlson and Putnam Dissociative Experiences Scale.217 Mothers were asked about potential traumatic life events experienced by their child during the 12, 18, 30, 42, 48, 54, and 64 month assessments and when their child was 19 years old. At 19 years, participants themselves were also asked to complete this assessment. Ogawa, Sroufe, and colleagues found that in preschool, dissociative symptoms were predicted well by neglect in infancy.218 In elementary school, 27% of variance in children’s dissociative symptoms was predicted by their mother’s experience of abuse as a child, physical abuse experienced by the child, and avoidant attachment in the Strange Situation. In both preschool and elementary school, the child’s IQ made a small negative contribution to the score. In adolescence, the best predictor of dissociative symptoms was avoidant attachment assessed in the Strange Situation. Additional contributions were made by exposure to domestic violence in infancy, disorganised attachment as assessed in the Strange Situation, and experiences of physical abuse in adolescence. Together these variables accounted for 19% of variance in dissociative symptoms. At age 19, 30% of variance in dissociation could be accounted for by only three variables, all from infancy: caregiver psychological unavailability, infant disorganised attachment, and maternal-report measures of infant distractibility and attention-span. Psychological unavailability in infancy alone accounted for 19% of variance. It was curious that, besides infant attachment, the 19-year assessment was predicted so powerfully by two variables that had previously not proven relevant, both from infancy. However, it should be noted that the 19-year assessment was a self-report of dissociative symptoms by the young persons themselves, whereas previous waves had been maternal or teacher report.

Ogawa, Sroufe, and colleagues also examined factors that predicted dissociative symptoms in the clinical range, participants with high but not clinical levels of symptoms, and participants with few symptoms. In preschool, they found that infant neglect predicted symptoms in the clinical range, whereas witnessing domestic violence and parental psychological unavailability predicted high but not clinical levels of symptoms. The two groups were quite distinct in terms of their antecedents. Witnessing domestic violence was also associated with high but not clinical levels of symptoms in elementary school. Symptoms in the clinical range were differentially predicted by neglect in infancy, infant disorganised attachment in the Strange Situation, and concurrent physical abuse experienced by the school-aged child. In adolescence, the clinical group was distinguished strongly by infant avoidant and infant disorganised attachment in infancy. High but not clinical levels of symptoms were predicted by concurrent physical abuse. In early adulthood, the clinical group was distinguished from the others on the basis of maternal psychological unavailability in infancy and (p. 377) disorganised attachment in the Strange Situation. The authors interpreted their data as suggesting two pathways. A first pathway was to clinical levels of dissociation, based to a surprising degree on experiences in infancy. These include experiences of neglect in infancy, and of avoidant and disorganised attachment. In fact, ‘an examination of the zero-order correlations between disorganized attachment and dissociation at each time period reveals a linear trend with a positive slope (Time 2 r = .14, n.s.; Time 3 r = .20 p < .05; Time 4 r = .25 p < .01; Time 5 r = .35 p < .001)’.219 A separate analysis was run which found that disorganised infant attachment alone marginally elevated dissociative symptoms in early adulthood, but that disorganised attachment and the experience of at least one form of trauma had a strong link to dissociative symptoms (F[2, 125] = 12.0, p < .001).220

The second pathway was to subclinical but substantial levels of dissociative symptoms. This pathway was predicted by concurrent physical abuse, and tended to stop once the participants reached the greater independence of young adulthood. Ogawa, Sroufe, and colleagues concluded that ‘psychopathological dissociation should not be viewed as the top end of a continuum of dissociative symptomatology, but as a separate taxon’.221 However, they highlighted that avoidant attachment appeared to make an independent and robust contribution to dissociative symptoms, except to the participant self-report measure. Whereas dissociation reduces environmental responsiveness by blocking a form of perception, avoidance retains awareness of the environment whilst bracing against the attachment system, directing attention away from cues that might activate it (Chapter 1). Nonetheless, Ogawa, Sroufe, and colleagues reflected that the contribution of infant attachment to later observed dissociative symptoms perhaps lies in the fact that avoidance may at times shade into dissociative processes, or may predispose a vulnerability to dissociation. Nonetheless, the lack of correlation between avoidance and self-report of dissociation required explanation. The Minnesota group drew on Bowlby’s published reflections to speculate that avoidant attachment as a strategy of defensive exclusion may be more mature in young adulthood, in the context of cognitive maturation and social independence, allowing the strategy to work more effectively, with less ‘leakage’ in the form of dissociation. It might also be supposed that avoidant attachment strategies could suppress awareness and/or reporting of dissociative symptoms.

Giovanni Liotti had proposed that infant neglect and avoidant and disorganised attachment may all have in common that they disrupt the early integrative processes that pull together behavioural systems (Chapter 3).222 Building on these ideas from Liotti, Carlson, Yates, and Sroufe suggested that dissociation may be regarded as a strategy of self-regulation in the face of threat, where such integration has been poor, and where the caregiver cannot be approached directly to offer a safe haven and an external source of integration.223 Experiences of domestic violence and physical abuse that a child cannot escape are liable to cause some dissociative symptoms, as a strategy developed for the purposes of (p. 378) self-regulation. However, clinical levels of symptoms are more likely when the need for this strategy occurs in infancy, since it may be used pervasively in the absence of other strategies and because infancy may be a period of special importance for the formation of behavioural systems and psychological coherence. This might account for the interaction between trauma and disorganised attachment. Traumatic events are those that can be anticipated to overwhelm established coping strategies. They may therefore be a catalyst for dissociative symptoms for those children who were unable to lay a solid foundation for integration and emotion regulation in infancy.224

More recently, Haltigan and Roisman reported findings from the NICHD Study of Early Child Care and Youth Development of 1,149 families.225 Dissociative symptoms were assessed by mothers at six time-points and by teachers at three time-points between age 4 and age 15. Participants themselves reported symptoms at age 15. The researchers found no evidence that infants from dyads classified as disorganised in the Strange Situation later showed dissociative symptoms. Infants from dyads classified as avoidant did have elevated dissociative symptoms as rated by their teachers and mothers, though not in their self-report. Haltigan and Roisman commented that ‘the claim that infant attachment disorganization represents a unique risk factor for the development of dissociative symptomatology has been given much attention and cachet in the literature, perhaps because of a theoretical account that provides an especially compelling rationale for this association. That said, the bedrock of science is the replicability of its findings.’226 Given that the NICHD study is a community-risk sample, and the Minnesota sample was selected specifically on the basis of adversities, Haltigan and Roisman raised the prospect that infant disorganised attachment may contribute to later dissociative symptoms in interaction with severe or chronic trauma. A link might be drawn here with Main’s prediction (Chapter 3) that infant disorganised attachment will usually resolve by adulthood unless the basis of the disorganisation lies directly in child maltreatment.227 Most researchers have, however, responded by ignoring the Haltigan and Roisman study, which remains relatively comparatively cited four years after its publication. There has been informal discussion of whether the NICHD sample was coded effectively for disorganisation, given that the classification does not predict well in this cohort study. However, a further possibility, one with quite serious implications for attachment research, is that the forms of disorganised attachment may differ between high-risk and low-risk samples (Chapter 3). These questions will likely be resolved over the coming years as more recent cohort studies enter adolescence and adulthood, if funds can be secured for later follow-up and measures are used to assess dissociative symptoms.

(p. 379) Developmental psychopathology

The domain of developmental psychopathology

The work of Sroufe and Egeland was fundamental to the emergence of developmental psychopathology, as a movement within developmental psychology.228 In turn, developmental psychopathology has supported the continuation of attachment theory as a significant reference-point within contemporary developmental psychology. An important part of the legacy of the Minnesota Longitudinal Study and of Sroufe and Egeland as attachment researchers is therefore revealed through consideration of the rise of developmental psychopathology.

Though based on earlier ideas,229 the movement was initiated primarily by a conference hosted by Norman Garmezy and Michael Rutter at the Center for Advanced Study in Behavioral Sciences at Stanford from 1979 to 1980.230 The conference was sponsored by the William T. Grant Foundation, as part of a growing public and policy concern about child abuse and maltreatment, contributing to an increased availability of research funds in this area.231 The meeting was prompted especially by findings from high-risk longitudinal studies set up around specific diagnostic entities such as schizophrenia that nonetheless highlighted the role of poverty, contextual risks and conflictual family relationships in the prediction of a variety of mental health outcomes for individuals and across generations.232 The compounding of adversities seemed to be much more predictive than any ‘main effect’ from imputed single causes. The researchers’ concerns were also galvanised by interest in the potential role of individual and social protective factors in buffering risks faced by individuals and families.233 Garmezy and Rutter drew on the term ‘resilience’ in conceptualising (p. 380) this process.234 The proceedings of the Stanford conference were published by Garmezy and Rutter as Stress, Coping, and Development in Children in 1983.235

Also in 1983, Sameroff proposed the dual-risk model of psychopathology, in which the particular role of multicausal models of cumulative risk factors was highlighted in the role of mental illness.236 Emergent longitudinal findings appeared to show that any one risk factor had little association with later disturbance. What seemed to be critical was the combination of liabilities and supports. Additionally, in his 1983 chapter, Sameroff began to sketch the ideas that would later form his distinction between ‘protective’ supports and ‘promotive’ support for development, the former contributing to beneficial stability and the latter contributing to thriving.237 Overall, this account firmly emphasised the contextual embeddedness of mental health, and the transactional relationship between different obstacles and difficulties individuals face in their lives. As mentioned above, a presentation by Sameroff of his emerging ideas prior to this publication proven influential in the choice of initial measures in the Minnesota Longitudinal Study of Risk and Adaptation.

Garmezy and Sroufe were colleagues and friends, and taught classes together at Minnesota, where there had been growing momentum around developmental approaches to mental health since the 1970s.238 Sroufe and Rutter’s 1984 article ‘The domain of developmental psychopathology’ was a turning-point for developmental psychopathology, as the movement resolved into an explicit and self-identified alternative paradigm for the study of mental health. Whereas the dominant medical model started from adult diagnostic (p. 381) categories and explored precursors, developmental psychopathology was defined by a concern to study ‘the origins and time course of a given disorder, its varying manifestation with development, its precursors and sequelae, and its relation to nondisordered patterns of behaviour. Thus, developmental psychopathologists may be just as interested in a group of children showing precursors of a disordered behavior pattern, but not developing the disorder proper, as the group that in time manifested the complete pathology.’239

Sroufe and Rutter highlighted that the distinction between adaptation and pathology was context-dependent. They give the example of an avoidant response to an abusive attachment figure, which might include not only physical avoidance but also ‘blunting or controlling emotional experiences’. This response is ‘adaptive’ in the sense of being intelligible and helpful for a young child, for whom proximity or displays of distress or desire for comfort may not be safe. However, the very same adaptation ‘may later compromise the child’s ability to maximally draw upon the environment’.240 Even stereotypic behaviour (Chapter 3) could be seen in this light: these odd, seemingly purposeless behaviours can best be regarded as ‘meaningful’ responses that at some point helped modulate arousal, and which have therefore been retained as a response to stress.241 Sroufe and Rutter argued that adult mental health should be understood as, in part, the residue of the history of ways in which an individual responded to earlier challenges, and in part a new, inventive attempt to turn existing expectations, skills, and strategies towards contemporary problems and opportunities.242 Where continuities are seen across periods of development these will rarely be in the display of identical forms of behaviour, but rather in the meaning of the behaviours as a way of engaging with the environment. For instance, there are vast differences between weeping and clinging in infancy, on the one hand, and impulsivity, jealousy, and continual dramas in adulthood, on the other. But both profiles may serve similar interpersonal functions and have their basis in a low threshold of activation of the attachment system.

As such, Sroufe and Rutter made a plea for broad-band indices of adaptation, whether in peer-relations or mental health. In line with Sroufe and Waters’ earlier ‘Attachment as an organizational construct’ (Chapter 2), Sroufe and Rutter argued that little stability of behaviour would be expected in the context of an individual’s changing developmental challenges and environment. By contrast, broader indices would predict well to later adaptation in general, since they would capture the extent to which adaptation had required significant compromise or even led to failures in meeting challenges in earlier life. In this, their argument lay the ground for proposals by Rutter’s former collaborator Avshalom Caspi and colleagues in 2014 for a general ‘p-factor’ representing general mental ill-health, rather than the division of symptoms into specific diagnoses.243

(p. 382) Though Sroufe tended to regard positive internal working models of self and relationships as characteristic of mental health,244 in his writings on developmental psychopathology he was keen to argue that there is no content of internal working models that is itself adaptive. Positive models are primarily beneficial when combined with environments in which expressions of positive models such as self-esteem or a capacity to trust can generate opportunities and rewards, which is not always the case. Against essentialism about either adaptation or pathology, in their 1984 article Sroufe and Rutter argued that it is the interaction with context that is critical for the meaning of even patterns of attachment.245 For example, though generally a protective factor over time in the wider context of development, secure attachment may in theory function variously as a vulnerability, protective or a risk factor, depending on local circumstances.246 Indeed, attachment was a central concern for developmental psychopathology, since it provided both theory and empirical results supporting the role of transdiagnostic, developmental processes for mental health, with any factor compounded by the network of other risk and protective factors experienced by a group or population.247

Though an outspoken critic of Bowlby’s notion of ‘maternal deprivation’, Rutter nonetheless praised Bowlby’s work as having ‘ushered in the era’ in which the developmental psychopathology perspective was possible, especially through his adoption of the concept of ‘developmental pathways’ from embryology (Chapter 1).248 Rutter also observed that ‘Bowlby’s (1951) World Health Organization monograph had raised expectations of the power of prediction from experiences in infancy and on the preventive value of psychosocial interventions, but also it had run into a storm of academic criticism. Both proved only half-right. The over-riding potency of infantile experiences has not been supported by subsequent research, but the focus on children’s attachments to their parents has paid off richly.’249 Sroufe, perhaps, if forced to give proportions, would likely have said that Bowlby was three-quarters right.250 Even if the direct effects of early attachment have been smaller (p. 383) than anticipated by Bowlby, the Minnesota Longitudinal Study of Risk and Adaptation provided ample evidence of the continuing importance of early experiences in some areas of life, such as enduring associations between early experiences of abuse and later relationships with peers and romantic partners.251 Looking back on the results of the study, Sroufe, Coffino, and Carlson concluded that ‘early experience can be conceptualized in terms of creating vulnerabilities or strengths with regard to later experience, including what experiences are sought and how they are interpreted, rather than as directly producing particular outcomes’.252

This emphasis on the power of prediction, for which Bowlby was a forerunner, was central to developmental psychopathology as a domain of inquiry. It was assumed that reasonable certainties were possible regarding the probabilities entailed by developmental pathways. In particular, hypotheses could be developed for testing against prospective data, to explore the ‘characteristics or histories that buffer individuals against stress or that provide them with attitudes, orientations, and skills that promote successful coping with stress, and … factors that produce vulnerability to stress or coping failures’.253 In wider social and political context, the emergence of developmental psychopathology can be regarded as aligned with a rise of public health and primary prevention approaches to mental health.254 It was also supported by a broader trend in social and policy science towards a concern with the capacities of individuals to handle the potential fallout of societal and economic uncertainties and problems.255 In turn, the prominence and credibility given to ‘resilience’ by the emergent field of developmental psychopathology helped contribute to its resonance and deployment across contexts in the 1990s—albeit generally completely shorn of developmental concerns, and formulated instead as a quality of individual ‘invulnerability’.256 In their contribution to Anthony and Cohler’s influential book on The Invulnerable Child, Farer and Egeland castigated the editors and other authors for what they regarded as a false and unethical framing of the question of adaptation in the context of adversity.257

Though considered one of the ‘founders’ of developmental psychopathology, Sroufe was also regarded as a distinct voice. He was known for his characteristic criticisms of medical models of diagnosis; his emphasis on lifespan development and intergenerational processes; (p. 384) his concern with the interdependence of different family relationships; his hostility to temperament researchers where they neglected the role of caregiving factors; and his dismay at images of ‘resilience’ that constructed it as an individual quality rather than a social effect.258 He and Egeland were also among the strongest advocates for the value of considering typical and atypical forms of development within the same frame, based on the assumption that the processes involved would be mutually revealing. Sroufe and Egeland argued that the role of many social factors in ordinary life is best revealed when they are absent or intensified to an unusual degree. They pointed, for instance, to the importance of the family and friendship networks available to the mothers in their sample: when there were improvements of social support to mothers who had otherwise been especially isolated, their children’s mental health and conduct in school saw dramatic improvements. However, in the sample in general, social support had only a modest overall relationship with child behaviour.259

Furthermore, the Minnesota Longitudinal Study of Risk and Adaptation led Sroufe, Egeland, and colleagues to a distinctive position on the role of risk factors across the lifespan. They agreed with the consensus in developmental psychopathology that risk factors were cumulative. However, they were unusual in offering some qualifications to this position. Kaufman and Zigler had argued that various risks were interchangeable, and what mattered was their number: for instance, various forms of child abuse were argued to have the same implications.260 The Minnesota group were definitely interested in the use of summary variables and the study of cumulative effects: they generally found stronger effects when multiple forms of abuse compounded. However, they held that risk factors would often interact in specific ways. Egeland reported that two-thirds of participants who had been physically abused showed oppositional defiant disorder. This was in line with other research groups who had found that physical abuse predicted clinically significant levels of conduct problems. However, the multi-assessment and longitudinal nature of the Minnesota data allowed Egeland to show that this association was, at least in his sample, a secondary effect of the links between physical abuse and other forms of maltreatment and inadequate care.261 Conduct problems were in fact best predicted by the co-occurrence of caregiver psychological unavailability, neglect, and sexual abuse. The Minnesota group also reported findings showing the implications of specific forms of abuse. Berzenski, Yates, and Egeland reported that emotional abuse, whether alone or in combination with other forms of abuse, made a particularly powerful contribution to anxiety and depression, while the combination of physical and emotional abuse was especially strongly associated with conduct problems.262 They also (p. 385) highlighted research findings from other groups that indicated that parental depression reduces the likelihood of physical abuse, but increases the likelihood of sexual abuse.263

The Minnesota group were distinctive among developmental psychopathologists for their interest in the capacity of risk factors at one stage to go dormant until the developmental challenges of a later stage made them salient again: ‘Even following change, early patterns of attachment retain a potential for reactivation … Certain issues and certain arenas of functioning—those tapping anxiety about the availability of others or apprehension regarding emotional closeness—will be especially likely to reveal the legacy of early attachment, even during periods of generally adequate functioning.’264 For instance, Yates, Dodds, Sroufe, and Egeland found that exposure to interpartner violence in the preschool years reappeared as a powerful predictive factor in contributing to behavioural problems at age 16, at a time when the key developmental challenges are forming intimate relationships, handling commitment and its conflicts, establishing boundaries with peers, and coordinating same- and cross-gender relationships.265 The Minnesota group later reported that exposure to interpartner violence in the preschool years predicted later partner violence perpetration and victimisation for both male and female participants. Participants who initiated partner violence perpetration and/or victimisation between age 26 and age 32 had greater exposure to partner violence in the preschool years than participants who showed no or declining domestic violence over these years.266 Collins, LaFreniere, and Simpson argued that such findings confirm that ‘one must understand the trajectory of an individual’s relationship history to fully appreciate, situate, and comprehend his or her (p. 386) adult relationships. This view of how development and relationships continually intersect across the lifespan is one of the major and lasting legacies of Sroufe and Egeland’s work on the entire field of psychology.’267

Another finding illustrates the ongoing role of early experience as a ‘resource’ to later adaptation.268 A clinical interview was conducted when participants in the Minnesota study were aged 28, assessing ‘global adaptive functioning’. This scale was based on the clinician’s impression of the individual’s overall psychological, social, and occupational functioning, and scored on a continuous scale. This measure is not just an assessment of mental illness or difficulties, then, but also positive adaptation and thriving. Infant Strange Situation attachment classifications had a very substantial association with this measure of adaptation and thriving several decades later (r = .41). Path analysis revealed that this association was partly direct (r =.26), even taking into account a large variety of mediators. Infant attachment appeared to remain, even after so many years, directly entangled in the coping and thriving of the Minnesota study participants at the most general level. The remainder of the association between Strange Situation classification and global adaptive functioning was mediated through a complex path including peer relationships at age 16, effectiveness in handling issues in romantic relationships in early adulthood, and total experience of life stress. Englund and colleagues within the Minnesota group characterised infant attachment as a species of ‘social capital’, and cumulative with other forms of social capital such as supportive friendships and romantic relationships in contributing to psychological, social, and occupational functioning.

Developmental pathways

The journal Development and Psychopathology was established in 1989 by Dante Cicchetti. The initial idea for the journal had come to Cicchetti during his years teaching seminars on developmental psychopathology for Sroufe’s classes, and then coteaching with Sroufe in his final year of graduate school.269 Sroufe’s articles for Development and Psychopathology in the 1990s are magisterial and programmatic. In particular, they set out his particular problems with medical models of diagnosis. The ‘essence of developmental psychopathology’, Sroufe argued, was attention to developmental processes that span both mental health and illness.270 He agreed with Egeland that developmental psychopathology needed to have an operational definition of targets of inquiry and intervention, like child abuse or conduct disorders.271 A collectively agreed definition would permit convergent validity in measurement, stability in theory-building, and clarity of message in communicating with policy-makers. However, (p. 387) Sroufe argued that too much research to date had assumed that medical diagnoses could serve to define these targets of interest:

The disease model requires syndromic integrity. If the disease model is apt for children’s behavioral and emotional problems, children generally should manifest tight clusters of symptoms, with unique indicators of other syndromes being absent. But in reality children commonly manifest problems that cut across established categories. To be sure, one disorder may potentiate another in medicine as well, but not nearly to the extent implied by the prevalence of co-morbidity of childhood disturbances … Comorbidity is the rule, not the exception. Moreover, broad classes of problems such as externalizing behaviors are predictive of a myriad of later conditions, including depression.272

Sroufe, Carlson, Levy, and Egeland argued that these concerns should be approached through the lens of Bowlby’s metaphor of developmental pathways, which implies a series of specific hypotheses:

(a) At any age, current quality of care will add to early attachment history in predicting pathology, given that adaptation is always the joint product of current circumstances and early history. (b) Likewise, broader aspects of current contexts, including relationships outside of the family and stresses and challenges of the period, also will increase prediction beyond early attachment; (c) a cumulative history of maladaptation will be more pathogenic than a single early period of poor functioning with pathology ever more likely the longer a maladaptive pathway has been followed, and (d) change itself will be predictable in light of changes in stress and/or support.273

This approach led Sroufe and colleagues to highlighted three ways that a developmental psychopathology perspective was a necessary counterweight to the disease model that dominates medicine.

  1. 1. Sroufe and colleagues argued that a diagnosis-focused science would end up missing broad causal factors that cross-cut and precede mental health and illness. Though ignored by a focus on diagnostic categories, attention to subclinical symptoms is critically important, both for understanding how clinical conditions develop and for understanding how mental health can be maintained. In this regard, Sroufe and colleagues claimed that family relationships are not a risk factor like any other, but play a fundamental role in shaping patterns of emotion regulation. This means that family relationships are especially important for the initiation and prolongation of most mental health symptoms.274 For instance, Carlson, Jacobvitz, and Sroufe found that relationship stability and social support for the mothers in (p. 388) their sample emerged as a powerful predictor of ADHD symptoms in their child participants.275

  2. 2. Second, they highlighted that the same diagnosis could reflect various diverse developmental pathways, with varying implications for treatment. ADHD was a clear example: Sroufe and colleagues felt that problems around attention and the integration of affect, cognition, and behaviour were best regarded as common signals of a developmental pathway rocked by adversities, rather than primarily a diagnostic entity in themselves.276 Supporting the idea that the same diagnosis could reflect different pathways, Aguilar, Sroufe, Egeland, and Carlson found two quite distinct routes to conduct problems and aggression in adolescence: one based on factors beginning in early childhood and leading to early-onset and persistent symptoms, another beginning in adolescence and associated with concurrent life stress and internalising symptoms.277 Sroufe and colleagues expressed concern that a diagnosis called ‘attachment disorder’, applicable only to a tiny proportion of children who had not formed a stable selective attachment, would magnetise attention away from the broader relevance of patterns of attachment to developmental pathways.278

  3. 3. Third, Sroufe and colleagues suggested that developmental pathways could result in various diagnoses. A pathway beginning from an avoidant attachment relationship in infancy could help predispose various forms of mental illness, depending on what then ensued in that child’s life and what other coping tactics were elaborated.279 Avoidant attachment suggests that key matters of concern cannot be brought to the attention of the caregiver, that the child must suppress or redirect distress, and that they may receive less support from caregivers in handling difficult feelings when they do unavoidably arise. Whilst not a clinical pathology in themselves, these aspects of avoidant attachment might contribute to the initiation of mental health symptoms and/or their prolongation. Clinical interventions and support focused on particular diagnosis could well, Sroufe and colleagues argued, miss the underlying developmental trajectory and hence the causal processes in play. This would reduce the effectiveness of the clinical work.280 In an important paper from 2005, Appleyard, Egeland, Dulmen, and Sroufe examined the impact of six risk factors in childhood (child maltreatment, interparental violence, family disruption, low socioeconomic status, and high parental stress) on symptoms relating to conduct problems and aggression in adolescence. The study showed that the number of risks in early childhood predicted the extent (p. 389) of symptoms in adolescence in a linear fashion. The authors therefore concluded that ‘the experience of an additional risk factor does not increase the odds of problems in a multiplicative way. For interventionists, such information might imply that there does not appear to be a “point of no return” beyond which services for children are hopeless, and that every risk factor we can reduce matters.’ The implication was that ‘interventions should be designed as comprehensive programs that enhance as many aspects of family life as possible’.281

The approach to developmental psychopathology of Sroufe, Egeland, and colleagues is well illustrated by their study of dissociative symptoms, discussed earlier. It was characteristic that the starting point of the Minnesota group was inquiry into dissociative processes in a population-based perspective. Rather than contrasting a group of patients with a diagnosis of a dissociative condition with healthy controls, Sroufe and colleagues explored the factors that can contribute to clinically significant and subclinical levels of dissociative symptoms. They leveraged the collection of data at multiple time-points, from multiple informants, to provide a rich picture of developmental pathways. The factors explored ranged from what may be individual-level factors such as infant distractibility, to relational factors such as infant attachment or present-day physical abuse, to the mother’s own childhood experiences—which her child might not even know. The researchers found that relational factors, more than individual factors, were important for predicting dissociative symptoms, which supported the researchers’ broader conceptualisation of risk and resilience.

The report on the antecedents of dissociative symptoms, however, also illustrates a limitation of the approach adopted by Sroufe and colleagues. Their analysis was solely within-sample. Without justifying the exclusion, they did not include structural factors such as socioeconomic disadvantage as a possible predictor within their analyses of dissociation. Socioeconomic disadvantage was characteristic of their whole sample as a consequence of the original recruitment strategy, which reduced relevant variance. Nonetheless, Sroufe and colleagues included socioeconomic disadvantage as an independent variable in other analyses, such as the sequelae of abuse and antecedents of ADHD, and found that it contributed to prediction.282 It was not as if the Minnesota group were uninterested in socioeconomic disadvantage, even if they regarded it as a distal risk factor and always mediated by proximal factors such as family behaviours and social support.283 As Egeland observed, ‘even in (p. 390) our poverty sample, socioeconomic status was a confound’.284 Many of the findings from the the Minnesota Longitudinal Study of Risk and Adaptation have replicated with low-risk samples.285 However, Haltigan and Roisman found that infant avoidant attachment, but not disorganisation, was associated with later dissociative symptoms in the NICHD study. In cases such as this, where findings differ, the exclusion of socioeconomic factors as a potential confound by the Minnesota group has contributed to difficulty in interpreting results.286

Resilience

‘Resilience’ was one of the headline concepts from the emergence of developmental psychopathology as a movement within developmental science. Colleagues of Sroufe and Egeland at Minnesota, such as Ann Masten, Norm Garmezy, and Auke Tellegan, played especially important roles in fostering this discourse.287 Some members of the developmental psychopathology movement were interested in examining individual-level factors that could lead research participants to do better than expected in response to adversity.288 This is the concept of resilience that would generally get absorbed by cognate disciplines, and by public and policy discourses.289 Sroufe and Egeland regretted this interpretation of the concept. In what became the dominant interpretation within developmental psychopathology over time, they took an opposing stance, emphasising that resilience is a relational process.290 They demonstrated that many of the studies that have shown links between temperamental robustness and positive outcomes in the face of adversity have had significant methodological flaws. For instance, a key work in the characterisation of resilience as a form of individual ‘invulnerability’ was Werner and Smith’s cohort study of 505 men and women who were born in 1955 (p. 391) on the Hawaiian island of Kauai.291 Sroufe and Egeland noted, however, that only one temperamental variable in childhood was linked to positive outcomes, and this was the mother’s description of her child as ‘loveable’ at two years old. They expressed doubt that this can confidently be regarded as reflecting inherent child variation.292

The term ‘resiliency’ implies an individual trait, and was therefore regarded by the Minnesota group as unacceptably misleading. But the term ‘resilience’ may more readily imply a process, even if often it has been used to refer to a trait. In 2003, Yates, Egeland, and Sroufe set out their definition of ‘resilience’ as a description of the capacity of children to garner resources that allow them to negotiate the present effectively and set up internal and external resources for future challenges.293 The emphasis on both internal and external resources provided a place for individual-level factors. However, the placement of these factors in developmental perspective suggested that it is ‘as supports accumulate’ that an individual develops the ‘acquired attitudes, expectations, and capacities to marshal resources [to] enable them to cope better with the additional challenge they may face’.294

The characteristics usually ascribed to ‘resilient individuals’, such as a calm temperament or optimism, were firmly regarded by Yates, Egeland, and Sroufe as developmental and social outcomes, mediated through the inner calibration of behavioural systems and by ongoing family and peer relationships. This made their position a profoundly destigmatising one, since it emphasised that mental health symptoms are an ordinary and expectable response to the accumulation of adversities, especially through childhood and adolescence, in the context of inadequate support. They were also early critics of a unidimensional concept of resilience, and urged recognition that positive outcomes despite adversities in one domain of life may in fact corrode opportunities in another domain that requires quite a different set of resources and capacities to the first. The paradigmatic case for Yates, Egeland, and Sroufe was avoidant attachment, which was conceptualised as an intelligible and likely helpful response to a less-sensitive caregiving context, but as making later emotion regulation and peer relationships more difficult in certain regards.295 Yates and colleagues suggested that keeping our worries and desires to ourselves reduces our exposure to hurt and disappointment, but if we lack the capacity to do otherwise then this saps something from us that might have otherwise been used elsewhere in building a richer and more stable life, as we lose heart in the prospects of attaining or sharing anything more than what we can muster alone. This then may fuse the avoidance into place, as we miss the opportunities that may arise for recognition and response to our worries and desires by others.296

(p. 392) Yates, Egeland, and Sroufe illustrated their argument by pointing to the developmental trajectories of children who displayed substantial behavioural problems at school. Causadias, Salvatore, and Sroufe found that, of children showing conduct problems, those who had experienced the most favourable early care showed the fewest mental health problems and had more supportive friendships in adolescence.297 Yates, Egeland, and Sroufe argued that a positive foundation had served as a resource for meeting the challenges of adolescence. The early foundation may have proven a ‘promotive factor’, enhancing the probability of later adaptation, not merely buffering the effects of contemporary adversity. Such findings were taken as showing that resilience is a description of the availability and utilisation of resources, rather than an inherent quality of individuals: ‘Consider the alternative interpretation that an early history of positive adaptation reflects an underlying individual trait called resilience. In this view, some of these children were resilient, then were not, and then were again.’298 Had the longitudinal study begun in preschool, the fact that some of the children had better outcomes than others would have seemed to be an effect of their individual characteristics; viewed in developmental context, these differences in fact were predictable from their experiences of early care.299

Perhaps the single most famous finding regarding resilience of the Minnesota Longitudinal Study of Risk and Adaptation was the identification of factors that could ‘break the cycle of abuse’, the transmission of abusive caregiving from one generation to the next. When parents who were abused as children were asked whether they would raise their child differently, all but four said yes, vowing that they would not repeat their own maltreatment. Yet based on interviews with the mothers in their sample, and multimethod assessments of child wellbeing and safety, the Minnesota group found that some mothers were not able to fulfil these vows. They found 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 but overall a minority of cases. There was a substantial gap in intergenerational transmission, a finding that has subsequently been well replicated.300 Egeland, Jacobvitz, and Sroufe reported that depression was exceptionally prevalent among the abusing mothers, and they ‘tended to view their children in an entirely negative or positive light and failed to recognize the ambivalence, which, of necessity, accompanies child care’.301

(p. 393) Egeland, Jacobvitz, and Sroufe 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. Egeland and colleagues observed that what these three experiences have in common is that they are experiences in which a secure base and safe haven has been provided, a ‘basis for developing alternative or transformed models of relationship’.302 One of the case studies they provide is of a mother ‘VF’.303 VF grew up in a family of ten children, all of whom ended up in foster care. Her mother would regularly beat the children, and as an adult VF still had scars from the physical abuse. Scared of being taken away from her mother like her older siblings, VF lied repeatedly at school to account for her bruises and cuts. However, when she was 11 years old, her mother asked child protection services to take VF. At the point that she left her mother’s care she had never been to a doctor or dentist—and did not know what these were. She spent two years in a first foster home, then two more foster homes. She then spent two years in a centre for treatment of substance dependency. During this time, she had psychotherapy. At age 18 she moved into an apartment with another girl, and very shortly after that became pregnant. When her son was a year old, she entered into a long-term relationship and eventually married. She and her husband both found work and had three more children together. Egeland and colleagues reported that ‘VF seems particularly aware of her past’, and was able to use this awareness to consider how it affects her caregiving. VF regarded her period of therapy and the support of her husband as important for helping her care for her children.

The researchers were powerfully struck that certain social experiences helped some mothers interrupt the continuities in abusive caregiving between generations.304 However, a later report by Enlow, Englund, and Egeland found that maternal experiences of child maltreatment predicted behavioural problems in their children at age seven, even if the mothers were not maltreating. This pathway was mediated by social isolation and negative life events such as financial difficulties, bereavements, and illness. The researchers concluded that ‘mothers able to “break the intergenerational cycle of maltreatment” may still be at risk for creating a negative caregiving context through increased stress exposures and reduced social support, part of the developmental sequelae of their maltreatment histories’.305 In a review chapter, Egeland and colleagues drew attention to research from other laboratories showing that, when abused parents do abuse their children, the kinds of abuse can be quite dissimilar.306 (p. 394) For instance, Thompson and colleagues found that mothers who had experienced sexual or physical violence as children were substantially more likely to have a child subject to inquiry from child protective services. Yet 75% of these referrals were for neglect.307 The movement of abuse between the generations appears not to be mimetic, as Bowlby had assumed (Chapter 1).

The Minnesota group emphasised that resilience is not a singular entity. Individuals can do well in one area of their life, but less well in others, as a result of the internal and external resources available to them—as well as the cut and colour of the challenges they face. The availability of a secure base as a form of social support helped promote non-abusive caregiving behaviours in mothers in their sample who had themselves experienced abuse. However, the early life stress these mothers experienced in receiving abuse nonetheless had implications for the mental health of their children. Sroufe and Egeland emphasised the importance of both life stress and social support in developmental perspective. When individuals responded to adversity with more positive outcomes, Sroufe and colleagues claimed that in regression analyses—not reported in print—the combination of life stress at different points and social support at different points could account for 80% of variance in outcomes: ‘Thus, little is left to mystery with adequate developmental data. Our research has found, time and time again, that resilience is not simply a function of good outcomes despite bad experience, but also an example of prior good experience facilitating the mobilisation of resources.’308 However, the analyses that lead to the 80% figure are not described in the text or in other published work.

Gender

In ‘The domain of developmental psychopathology’, Sroufe and Rutter proposed that gender is an important category to investigate.309 Both researchers found in their longitudinal studies that after puberty males facing risk factors were more likely to display aggression, conduct problems, and other ‘externalising’ forms of mental health problems, whereas girls were more likely to display depression, anxiety, and other ‘internalising’ symptoms. Sroufe and Rutter acknowledged that biological factors play a role, but overall their emphasis fell on social and cultural causes of this difference. This conclusion was later reinforced by Bakermans-Kranenburg and colleagues, who found no such gender differences in the distribution or trajectory of depressive symptoms in a Chinese sample, in contrast to European and American samples.310 Sroufe and Rutter pointed to the role of powerful cultural norms (p. 395) around gender that shape the expression of emotions, the activation of the caregiving system, the modulation of closeness and intimacy, and acceptable bases for self-esteem. They give particular emphasis to gender socialisation, since ‘girls in our culture are socialized toward compliance, inhibition, passivity, and reliance on others’. Depression and anxiety are congruent with these forms of socialisation in European and American societies. Sroufe and Rutter also refer to the experience of girls of sexual vulnerability, harassment, and the threat of rape.311

In the Minnesota Longitudinal Study of Risk and Adaptation, Pianta, Sroufe, Egeland, and colleagues observed aspects of gender socialisation. For instance, boys who were rated by observers and by their mothers as less socially skilled received more sensitive care at later follow-up, whereas this compensation by mothers was not provided to girls. In fact, girls who were more self-controlled received more sensitive care, which the researchers suspect may have been a form of reward and reinforcement.312 Gender also played a role in shaping the context of mental health symptoms. For instance, Pianta, Sroufe, and Egeland found profound effects of the kind of role model provided by mothers to their daughters, both as a potential risk and a potential protective factor for internalising symptoms. There was no such effect for boys.313

Sroufe and colleagues argued that ‘gender is a key aspect of the preschooler’s emerging self concept. Being a boy or a girl is central to the definition of the self. Development of a gender-based self-concept involves three steps. First, children gradually adopt sex-typed behaviour—actions that conform to cultural expectations regarding what is appropriate for boys or for girls. Second, children simultaneously acquire a gender-role concept—a beginning knowledge of the cultural stereotypes regarding males and females. Finally, children develop an emotional commitment to their particular gender.’314 Pursuing this reasoning, the Minnesota group conducted research to examine antecedents of the extent to which children regarded gender as a permanent characteristic.315 They asked preschool participants whether someone’s gender could change in different situations, such as if they put on different clothes. They found that maternal tendencies towards angry or depressive mood reduced girls’ perception of gender as a permanent characteristic. By contrast, mothers’ support and guidance for problem-solving and their responsiveness at home predicted greater gender concept stability for boys. There was no effect for attachment classification in the Strange Situation.

Indeed, it appeared to Sroufe and Egeland that gender and the attachment system were mostly unrelated phenomena. Given the importance of gender for so many aspects of life, the relatively modest role of gender differences in attachment research is (p. 396) curious.316 This was already remarked upon by Ainsworth. In correspondence with Bowlby, Ainsworth wrote:

I believe that the main reason why the issue of sex differences has been so neglected in regard to attachment is that such differences seem to be absent in infancy. Certainly, the three main attachment patterns (strategies) do not differ in incidence among males and females. That makes sense, because in infancy protection is just as important for males as for females, and the kind of protection that is needed is essentially the same for both sexes.317

Yet gender would similarly prove of relatively little relevance to the distribution of attachment classifications in middle childhood or adolescence, at least when observational measures were used.318 Whilst developmental psychopathology has generally been concerned with the interaction of risk and protective factors, there is no implication that all aspects of life will interact to the same extent. Gender and attachment were regarded by Sroufe and Egeland as both important aspects of a child’s experience and relevant to their developmental outcomes. In particular, both gender socialisation and attachment were regarded as contributing to the elaboration and calibration of the caregiving and sexual behavioural systems. However, they considered these contributions as largely independent, rather than in interaction—though this was not formally tested. After Sroufe and Egeland’s retirement, some small associations were found by Raby, Roisman, and colleagues on the Adult Attachment Interview (AAI) when this was coded dimensionally rather than categorically: dismissing is slightly more common in male participants (r = .17), and preoccupation in female participants (r = .22).319 However, such associations are not large enough to imply a need to substantially renegotiate Sroufe and Egeland’s position on gender and attachment.

Two case studies

Case descriptions are woven in elegantly to the reports of the Minnesota group, particularly in Sroufe’s early book chapters.320 However, it was rare for the group to describe in detail the result of the assessments that had taken place over the years with particular dyads. Two case studies, however, were reported in substantial depth at the end of the 1990s. One was the case (p. 397) of Beth. At 12 months, Beth and her mother received a classification of ambivalent/resistant attachment, and at 18 months a classification of secure attachment. Beth’s father was killed in a car crash, but at 24 months Beth and her mother both seemed to be doing well under the circumstances. Beth was able to cooperate effectively with her mother in the 24-month tool-use situation, and at 42 months was persistent and enthusiastic in attempting the difficult Barrier Box task.321 At the end of first grade, Beth’s mother had remarried. Her teacher described Beth as a bit of a loner, but doing quite well academically. In an interview during second grade, the interviewer noticed that Beth seemed afraid of her stepfather, who was regularly drunk. A year later, Beth and her mother moved into a women’s shelter. In the sixth grade follow-up, Beth and her mother were living with the stepfather again.

At the follow-up when participants were 17 years old, it took some time for the researchers to locate Beth. In interview, she described that her stepfather had been subjecting her to multiple forms of maltreatment since she was 12 years old. Child protection services inquiries had been deflected by her mother and father, who presented Beth herself as the problem and disengaged when this narrative was questioned. She had left home at 16, and her parents left the state in order to avoid a court order that the stepfather leave the home to protect the younger siblings. Beth first lived with a foster family, then a boyfriend, then her grandparents. She had remained in school throughout this period. The clinical interview at age 17 revealed no current psychiatric diagnosis. The interview documented depression in early adolescence, and a subclinical phobia of small spaces. She also reported one past suicide attempt. At age 19, Beth had been awarded a scholarship to attend college and had completed one semester. She had recently withdrawn due to problems with her boyfriend, including abuse. However, she planned to return to school, had a part-time job, and seemed to be doing very well in other areas of her life. She was in the process of seeking counselling.

A second detailed case study offered by the Minnesota group was that of Laura Miller (a pseudonym). In the 1990s, a furious controversy raged regarding whether adult ‘recovered memories’ of child sexual abuse were fabulations or truthful recollections of previously repressed experience.322 In 1998, the journal Development and Psychopathology dedicated a special issue to this concern, which highlighted the need for thinking about how memory processes following trauma might be drawn upon for coping and for other adaptive challenges.323 In line with this interest, in an article from the same year, Duggal and Sroufe reported the first ever prospective report in which sexual abuse was documented by multiple sources, and the young person subsequently reported no recall of the abuse for a time, and then ‘recovered’ the memory. Data for the case study were drawn from the longitudinal study, from a follow-up interview with Laura and her mother, from the records of a (p. 398) therapist who worked with the family after child protection services became involved when Laura was four years old, and from interviews with Laura’s child protection worker and her therapist.

A first sign of sexual trauma was documented in preschool. Laura was one of the children who was part of the preschool class made up of members of the sample. The head teacher of the preschool reported sexual behaviours shown by Laura in the preschool classroom. In the interview with mothers when their children were 54 months, Ms Miller (Laura’s mother) stated that Laura had described a dream in which her father, who had some limited custody, had come into her bed while she was sleeping and touched her, while she pleaded with him to stop. When confronted, Laura’s father denied sexually abusing Laura, but agreed to see Laura for only three hours per week and only with a third party present. Ms Miller asked child protection services if visitation could be prevented entirely, but physical examination for penetration was negative and by this time Laura would not talk any further about the incident, so it would have been difficult to evidence by the legal standards of the time. However, Laura was referred for therapy. The therapist notes from the period agree with Ms Miller’s account. They also state that Laura spontaneously announced to the therapist, a month after intake, that ‘Daddy knows he did it, he’s just scared’. The notes included record of Laura’s anxiety and anger towards her father.324 The father was interviewed by the therapist and said that he had no memory of sexually abusing his daughter, and that if it occurred it would have been while he was on drugs. At the end of first grade, Laura’s teacher related that Laura had told her that the reason her parents had divorced was that her father had abused her, but that he was not doing it any more. Duggal and Sroufe concluded that ‘Multiple sources of evidence discussed in this section consistently support the conclusion that Laura experienced trauma of a sexual nature during early childhood’.325 The last record in which Laura evidenced knowledge of abuse was at age eight, in the therapist notes.

In her engagements with mental health services in seventh and ninth grade, she did not mention childhood sexual abuse as a presenting issue. At age 16, participants in the study were asked to complete a broad-ranging written health interview. One of the items was the question: ‘Have you ever been sexually abused? Sexual abuse is when someone in your family or someone else touches you in a place you did not want to be touched, or does something to you sexually which they shouldn’t have done.’ Laura indicated that she had not been sexually abused. The interviewer noted that Laura appeared unguarded in her replies, and disclosed various other personal information with candour. At age 17, in a clinical interview to assess symptoms of post-traumatic stress disorder, Laura was asked whether she had ever had any terrible and unusual experiences. She reported, without hesitation, that she had experienced nothing of this kind. In response to a query regarding experiences of abuse, including sexual abuse, Laura replied that this had not happened to her. She was again candid in discussing other difficult past and present experiences, including contemporary mental health problems. Ms Miller recalled that during adolescence, Laura reported being happy to spend time with her father and wanted to see him more. Ms Miller did not bring up the issue of abuse or ask direct questions because she felt that she had been warned by child protection services to avoid ‘putting words in Laura’s mouth’.326

(p. 399) When Laura was 18, she took part in an AAI (Chapter 3), which includes a question about experiences of abuse. In response to this question, Laura described a recent conversation with her boyfriend about her first memories of her parents. This triggered a ‘really overwhelming feeling’, and led Laura to seek out a trusted teacher. During the conversation with the teacher, she reportedly recovered memories of her father kissing her while she was in bed, with a sense that there was a sexual component to the interaction:

I told her this first memory and then it just came to me. Like, it just was so clear in my mind, um, you know he like did this to me, whatever. And um, and I like went into complete shock … I just felt like I was really dizzy and, and I couldn’t really see very well and, and there was like all these noises going on, but I didn’t know what they were, and and it wa- I was really confused and, like I couldn’t cry but I felt like I should be.327

She added: ‘You know he bent, he was like walking over to me and he bent down and kissed me. And there’s something after that and I don’t know what it is.’328 Following the conversation with the teacher, Laura’s relationship with her father changed strikingly. She felt scared of him most of the time, interspersed with missing him terribly. Duggal and Sroufe argued that exclusion of memory of the event made sense if Laura was forced to see her father during her childhood. There was a logic to the memory resurfacing at age 18, since this was the threshold of Laura’s independence from her parents and the need for visitation. Duggal and Sroufe situated Laura’s case as illustrating the role of defensive processes as adaptive responses to situations that cannot be controlled or escaped. They conjectured that it seemed a good sign for Laura’s later development that she could release the exclusion of the painful memory at the point that she was no longer required to see her father.

There is no sense that the two published case studies are representative of the sample as a whole. They nonetheless offer a valuable window into the way that the multiple assessments of the Minnesota Longitudinal Study of Risk and Adaptation together offered the research group an encompassing picture of human lives, each hard-won and remarkable. The cases illustrate the approach to child development that framed Egeland and Sroufe’s work on the Minnesota study. The assessments conducted by the Minnesota group allowed lines of continuity and discontinuity in development to be examined, alongside exploration of the interdependence and lurching along of different social relationships. Comparison of the two cases illustrates the important role of changing adversities and social supports for the kinds of resilience they could muster. For instance, the cases of Beth and Laura illustrate disinhibiting parental drug and alcohol use as a risk factor for the children, which compounded with other adversities and forms of unforgiving isolation they faced.

Both case studies were reported in the late 1990s. There have subsequently been several further waves of data collection. The field has also seen a ‘changing of the guards’ with the retirement of Sroufe and Egeland. Such a significant developmental transition has come with new challenges, and, as could be anticipated, has required the elaboration and change in the field’s inner organisation. One angle on the lines of developmental continuity between past and present is to consider one of Sroufe and Egeland’s former students, Dante Cicchetti, who now holds Sroufe’s William Harris Professorship in Child Development at the University of (p. 400) Minnesota. Cicchetti did not primarily regard himself as an attachment researcher. However, the approach of Sroufe and Egeland and the findings of the Minnesota Longitudinal Study of Risk and Adaptation have been a central component of his influential approach to developmental psychology. Consideration of Cicchetti’s work offers an illustration of one way that the legacy of the study of attachment at Minnesota has been inherited by subsequent researchers.

Two former students

Dante Cicchetti

Part of Mary Ainsworth’s achievement was to have translated a new and exciting methodological innovation into a new generation of attachment researchers in the 1970s and 1980s, including many important future leaders (Chapter 2). The other primary hub of attachment research during the period was the Minnesota group, which was also critical for mentoring a second generation of attachment researchers. Sroufe commented that his proudest achievement has been the contribution made by his former students to the field.329 One of Sroufe’s last doctoral students was Glenn Roisman, whose work was discussed in Chapter 3. At the other end of his career, one of his first doctoral students was Dante Cicchetti. Cicchetti had a drive to examine experiences of risk and improve outcomes for families rooted in his own childhood. As he later recalled in interview, ‘much of my research has been influenced by my own experiences. Early encounters with poverty and harsh conditions played a major role in fuelling my research interest in child maltreatment.’330 He later dedicated his Award from the American Psychological Association for Distinguished Senior Career Contribution to Psychology in the Public Interest to his younger sister ‘Candace, who died in the fall of 2000. She represents one of all too many individuals who unfortunately lose battles due to adverse circumstances and mental illness.’331 Cicchetti has also written about his own struggles around mental health.332

Cicchetti’s doctoral thesis examined the affective and cognitive development of infants with Down syndrome.333 Cicchetti badged his work as an application of the organisational perspective, announced in the Sroufe and Waters 1977 paper ‘Attachment as an organizational construct’ (Chapter 2).334 Over four decades there is an explanation of this (p. 401) perspective in nearly every first-author article and book chapter he has written. Following Sroufe, Cicchetti emphasised that individual behaviours should not be the unit of analysis in the study of child development. Instead, researchers should focus on the integration of resources in responding to the most important challenges from the environment relevant to a person at that time in their lives.335 Bowlby’s concept of probabilistic developmental pathways had just been introduced during Cicchetti’s time as a graduate student, and this proved an important influence on his exploration of the interaction of factors in both risk and resilience, and the reciprocal interpretation of typical and atypical development in light of one another.336 As a graduate student at Minnesota, Cicchetti co-authored a review paper with Egeland on antecedents of child maltreatment, with particular attention paid to the role of poverty, family conflict, and alcohol and substance use.337 He documented that parents who maltreat their children show no single set of personality traits or fit into specific psychiatric diagnostic categories. Instead there are a wide variety of factors that together contribute to the likelihood of maltreatment, as well as other potential difficulties for children and families. The fact that a diversity of pathways could lead to the same outcome would be dubbed by Cicchetti as ‘equifinality’; the fact that a single pathway could contribute to diverse outcomes would be dubbed ‘multifinality’.338

After leaving Minnesota in 1977, his central focus in the 1980s was basic research on child maltreatment. Cicchetti took up a faculty position at Harvard, where he won a grant for over $2 million from the National Center on Child Abuse and Neglect for a study of the etiology, intergenerational transmission, and developmental sequelae of children. Funding for follow-up research with the sample was provided by the National Institute of Mental Health and the William T. Grant Foundation. He became Director of the Mt. Hope Family Center at the University of Rochester in 1985. During the 1980s, Cicchetti played a pivotal role in the initiation of developmental psychopathology as a movement within developmental science. He sustained a punishing schedule of productive work during the decade as the movement was established. One metric is his work as an editor: as well as initiating the journal Development and Psychopathology and the annual Rochester Symposium on Developmental Psychopathology, Cicchetti co-edited 10 books through the 1980s. He also published 26 articles as first author.

By the end of the decade, the National Academy of Sciences had embraced developmental psychopathology as the overarching framework for its report Research on Children (p. 402) and Adolescents with Mental, Behavioural, and Developmental Disorders,339 and the National Institute of Mental Health had situated developmental psychopathology as the organising framework for future research funding priorities for child and adolescent mental health.340 In 1990, the eminent developmental researcher Robert Emde described the work of Cicchetti, Mary Main, and others as commencing the ‘third phase of attachment research’.341 If Bowlby’s initiation of attachment theory had represented the first phase, and the validation of the Strange Situation by Ainsworth, Sroufe, Egeland, and others in the late 1970s and 1980s had represented the second phrase, the third phase was characterised by innovations in methodology to examine attachment beyond infancy and new insights into risk and protective factors for mental illness.

Yet the continuities between the work of Cicchetti and Sroufe were strong and fundamental. Cicchetti picked up from Sroufe and colleagues an abiding concern for the implications of early attachment relationships, even if this concern would be integrated within a broader frame. For instance, he argued for greater attention to the implications of cultural differences in constituting and shaping risk and promotive factors, and individual and group responses to challenges.342 However, the prime extension of Sroufe’s model was the inclusion of biological factors. Cicchetti was led to a concern with biological differences from the start of his career in making sense of the experiences of families with an infant with Down syndrome. However, he was also part of a generation of researchers with much more ready access to a diversifying array of biomedical measures. Cicchetti argued that the account of resilience provided by Sroufe and colleagues could be expanded and enhanced through attention to biological factors, such as neurological, endocrinal, genetic, and immunological processes, as sources of stability and change in developmental pathways.343

Behavioural systems such as attachment were reconceived as having affective, cognitive, behavioural, cultural, and neuroendocrinal components. Cicchetti therefore argued that assessments of maltreated children by child protective services as well as researchers should include both biological and psychological measures. However, for Cicchetti, this was only as a more fiercely realised version, not different in kind, from Sroufe’s stance that the development of the ‘whole person’ must be kept in view.344 It was also an extension of Sroufe’s focus on the emotional components of behavioural systems. Cicchetti regarded emotions as having four components: non-verbal components, verbal components, experiential components, (p. 403) and biological components.345 For instance Cicchetti and colleagues utilised the Egeland and Sroufe methodology of inviting their child participants to take part in summer camps in order to see how individuals form and respond to the developmental challenge of peer relationships. However, unlike Egeland and Sroufe, Cicchetti and colleagues interpreted the behaviour of the individuals and the interactions of the groups in light of neuroendocrinal data, as well as self-report and observation.346

Egeland and Sroufe did not have available such neurological and hormonal measures and did not give biological processes a specific place in their thinking about behavioural systems.347 However, to some extent both researchers were attentive to biological factors. For instance, Sroufe and Waters pioneered the use of physiological measures of cardiac response to show the distress of the apparently unruffled infants in avoidant dyads. And there is every reason to assume that if other physiological measures of sympathetic nervous system response had been available, for instance saliva cortisol, Sroufe and Waters would have measured these too.348 Nonetheless, of the two principal investigators, Egeland appears to have been the more sympathetic to biological assessments. He drew on assessments of puberty development in elaborating structural equation models of the antecedents of teenage sexual behaviour and alcohol use. Attentive to the multicausal processes involved in puberty timing, Egeland and colleagues regarded physical maturity as a ‘combined effect of hormones and social interactions’.349 He also offered speculations on the neurophysiological processes that provide the infrastructure for avoidant and resistant attachment.350 Cicchetti and Egeland have subsequently worked together on a study of contribution of genotype to insecure attachment in the Minnesota sample. The researchers found that variance in the serotonin-transporter-linked polymorphic region (5-HTTLPR) had no association with whether a child was classified as secure or insecure. However, for those classified as insecure, it contributed to the prediction of an ambivalent/resistant or avoidant attachment classification, through children’s emotional reactivity.351 5-HTTLPR also predicted the subtype of secure attachment, contributing to the more distressed (B3 and B4) or less distressed (B1 and B2) form of security. The effect seemed to be independent of the contribution made to (p. 404) attachment classification by caregiver sensitivity: there was no interaction effect between the variables. On the basis of such findings, Raby, Cicchetti, Egeland, and colleagues have called for ‘reconciliation’ between Sroufe’s focus on attachment and Kagan’s focus on temperament, arguing that both can make a contribution.352 However, it should be noted that Raby’s later work on the NICHD sample has not offered support for the role of molecular genetic factors in predicting attachment classifications.353

In agreement with Sroufe and Egeland, Cicchetti viewed it as an important intervention against the stigma attached to mental illness to view symptoms as ordinary and statistically expectable responses to a dynamic accumulation of adversities, and relative lack of supports, in the course of development.354 Whilst less adverse to use of diagnostic categories in assessments than Sroufe, Cicchetti has been mindful of the sociological processes that led to the dominant role of diagnoses within psychology, and in his own work on children and adolescents has focused much more on cross-diagnostic developmental and relational processes.355 Cicchetti and Toth have been extensively involved, for example, in studies that measure traumatic experiences, but have expressed doubts about the criteria used to diagnose trauma in clinical practice, for instance the need for a precipitating single event.356 They have also expressed concern regarding the varied connotations of the term ‘trauma’, which can lead to miscommunication between researchers, for instance in discussion of the implications of trauma across developmental periods.

The decision to not intervene with the families in the Minnesota study, even the maltreating families not known to child protective services, in their study, but to watch and try to understand was one that weighed heavily on Sroufe and colleagues.357 They felt that their participants had signed up for a cohort study, not an intervention study, and that it was also important to understand the naturalistic antecedents and implications of child maltreatment in order to inform future intervention efforts. By contrast, Egeland was actively involved in the creation of the STEEP intervention (which is discussed in the section ‘Holism in interventions’). Like Egeland, Cicchetti was concerned with both epidemiology and clinical intervention, a position he and Toth described as ‘Bowlby’s dream come full circle’ in contributing to clinical practice.358 However, Cicchetti and Toth argued that Bowlby’s attention to psychological unavailability must be elaborated into a more multicausal model, capable of acknowledging factors such as economic, medical and housing difficulties, without (p. 405) collapsing levels of analysis.359 Cicchetti and Toth also supplemented Bowlby’s approach with Sroufe’s more fine-grained attention to different developmental challenges across the lifecourse. For instance, Cicchetti and Toth argued that it follows from an organisational perspective that even when a child has been supported to achieve good outcomes following maltreatment, developmental reorganisations—such as the transition to parenthood—could result in vulnerabilities. The reorganisation may well draw upon earlier affective, cognitive, behavioural, and neuroendocrinal components that had previously not been much in evidence.360 On the other hand, Cicchetti and Toth emphasised that developmental transitions such as adolescence, and their ensuing shifts in challenges and relevant resources, can also represent opportunities for reorganisation away from pathology, especially if there are resources available on the basis of past experience or present support.361

Additionally, intervention research offered the prospect of experimental rather than naturalistic evidence for theoretical hypotheses. Intervention studies allow mechanisms of change and stability for families to be specified, and causation more readily distinguished from correlation.362 Cicchetti, Toth, and colleagues were early adopters of the use of randomised control trials to evaluate interventions with families and the use of attachment measures as intervention outcomes.363 Even as attachment theory has seen a decline in authority and acceptance within the developmental science community, Cicchetti has remained supportive of research in this area. In his role as a prominent and influential editor, he has allocated space to attachment theory as a key paradigm within developmental psychopathology. From his perspective, the preferential attachment relationships with a caregiver or caregivers established in infancy ‘provide a context for children’s emerging bio-behavioral organization’, influencing the integration of ‘neurobiological, cognitive, affective, and behavioral capacities that influence ongoing and future relationships, as well as the understanding of self’.364 Stating their agreement with Sroufe, Egeland and colleagues, Cicchetti, and Boyle have argued that ‘variations in early attachment are not considered to be pathology, or as directly causing pathology. However, they do lay the foundation for disturbances in developmental processes which can eventuate in psychopathology’.365 For instance, the researchers (p. 406) found that secure attachment at 26 months fully mediated the association between a psychotherapy intervention for parents suffering from depression and their child’s display of conduct problems at school eight years later.366

Cicchetti and colleagues were the first researchers to use the Main and Solomon disorganised attachment classification in a sample of maltreated children.367 The Harvard Child Maltreatment Project was initiated in 1979 by Cicchetti and Aber, as a large longitudinal study, modelled on the Minnesota Longitudinal Study of Risk and Adaptation.368 A small subsample were seen in the Strange Situation and classified using the Ainsworth categories.369 Twenty-two dyads were known to child protection services for abuse and/or neglect of the child or an older sibling; 21 were a matched comparison group. However, a third of the sample were classified as securely attached, which was an unexpected and concerning finding. Like Egeland and Sroufe, the infants in these dyads displayed disrupted behaviour that seemed to interrupt or contradict their use of the caregiver as a secure base. There were a few different proposals available for additions to Ainsworth’s system. One was that of Ainsworth’s student Patricia Crittenden, who observed that many abused infants show both avoidant and resistant attachment (A–C).370 Another was the Main and Solomon disorganised attachment classification, which appealed to Cicchetti and colleagues as more encompassing; it was assumed that finer-grained distinctions could be worked out later. The disorganised classification also had the advantage of evidence from Main’s laboratory of relevant correlations with interviews with the caregivers and with the child’s later behaviour, offering a source of predictive validity.371 The analysis used a coding of the data conducted by Cicchetti and Barnett; Barnett was blind to the children’s maltreatment status: 82% of the infant–caregiver dyads known to child protection services were classified as showing disorganised attachment and 14% classified as secure; in the comparison group 19% were classified as disorganised and 52% classified as secure. At 18 months, 61% of dyads known to child protection services were classified as showing disorganised attachment compared to 29% in the comparison group.372

(p. 407) In a later study, Cicchetti, Rogosch, and Toth used the Strange Situation as part of the evaluation of two different forms of intervention with 137 families known to child protection services for infant abuse and/or neglect.373 They also included 52 matched families as a non-maltreating comparison sample. The maltreating families were allocated to one of three conditions. In the first condition, they received usual social work support. In the second condition, they received an intervention that focused on teaching parents about infants’ needs. In the third condition, the mother–child dyad received once-a-week psychotherapy, using the model of child–parent psychotherapy developed by Selma Fraiberg374 and Alicia Lieberman (a former student of Ainsworth).375 In child–parent psychotherapy, the intervention focuses on exploring the caregiver’s personal history of adversities and difficult feelings, especially as evoked in the context of the present-day relationship with their child. Therapeutic interactions are responsive to spontaneous child–parent interactions and play, and aim to provide a safe base for the caregiver in providing a safe base for the child. As in their earlier study, Cicchetti and colleagues found high rates of disorganised attachment in the maltreatment group (89%, compared to 42% in the non-maltreating comparison dyads). At 26 months, the Strange Situation was repeated. For the maltreating families who had seen social work intervention as usual, 78% received a disorganised attachment classification and the rest of the dyads were classified as avoidantly attached. In the parent education condition, 45% received a disorganised attachment classification and 54% were securely attached. In the child–parent psychotherapy condition, 32% of dyads received a disorganised attachment classification and 61% a secure classification. To the researchers’ surprise, the change was not mediated by measures of adult representations of their child, caregiver sensitivity, parenting attitudes, child-rearing stress, or social support. Cicchetti and colleagues described that the results of the trial were ‘gratifying and sobering’:

The fact that plasticity is possible during infancy and that even the most disorganized form of attachment is modifiable in extremely dysfunctional mother–child dyads offers significant hope for thousands of young children and for their families. By fostering secure attachment, costlier interventions such as foster care placement, special education services, residential treatment, and incarceration can be averted. Unfortunately, our results also shed light on the harsh reality of the ineffectiveness of services currently being provided.376

More recently, Cicchetti, Rogosch, and Toth reported results from a study of genotypic variation in their sample, examining the 5-HTTLPR (serotonin-related) and DRD4 (dopamine-related) polymorphisms.377 This followed up a hypothesis Cicchetti had put forward in the (p. 408) 1980s that dyads would be most likely to receive a disorganised attachment classification if the infant was both abused and had a genetic predisposition towards affective reactivity.378 The hypothesis was not precisely supported. The researchers found that among the non-maltreated infants, 5-HTTLPR and DRD4 polymorphisms influenced attachment security and disorganisation at 26 months and the stability of attachment disorganisation between 12 and 26 months. However, among the maltreated infants, they found no such associations between genotype and attachment classifications. The researchers concluded that the high rate of disorganisation ‘overpowered the genetic contribution’ to attachment patterns among the maltreated infants.379 Genetic variation between infants was not found to have any implications for the effectiveness of the interventions.

In a later follow-up of the sample 12 months after the end of treatment, dyads who had received psychotherapy had rates of secure (56%) and disorganised (26%) attachment that resembled the distribution in non-maltreating samples facing financial adversities. By contrast, the parent education intervention had high rates of disorganised attachment (59% disorganised; 23% secure), as did those who had received social work intervention as usual (49% disorganised; 12% secure). Cicchetti and colleagues proposed that the parent education intervention had supported the parents only in dealing with the needs of infants. It had not supported them in responding to past adversities and traumas that might disrupt their ongoing and unfolding relationship with their growing child.380 Aligned findings come from another study by Toth, Cicchetti, and colleagues who studied a group of maltreated preschoolers who received parent–child psychotherapy. The researchers found that the preschoolers’ beliefs about themselves and about their caregivers became more positive and less negative over the course of the intervention, in contrast to the parent education condition or the standard child protective services intervention.381

Glenn Roisman

As discussed in Chapter 3, another student of Sroufe’s later in his career was Glenn Roisman. Roisman is now one of the leaders of the Minnesota Longitudinal Study of Risk and Adaptation. Roisman has been fascinated by the discoveries of Main and colleagues. Yet on several occasions he has identified that terms in Main and Hesse’s writings were extrapolated as theory on the basis of demanding observation of small samples. Main’s most prominent innovations—the disorganised infant attachment classification and the AAI—were based on video and audio recording as new technologies of exactitude and repetition, combined with meticulous work with a limited number of cases. Roisman expressed concern that such use (p. 409) of small samples, among other factors, meant that little or no psychometric analysis was ever conducted by Main on her new assessments, to see if the categories are well adapted to capturing what they seek to measure. He suspected that the scales that Main and colleagues had created to help coders come to category decisions—such as the scales for coherence, idealisation of the parent, and preoccupying anger—might have psychometric properties as good, or perhaps better, than the overarching categories. Additionally, examination of the scales could permit taxometric inquiry into the architecture of individual differences in states of mind regarding attachment in adulthood, with the potential to contribute to theory and to improvements in how findings could be analysed.

Roisman was by no means the first to ask psychometric questions of measures such as the Strange Situation and the AAI. This was a major concern of Waters (Chapter 2) and van IJzendoorn and colleagues from the 1980s onwards, including the subject of Marian Bakermans-Kranenburg’s doctoral thesis.382 However, Roisman has been distinctive in repeatedly raising questions about these measures and their theoretical categories specifically in terms of construct validity and psychometric precision. The first object of this criticism was, as we have seen, the ‘earned-secure’ classification (Chapter 3). However, Roisman’s questions also raise the wider issue of whether Main’s semi-inductive approach to measure design might have a systematic drawback. On the one hand, her approach may contribute to creative discoveries of previously unrecognised associations. On the other hand, she may inadvertently mislead later researchers by creating named categories and then interpreting the association in light of that name rather than critically pursuing the mechanism behind the pattern. The ‘earned-secure’ classification was discovered when some parents in the Berkeley sample were able to speak coherently as they reported very difficult childhoods. However, Roisman argued that scientific pursuit of the meaning of the phenomenon has been overly shaped by a commitment to the theory implicit in the label given by Main and Goldwyn on the basis of the reports of a small number of subjects. The theory and methods developed by Main and colleagues has rested at the foundation of so much activity over decades in the developmental attachment tradition that criticisms have often been treated as a general attack on the paradigm in general. Were it not for the consecration provided by Sroufe’s involvement as doctoral supervisor and co-author, it is likely that Roisman’s questions would have been treated as an unlicenced attack on Main and Hesse.383 However, the criticisms of the ‘earned-secure’ classification were made by Roisman and Sroufe together: not only was Sroufe’s standing within attachment research unimpeachable, but the claims were made on the basis of data drawn from the authoritative Minnesota Longitudinal Study of Risk and Adaptation, and then later the massive NICHD study.

Following his doctorate, Roisman took up a faculty position at the Department of Psychology at the University of Illinois in 2002. Chris Fraley had arrived at Illinois-Chicago in 2000, following a PhD at the University of California, Davis with Phil Shaver. He joined the University of Illinois at Urbana-Champaign in 2004, and it was there that Roisman and Fraley became close friends and collaborators. Shaver and Fraley had worked on the development and refinement of the Experiences in Close Relationships scale (Chapter 5). An (p. 410) overwhelming majority (80%) of the variance in this analysis was accounted for by two dimensions: anxiety and avoidance. Fraley was confident that attachment would be dimensionally rather than categorically distributed. If multiple factors could impact to varying degrees on attachment patterns, then for him a logical implication was that differences in attachment between individuals would be of degree, rather than of kind. At a statistical level, as well, Fraley felt that continuous models offer more effective prediction. In conducting correlations and regressions, dimensions allow all the variance to be used, rather than forcing arbitrary cut-offs at both ends and losing relevant variation.

Most attachment researchers now agree that attachment can be modelled dimensionally.384 However, a fundamental controversy lay in the proposal, put forward by Fraley and Speiker (Chapter 2), that proposed two dimensions—the Ainsworth avoidance and resistance scales—as the best way to capture differences in the Strange Situation.385 Fraley argued that since Ainsworth had already created scales for resistance and avoidance, the field had already been collecting the relevant data over decades to make possible a switch from categories to dimensions. This assumes that security is essentially the co-occurrence of low levels of avoidance and anxiety (Chapter 5), and that a separate disorganisation dimension is unnecessary. Waters and colleagues had earlier developed the Attachment Q-Sort (Chapter 2) which assessed secure base behaviour using a single dimension of security-insecurity. However, Waters avoided implying, at least in print, that a disorganised category or dimension was unnecessary.

The use of scales rather than categories seeped into common practice for reporting results from the Strange Situation among attachment researchers through the 2000s, but rarely with explicit justification or reference to Fraley and Spieker. Their paper was generally treated by the developmental tradition of attachment research as a marginal view. However, conversations between Roisman and Fraley continued to develop the position, mostly with the AAI as their target. Given its much smaller number of scales, the Strange Situation procedure offers less purchase than the AAI for psychometric analysis. The AAI may also have been the less controversial coding system to raise for discussion. As time went by, Roisman and Fraley continued to agree with Main that the infant attachment classifications represented the basic forms of individual difference in human emotional life, across the lifespan. However, they became increasingly certain that these differences were best captured by two latent factors. A first factor was dismissing states of mind regarding attachment. A second was a combination of preoccupied and unresolved states of mind regarding attachment.386

(p. 411) Applied to infancy, this implied a change from operationalising the second dimension as resistance, as in Fraley and Spieker. Roisman and Fraley combined resistance and disorganisation scores, since both were understood to reflect overt anxiety.387 This decision appears to have been based on Fraley’s conceptualisation of individual differences in attachment on the basis of the Experiences of Close Relationships scale. Both resistance and disorganisation could be conceptualised as ‘anxiety’ about the availability of the caregiver, one of the two latent dimensions of the Experiences of Close Relationships scale. No taxometric analysis of the Strange Situation or AAI has been conducted on a clinical sample.388 Yet in multiple non-clinical samples Fraley and Roisman demonstrated that variance on the AAI can be best explained by these two weakly correlated latent factors.389 Roisman’s appointment in 2012 as Sroufe and Egeland’s successor at Minnesota, with leadership responsibilities for the Minnesota Longitudinal Study of Risk and Adaptation, has contributed to his powerful position within contemporary attachment research. Nonetheless, the Roisman and Fraley criticisms of the four-category model have been highly controversial, since they have appeared to some to undermine the conceptual and symbolic foundation for the field of attachment research as established by the first and second generations.

Van IJzendoorn and Bakermans-Kranenburg argued that unresolved states of mind may fall with preoccupied states of mind in non-clinical samples, since in this context they have much in common. However, in the clinical range the U/d category may convey extra information, and not be reducible to the preoccupied dimension.390 Responding to these concerns, Raby, Roisman, and colleagues recently conducted taxometric studies of AAIs in two high-risk samples.391 One study was with 164 participants from the Minnesota Longitudinal Study of Risk and Adaptation who completed an AAI at age 26. Another study was with 284 participants living in poverty and known to child protective services; 147 parents who had pursued an international adoption of their child; and 300 foster parents. An exploratory factor analysis with the Minnesota study indicated that Main, Goldwyn, and Hesse’s scales loaded on two latent dimensions reflecting dismissing and preoccupied states of mind, and that scores for unresolved states of mind loaded on the same factor as scales related to (p. 412) preoccupied states of mind. A confirmatory factor analysis with the three samples in the second study supported the viability of the model in a second high-risk cohort.

There remain questions about the model of individual differences in attachment in terms of two latent dimensions. Raby, Roisman, and colleagues excluded the Main et al. coherence scale from their factor analysis because it had a very strong relationship with security and had previously been found to cross-load on the dismissing and preoccupation dimensions.392 Yet in fact recent discussions in the social psychological tradition have questioned the two-dimensional model and argued for security as a partially independent dimension (Chapter 5). Furthermore, it seems to be a premise of the work by Raby and colleagues that three or more scales are required for a construct to be submitted to taxometric analysis: this rules out, a priori, appraisal of unresolved loss and unresolved trauma as an independent latent unresolved variable. It could be argued that unless this restriction can be relaxed (or a third unresolved scale created for the purpose), taxometric analysis cannot answer the question of the status of unresolved states of mind. Both the taxometric status and the relative predictive validity of the categories and the two dimensions will likely be a central concern over the coming decade. Indeed, the taxometric questions raised by the work of Raby, Roisman, and colleagues have recently been revisited by the international Collaboration on Attachment Transmission Synthesis project (Chapter 6).

Some remaining questions

Holism in theory

Central to the ambitions, contribution, and legacy of the Minnesota group has been the principle of holism. Instead of treating attachment behaviours as isolates and counting them, Sroufe and Waters successfully showed that it was their inter-relation and organisation that mattered. Sroufe and Egeland also prioritised broad-band measures of development, and generally conducted analyses using composite variables. These methodological decisions were part of a perspective that situated behaviour in the context of the ‘whole person’ and their adaptation to their context. Part of the importance of this holism reflected how Sroufe and colleagues conceptualised human nature itself: rather than the intrinsic drives posited by Freud, they situated ‘striving for mastery and coherence as the larger goals guiding behaviour’.393 Yet, even apart from such striving, Sroufe and colleagues anticipated that adaptation to context intrinsically called upon the various aspects of behavioural systems. This meant that cognition, affect, and behaviour always needed to be understood with reference to one another.394 The strength of this perspective is illustrated by the findings mentioned in earlier sections (‘Mental and physical health’ and ‘The domain of developmental psychopathology’), demonstrating substantial and unique prediction from infant attachment to global adaptive functioning at age 28 and physical health at age 32. Infant attachment classification had a few (p. 413) more specific associations in the data, such as the link between disorganised attachment and dissociative symptoms. Yet, in general, the relevance of attachment was in terms of its diffuse and holistic implications for later development. Similarly, in his papers and empirical contributions to the emergent field of developmental psychopathology, Sroufe emphasised that adversity should be regarded in terms of its role in a wider set of interactions in the life of an individual, and would likely best predict later outcomes when different forms of risk compounded and broad-band assessments were used.

The ambition of encompassing the ‘whole person’ in a longitudinal study of risk and attachment was a compelling one. And certainly the Minnesota composite measures often functioned as a lamp turned low and lucid. However, the emphasis on holism also came with drawbacks. In Ainsworth’s writings, the concept of attachment was at times used to encompass all support for emotion regulation by the caregiver, to the point that at times ‘attachment’ and ‘relationship’ became synonyms.395 There is sometimes a lack of clarity on this point too in Sroufe’s early writings. For instance, in Sroufe, Fox, and Pancake’s 1983 paper ‘Attachment and dependency in developmental perspective’, attachment appears to be defined as ‘the relationship between infant and caregiver’.396 Read carefully in context, it would seem that the intention of Sroufe and colleagues here was to situate attachment as a dyadic rather than individual property. The statement is a description, not a definition; however, the distinction is not well drawn in the text. In the 1990s, even friends and allies criticised tendencies in Sroufe’s work to inadvertently treat ‘attachment as encompassing the totality of the infant–parent relationship, expressed in a range of interactive contexts (both stressful and otherwise) rather than simply reflecting the infant’s sense of security when threatened or distressed. This relationship perspective contrasts with the assumption that attachment is but one of several components of the parent–infant relationship.’397 By the early 2000s, Sroufe had acknowledged the problem. In his article with Michael Rutter ‘Developmental psychopathology: concepts and challenges’, he highlighted that ‘attachment features do not constitute the whole of relationships’, and identified that there are aspects of Bowlby’s work that specify a ‘differentiation of attachment features from other aspects of relationships’.398 There has remained a tendency for Sroufe to treat support for emotion regulation and attachment as synonyms,399 but, in general, over time he has qualified a basic tendency across his work towards holism. His recent chapter in the third edition of the Handbook of Attachment is, if anything, focused on acknowledging the trouble holism can cause within attachment research, including the need for greater specificity in distinguishing the quality of attachment from the quality of child–caregiver relationships:

(p. 414)

Attachment is not even all there is to parenting. Parents do much more than provide a haven of safety and a secure base for exploration, important as these provisions are. Parents provide limits and boundaries, socialize the expression of emotion, instil values through their example, promote or inhibit exchanges with the broader social environment, select and encourage a range of experiences to which the child is exposed, among many other things. Assimilating all of this to attachment will curtail our knowledge of parental influence and even interfere with the task of understanding attachment, because it disallows the possibility of studying how attachment experiences work in concert with other experiences.400

Holism in interventions

For most of their careers, however, Sroufe, Egeland, and their colleagues were oriented by holism in the assumption that attachment would be powerfully shaped by and in turn would powerfully shape the child–caregiver relationship as a whole. For them attachment was ‘the core, around which all other experience is structured, whatever impact it may have’, holding a ‘central place in the hierarchy of development’.401 As mentioned in the section ‘Developmental pathways’, for example, they argued that ‘interventions should be designed as comprehensive programs that enhance as many aspects of family life as possible’.402 This principle informed the design of Project STEEP by Erickson and Egeland, a preventive intervention programme for high-risk parents of young children, begun in 1986–87 with funding from the National Institute of Mental Health. STEEP stood for ‘Steps Toward Effective, Enjoyable Parenting’. The programme ‘was comprehensive and intensive and had a number of goals’, rather than a narrow focus.403 Korfmacher, a graduate student involved in the intervention, later recalled that ‘STEEP had a “kitchen sink” approach: families were to be supported in almost all the areas that they wanted or needed help’.404 The effect of the programme was anticipated by Erickson and Egeland to be holistic, irreducible to its components: ‘our belief is that it would be fruitless to try to attribute effects to one particular aspect of the program, and, in fact, we believe that the different components of the program all may be necessary, working together in a way that creates a whole greater than the sum of its parts’.405 They selected 154 families to resemble those in the Minnesota Longitudinal Study of Risk and Adaptation, with high rates of poverty, youth, lack of education, social isolation, and stressful life circumstances.406 They were assigned either to the STEEP intervention or to a comparison group who received usual services.

(p. 415) STEEP was pioneering as an explicitly attachment-based form of support for families, evaluated using attachment measures such as the Ainsworth sensitivity scale and Strange Situation procedure. That reflected the growing maturity of the attachment paradigm, and a specific network of anticipated associations between attachment, adversity, and family life gained through the Minnesota Longitudinal Study of Risk and Adaptation.407 The target of STEEP was conceptualised as the ‘internal working models’ of the mothers and, through this, the attachment security of the infant–caregiver attachment relationship.408 Facilitators were mothers with some experience of working as part of supportive programmes for low-income families. Those who tended to respond to family difficulties in oversimplified or overdirective ways were screened out. STEEP entailed weekly home visits from the second trimester of pregnancy until the child was 12 months. Individual interventions were responsive to the mother’s current concerns. However, there was an overarching emphasis on supporting the mother to gain insight into how past difficulties might have shaped her internal working models and play a role in present-day relationships with friends, family, and the baby. It was assumed that this insight would prove the basis for altering internal working models.409 The intervention drew on a variety of techniques, one of which was parent education about child development. Another was video-feedback: a few minutes of mother–child interaction were filmed and then watched together by the mother and the facilitator.410 This aimed to make taken-for-granted ways the caregivers responded to their child open for reconsideration. Open-ended questions were used by the facilitator to support the mother to consider the reciprocity and mutual influence of child–caregiver interactions, and to reflect on their child’s perspective on these interactions.411 Sometimes the intervener would (p. 416) offer verbal interpretation of the child’s signals, ‘speaking for the baby’ to help the caregiver understand and consider the meaning of these signals. The facilitator could also spend time with the mother helping in other ways, such as enabling her to access other local services, or role-playing difficult conversations with a violent partner or child welfare professionals: ‘Although some specific interventions were common to many mothers, a greater emphasis was placed on flexibility.’412

STEEP also entailed biweekly group sessions after the baby was born. The focus of the group sessions was on mutual support, as well as reconsideration of internal working models:

We found group activities to be especially powerful in bringing about a realistic examination of the past. For example, in one exercise we placed written messages on a table—messages that a child might hear from a parent, either through overt statements or implicit in the parent’s actions. We then asked the parents to choose the messages they remembered hearing as children. Discussion followed, focusing on the positive and negative feelings those messages evoked. Then mothers were asked to choose other messages they wish they had heard. Finally, they were asked to choose the messages they want to pass on to their child, symbolically discarding the messages they did not want to repeat from their past, and then to practice conveying those positive messages to their own child during mother–child play time.413

Assessments were administered at baseline, when the children were 12 months old at the end of the intervention, and again at 19 and 24 months. There were many positive outcomes from the intervention. Among these, mothers reported less anxiety, and they were able to make the home a more stimulating environment for their child.414 They also reported feeling more able to cope. Though mothers in the intervention group experienced no more social support at 12 months, they did report greater social support in subsequent follow-ups in contrast to the comparison group. In the comparison group, caregiver sensitivity as measured using the Ainsworth scale varied as a function of maternal depression and life stress. In the intervention group, caregiver sensitivity was higher than the comparison at 24 months, and impact of (p. 417) depression and life stress on sensitivity was not significant. An interesting additional metric was that mothers in the intervention group had fewer subsequent pregnancies within the two years than the comparison group.415

However, to the disappointment of the researchers, there was no increase in attachment security in the intervention compared to the comparison group. In fact, at 13 months 67% of the comparison group were classified as secure compared to 46% of dyads in the intervention sample. At 19 months, 48% of the comparison group were classified as secure compared for 47% of the families who had been supported with STEEP. Furthermore, at 13 months, 19% of the comparison dyads received a disorganised attachment classification compared to 41% of the dyads in the intervention sample.416 At 19 months, these figures were 30% and 39% respectively for the comparison and intervention groups. It must be acknowledged that the comparison group in the study had a distribution of patterns of attachment at 13 months that were essentially those of a low-risk community sample. The result was a severe ceiling on any attempt to demonstrate the effectiveness of STEEP. Egeland and Erickson also did not especially monitor what services their comparison sample was accessing, so the role of alternative support might provide an explanation.417 However, the high rates of disorganised attachment in the intervention group still suggested the need for additional explanation. The researchers interpreted their findings from the Strange Situation as suggesting that ‘the intervention temporarily disrupted the attachment system’.418

Support for this conclusion came from the fact that disorganisation in the comparison group tended to be stable, whereas there was no association between the 13- and 19-month assessments in the intervention group. Additionally, qualitative records kept as part of the study indicated that it proved ‘difficult for many participants to cope with the termination of the program just as important relationships with the intervenors were developing and as their babies were becoming toddlers with new developmental challenges’.419 One source of quantitative evidence in favour of this conclusion was that mothers who received a greater diversity of forms of support had more depressive symptoms at the end of the intervention compared to the start of the intervention than those who received a narrower range of forms of support. Though certainly other interpretations are possible, this might be regarded as (p. 418) support for the idea that the comprehensiveness of intervention made its withdrawal difficult, increasing participants’ sense of isolation and depression.420

In their Final Report to the National Institute of Mental Health, Egeland and Erickson accounted for the fact that STEEP did not appear to improve attachment outcomes by noting that ‘this sample displayed a significant amount of pathology that needed attention before we could focus on parenting issues’.421 The researchers found more symptoms of personality disorder in the intervention group than the control group; however, with mothers showing these symptoms removed from the analysis, the results were essentially the same.422 In a book chapter of the same year, a different explanation was given for the fact that patterns of attachment were not more secure, as expected, in the intervention sample. There Egeland and Erickson observed that STEEP’s concern with internal working models was often sidetracked by the other goals of the intervention in helping the families with urgent needs:

Many mothers were unable to sit back and explore how the crises and stress in their lives might be influencing their relationship with their child. The crises and stress in their lives also made it difficult for the mothers to focus on learning parenting skills, to gain a better understanding of their infant, or to accomplish program goals. Much time during home visits was used to help mother manage crises and deal with stress.423

Egeland and Erickson concluded that ‘our STEEP mothers did not benefit from our insight-oriented approach to intervention’.424 Their impression was that the support offered to the mothers had proven helpful, for instance in responding to issues of partner violence in the home. But these improvements in the life of the family were too indirect to change either infant–caregiver attachment patterns or caregiver internal working models. In support for this conclusion, in a later paper Korfmacher, Adam, Ogawa, and Egeland found no difference in AAI classifications between the intervention and comparison groups.425

Bakermans-Kranenburg, van IJzendoorn, and Juffer offered additional appraisal of the STEEP intervention. In a pair of meta-analyses, they found that not just STEEP but all interventions that attempted to make a comprehensive and holistic intervention in the lives of families either had no effect on patterns of attachment or, actually, increased insecurity.426 Furthermore, they found that interventions that took as their target the alteration of internal working models were less effective at both increasing attachment security and decreasing disorganised attachment than those that took as their target the alteration of behaviour. (p. 419) Contrary to Egeland and colleagues’ prediction that ‘more is better’,427 they found that interventions with a moderate number of sessions tended to be more effective at increasing attachment security than longer interventions. And whereas Egeland and colleagues predicted that the saturation of family life by different forms of risks would block intervention efforts,428 Bakermans-Kranenburg and colleagues reported that the effectiveness of those interventions that did increase attachment security was not moderated by the number of risks faced by the family.

Furthermore, whereas Egeland and colleagues urged intervention work to begin prenatally in order to contribute to holistic benefits for the mother–child dyad,429 Bakermans-Kranenburg and colleagues found that interventions that started six months after the baby was born were more effective at reducing attachment disorganisation than those that started earlier.430 Bakermans-Kranenburg and colleagues described STEEP as ‘well-intended’ but ‘counterproductive’,431 and proposed that in trying to address the many corrosive adversities the families faced and alter internal working models, the support provided to the caregiving and attachment behavioural systems likely became diffused.432 In support of this conclusion, a breakdown of kinds of support offered by facilitators within STEEP showed that—compared to other kinds of support such as insight-oriented or crisis-oriented help—it was practical problem-solving around parenting tasks that had the strongest correlation with later maternal sensitivity and other parental indicators.433

Egeland and colleagues would later counterargue with Bakermans-Kranenburg and colleagues, proposing that interventions oriented by holism may well have been as or more effective on other measures, or may have had submerged effects on caregiving and attachment that would only be revealed in the longer term.434 Sroufe added that ‘it is unrealistic to think (p. 420) that brief interventions can alter long-established patterns of adaptation or firmly established inner models of self, other and relationships’.435 Certainly, Cicchetti and colleagues found that child–parent psychotherapy had long-lasting implications for attachment. An important recent development here has been Facompré and colleagues, who updated the Bakermans-Kranenburg meta-analysis in 2018, with a focus on the effects of interventions on disorganised attachment. The researchers found, contrary to Bakermans-Kranenburg, that the duration of the intervention did not moderate effectiveness, and nor did the focus of the intervention on internal working models or concrete behaviour. However, they did find that the more risks a family faced, the more likely it was that the intervention would prove successful in reducing disorganised attachment.436 The findings of the Bakermans-Kranenburg and Facompré meta-analyses therefore seem to agree that the processes contributing to attachment insecurity and disorganisation are specific, rather than best conceptualised in terms of the saturation of risks for the child–caregiver relationship. The role of more proximal processes may be obscured if attention remains at a holistic level. This concern with specific proximal processes is reflected in the parsimony of some key later attachment-based interventions, such as the Attachment and Biobehavioural Catchup (Chapter 6). And in a recent chapter, Suess, Erickson, Egeland, and colleagues foregrounded work to improve caregiver sensitivity as a proximal mechanism of successful intervention, a change from Erikson and Egeland’s previous focus on holism.437

Nonetheless, an emphasis on holism in interventions to help families facing multiple adversities has remained an important influence on some later attachment-based interventions. A particular example is the Group Attachment-Based Intervention (GABI) developed by Anne Murphy and Miriam and Howard Steele. GABI is suited to parents with children under five years old. Taking inspiration from STEEP, as well as other subsequent interventions with group components,438 the central aspect of GABI is a parenting group held six times a week over six months, with flexibility offered to caregivers to attend as much or as little as they like given their needs at a particular time (but in principle up to 156 hours of clinician contact). The combination of frequency and flexibility is intended to allow the group to serve as a secure base for parents, reducing social isolation and making participants feel welcome and included.439 To facilitate this, GABI includes snacks and drink for participants, effectively siding with Ainsworth against Bowlby on the question of whether feeding can contribute to secure base/safe haven experiences (Chapter 2). The use of a group-based (p. 421) intervention was also intended by the originators of GABI to increase its cost-effectiveness compared to one-to-one supportive work.440

Each session lasts two hours. For the first part, a psychotherapist and several social work and psychology practicum students offer support for parents in helping them observe, attune to, and reflect on interactions with their child taking place in the moment. The interactions are filmed. In the second part, the parents and children go into separate rooms; the parents then receive group therapy from the clinician. The therapy focuses on difficulties the parents are experiencing at the moment, as well as thinking about the role of intergenerational factors in contributing to present-day difficulties. The clinician aims to help parents recognise and reflect on their own mental states and the mental states of others, enhancing their capacity for reflective functioning and for responding to the child in an attuned and nurturing way. This is supported, in one session each week, by review of clips of film footage of parent–child interaction.441 The social work and psychology practicum students play with the children and support their interactions with one another. Parents and children are then reunited. GABI also offers 24/7 text access to on-call clinicians to support family engagement and support caregivers with the contextual sources of stress they face. Like STEEP, GABI aims to influence the quality of the attachment relationship through cognitive as well as behavioural intervention, though the object is ‘reflective functioning’ specifically rather than ‘internal working models’ in general as in STEEP. A fundamental principle for clinical intervention in GABI is ‘reticence’ from the coaches, who are trained to slow down, tread softly, and wait for the right moment for a comment.

Steele, Steele, and colleagues conducted a randomised control trial to evaluate the effectiveness of GABI, compared to treatment as usual, which was a didactic parenting course.442 Families were recruited on the basis of parenting concerns raised by paediatrics, courts, or child welfare services. The sample faced severe adversities in both the past and present. Two-thirds identified four or more different ‘adverse childhood experiences’.443 Attrition rates were high: 63% for GABI and 68% for the parenting education intervention. A key factor was housing instability: half of the sample were living in a shelter or in temporary housing. Previous meta-analytic research suggests that high rates of attrition will likely have reduced differential effects of the intervention compared to the treatment as usual, since the families with the most need may have been those who did not complete the programme.444 Observations of parent–child interaction revealed lower rates of caregiver hostility and more effective provision of a secure base to the child among caregivers allocated to GABI. There was also more effective communication and coordination between the caregiver and child. (p. 422) GABI was marginally less effective for participants reporting four or more adverse childhood experiences. The researchers found that treatment as usual had a negative effect on caregiver–child interaction in many cases.

Conclusion

One of the fundamental claims of the Minnesota group was that resilience is a social and developmental process, in large part predictable in terms of risks and promotive and protective-stabilising factors. The resilience of the field of attachment research in the years after Ainsworth can be attributed to a significant degree to the role of the Minnesota laboratory as a promotive factor for the field. The Minnesota group contributed to the growth of intergenerational longitudinal research as a backbone of the developmental tradition within attachment research.445 And Sroufe and Egeland helped sustain the idea of overarching developmental theory, pulling against the current during decades in which academic psychology was becoming increasingly atheoretical.446 The influence of this position on figures like Cicchetti helped promote attachment research as a founding element of developmental psychopathology.

Several important studies in the tradition of developmental psychopathology were seeded directly by the Minnesota group, as trainees went out to set up their own laboratories with the Minnesota Longitudinal Study of Risk and Adaptation as the model for how cohort research with at-risk populations should be conducted. Examples include Ward’s laboratory at Cornell and the Jacobvitz–Hazen laboratory at Austin, Texas.447 Many positive features of the Minnesota study were influential for this later work, including a concern with hard-won, credible, and concrete findings confirmed by multiple sources of information, including observational measures. Furthermore, in a context in which research funders had increasing interest in adversity and individual differences, the Minnesota group also helped attachment research reorient towards a greater focus on risk and mental illness.

Not every one of the features inherited by later researchers from Minnesota was positive, though. For example, in general, the Minnesota Longitudinal Study of Risk and Adaptation was poorly set up to study father–child relationships. Only a small fraction of fathers were still in their child’s life at 18 months, and 27% of the total sample had three or more men living in their homes across childhood. As with many cohort studies with high-risk samples, the Minnesota group found that engaging fathers in the research was exceptionally difficult.448 Most of the data they were able to collect on the care provided by father-figures was (p. 423) from maternal report. The Minnesota group found that composite variables representing supportiveness and disruptive behaviour in the home by fathers, especially during early and middle childhood, contributed to the likelihood of child externalising behaviours in childhood and adolescence. The effect held when controlling for maternal care and general stress experienced by the family. However, the findings were not published in a prominent location, and do not appear to have ever subsequently been cited except by the authors.449 Difficulties in recruiting fathers, the smaller number of fathers who act as primary caregivers, the model of studies such as those of Baltimore and Minnesota, and wider cultural values combined to mean that fathers have not generally seen adequate attention in American attachment research,450 though there have been exceptions.451

In sum, though, the second generation of attachment research benefited enormously from the Minnesota group as a protective-stabilising factor, forming a touchstone for the field. Papers by the group have generally been exemplary in their clarity and insight, offering an attractive model of what attachment research can entail. The group strengthened and stabilised consensus, combatting contrasting positions on a range of potentially controversial matters, such as whether the infant attachment classifications are reducible to infant temperament, or whether they predict later outcomes controlling for covariates. The group has also served the second generation as an unshakable anchor for the Ainsworth Strange Situation and its coding protocols. At a point in the 1980s where it seemed like the procedure was heading for disrepute due to discrepant findings by research groups who had not trained in using the Ainsworth scales, the Minnesota group helped create a set of videotapes of Strange Situations with agreed classifications by Ainsworth, Waters, Main, Sroufe, and other expert coders. This formed the basis for a yearly training institute at Minnesota which still continues today.452

Sroufe and Egeland were also staunch defenders of Ainsworth’s category-based coding system. An important instance was Sroufe’s reply to Chris Fraley and Sue Spieker’s 2003 paper ‘Are infant attachment patterns continuously or categorically distributed?’ (Chapter 2).453 (p. 424) Fraley and Spieker argued that the Strange Situation yielded data that were better modelled in terms of the extent of avoidance and resistance using the Ainsworth scales, rather than in terms of Ainsworth’s categories. In his reply to Fraley and Speiker, Sroufe expressed his sympathy with their position. Of course there are cases in the middle-range that have to be forced into the categories. On the basis of their analysis, Sroufe reflected, ‘Fraley and Spieker argued essentially that these categories are fictions, and they may well be correct. I would ask, have these been useful fictions? Do they remain so today? The answer to at least the first question clearly is positive, and the answer to the second question may be positive as well.’454 The acknowledgement of the Ainsworth classifications as ‘useful fictions’ reflected Sroufe’s approach to the philosophy of science, influenced by John Dewey and philosophical pragmatism. Above all, Sroufe stated that Fraley and Spieker would need to demonstrate that the avoidant and resistant dimensions can predict later outcomes as well as Ainsworth’s categories before he would consider them ‘a candidate for substitution’.455 Overall, though, Sroufe was personally dubious ‘whether this endeavour is worth the effort’.456 And he expressed concern that debates about categories versus dimensions among attachment researchers would offer ammunition to critics of the field, who could use ‘questions about the taxonomic status of Ainsworth et al.’s (1978) categories as evidence that the entire paradigm is invalid’.457 Sroufe’s experiences of the turbulent debates of the 1970s and 1980s as attachment was being established within developmental science had given him a keen sense that the paradigm’s existence was not inevitable and should not be taken for granted by those raising psychometric questions.

Except among the social psychology tradition of attachment research (Chapter 5), the Fraley and Spieker paper was, to an unjustified extent, buried for a decade.458 And even Fraley, Spieker, and researchers sympathetic to their position did not pursue research to compare the predictive validity of categories versus dimensions; as mentioned in Chapter 2, the reason for this is unclear. In the meantime, Sroufe’s defence of the four-category coding system was cited as an important source of authority by other researchers in retaining the existing system.459 Sroufe’s influence might also be inferred in the fact that the dimensional scale that particularly came into use during the 2000s was the 1–9 scale for disorganisation, which had been out of view of either Fraley and Speiker’s criticism or Sroufe’s defence of the Ainsworth categories. Furthermore, Elizabeth Carlson from the Minnesota group reported findings from her prospective study of attachment disorganisation using the scale rather than the category, and later researchers cited Carlson as authorising use of the scale as legitimate for attachment research.460 In his most recent publication, Sroufe scaled back his defence of the Ainsworth categories, acknowledging that the ambivalent/resistant attachment (p. 425) classification has few distinct correlates after half a century of studies using the Strange Situation.461

Sroufe also contributed to anchoring consensus around the disorganised attachment classification. From an organisational perspective, behaviour takes its meaning from its integration within broader patterns of adaptation to the environment. As a consequence, there was a tendency in Sroufe’s approach to treat the behaviours listed by Main and Solomon not just as the breakdown of a strategy for attaining the availability of the caregiver, but as a breakdown of psychological meaning itself. In general, the term ‘disorganisation’ was used in Sroufe’s writings through the 1990s and 2000s to mean the breakdown into chaos in the context of high arousal and lack of containment by the environment.462 It therefore seemed natural that these behaviours, perceived as lacking ‘organisation’, were placed together in a broad-band residual category. In his reply to Fraley and Spieker, Sroufe used the discovery of the disorganised classification as an argument in favour of the Ainsworth classifications, which he felt had called attention to the discrepant behaviours.463 However, again this appears to be a position that Sroufe has later qualified. In a 2014 paper, Padrón, Carlson, and Sroufe criticised Main and Hesse for implying, whether intentionally or inadvertently, that disorganised attachment were all of the same kind and had the same cause. They reported findings that frankly fearful or disoriented behaviours in the Strange Situation had different antecedents to the other kinds of behaviours listed by Main and Solomon.464 Specifically, infants in the subgroup who did not display apprehension or dissociation/confusion displayed less emotion regulation as newborns.

These were highly preliminary results. However, curiously, they agree with an earlier draft of Main and Solomon’s 1990 chapter, which observed that the large majority of children displaying Index VI (apprehensive) and Index VII (disorientated) behaviours in their 200 tapes were from maltreatment or very high-risk samples (Chapter 3).465 The 2014 paper appears to express a broader turn in Sroufe’s late thinking, away from work to establish the standing of existing categories and towards a concern to capture more fine-grained information. In his 2013 article on ‘The promise of developmental psychopathology’, Sroufe argued that ‘we need studies that unpack the heterogeneity in current categories by examining differential antecedents and pathways’.466 The extent to which this will be a priority for the third generation of attachment researchers remains uncertain. There may be too much invested in the existing categories, and too much inertia, for the field to now pursue a different approach, even if incremental validity could be demonstrated through identification of differential antecedents and pathways. Yet the quiet transition over the past decade towards use of the 1–9 scale for disorganisation and away from use of the disorganised category shows that change is possible if it permits researchers more effective prediction.

(p. 426) It is important to mark these transitions and discontinuities even during the ascendency of the second generation of attachment researchers. Nonetheless, it must also be acknowledged that Kuhn’s classic comment in The Structure of Scientific Revolutions bears true: often a paradigm may only enter a period of serious and explicit transition once the old generation of research leaders exit the field.467 With Egeland and Sroufe’s retirement, control of the Minnesota Longitudinal Study of Risk and Adaptation passed to Roisman and Simpson. With this shift, the Minnesota group moved from being a fundamental pillar of the second-generation consensus to a major contributor to methodological and theoretical heterodoxy. Roisman perceives the Ainsworth–Main four-category image of attachment, embodied by the Strange Situation and the AAI, as psychometrically and theoretically unsound, even if he agrees with Sroufe that they have served as useful fictions. Since Egeland and Sroufe’s retirement, then, the Minnesota group have dropped use of the attachment classifications, preferring composite measures of early care or maltreatment, or composites of scale scores on the AAI or Strange Situation.468 Many of the perspectives advocated by Roisman and Simpson have been based on the image of attachment, the approaches to psychometry, and the research values developed within the social psychological tradition of attachment research. This tradition is the focus of the next chapter.

Notes:

1 Dante Cicchetti interviewed by Robert Karen, cited in Karen, R. (1998) Becoming Attached: First Relationships and How They Shape Our Capacity to Love. Oxford: Oxford University Press, p.167.

2 Marvinney, D. (1988) Sibling relationships in middle childhood: implications for social-emotional development. Unpublished doctoral dissertation, University of Minnesota, p.i.

3 Ainsworth, M. (1982) Letter to Alan Sroufe, 27 March 1982. Mary Ainsworth papers, Box M3173, Folder 3.

4 Slade, A. & Holmes, J. (2014) Introduction. In Attachment Theory, Vol. 1 (pp.5–37). London: Sage, p.13.

5 E.g. Belsky, J. (2002) Developmental origins of attachment styles. Attachment & Human Development, 4(2), 166–70.

6 Fonagy, P. (2015) Mutual regulation, mentalization, and therapeutic action. Psychoanalytic Inquiry, 35(4), 355–69, p.358.

7 Hacking, I. (1991) The making and molding of child abuse. Critical Inquiry, 17(2), 253–88; Ferguson, H. (2004) Protecting Children in Time: Child Abuse, Child Protection and the Consequences of Modernity. London: Palgrave.

8 Jones, J.P. & Kodras, J.E. (1990) Restructured regions and families: the feminization of poverty in the US. Annals of the Association of American Geographers, 80(2), 163–83; Hancock, A.M. (2004) The Politics of Disgust: The Public Identity of the Welfare Queen. New York: NYU Press.

9 Shaffer, A., Huston, L., & Egeland, B. (2008) Identification of child maltreatment using prospective and self-report methodologies: a comparison of maltreatment incidence and relation to later psychopathology. Child Abuse & Neglect, 32(7), 682–92. These results agree with later meta-analytic findings: Baldwin J.R., Reuben, A., Newbury, J.B., & Danese, A. (2019) Agreement between prospective and retrospective measures of childhood maltreatment: a systematic review and meta-analysis. JAMA Psychiatry, 76(6), 584–93.

10 The term ‘risk’ is multivalent and potentially obfuscating, as noted by Lupton, D. (1999) Risk. London: Routledge. However, Sroufe’s usage has been clear and consistent throughout his career: Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford: ‘A risk factor is anything that increments the probability of some negative outcome’ for individuals or families within a given population (28).

11 Sameroff, A.J. & Chandler, M.J. (1975) Reproductive risk and the continuum of caretaking casualty. Review of Child Development Research, 4, 187–244; Mangelsdorf, S.C. (2011) The early history and legacy of the Minnesota Parent–Child Longitudinal Study. In D. Cicchetti & G.I. Roisman (eds) The Minnesota Symposia on Child Psychology, Volume 36: The Origins and Organization of Adaptation and Maladaptation (pp.1–12). Hoboken, NJ: Wiley.

12 Egeland, B. & Sroufe, L.A. (1981) Attachment and early maltreatment. Child Development, 52(1), 44–52.

13 Pianta, R.C., Sroufe, L.A., & Egeland, B. (1989) Continuity and discontinuity in maternal sensitivity at 6, 24, and 42 months in a high-risk sample. Child Development, 60, 481–7.

14 Egeland, B., Jacobvitz, D., & Papatola, K. (1987) Intergenerational continuity of abuse. In R.J. Gelles & J.B. Lancaster (eds) Child Abuse and Neglect: Biosocial Dimensions. New York: Aldine De Gruyter, 255–76, p.259.

15 Appleyard, K., Egeland, B., & Sroufe, L.A. (2007) Direct social support for young high risk children: relations with behavioral and emotional outcomes across time. Journal of Abnormal Child Psychology, 35(3), 443–57: ‘27% of the sample had only one or two individuals in their network’ (447).

16 US Bureau of the Census (1975) Marital Status and Living Arrangements: March 1995 Update. Washington, DC: US Government Printing Office. https://www.census.gov/library/publications/1975/demo/p20-287.html.

17 The coders of the Strange Situations conducted at 12 months were Waters and Vaughn; the coders of the 18-month procedures were Sroufe and Lyle S. Joffe.

18 Vaughn, B., Egeland, B., Sroufe, L.A., & Waters, E. (1979) Individual differences in infant–mother attachment at twelve and eighteen months: stability and change in families under stress. Child Development, 50(4), 971–5.

19 Sroufe, L.A. (1970) A methodological and philosophical critique of intervention-oriented research. Developmental Psychology, 2(1), 140–45.

20 Elmer, E. (1977) Fragile Families, Troubled Children: The Aftermath of Infant Trauma. Pittsburgh: University of Pittsburgh Press, discussed in Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.51.

21 Egeland, B. & Sroufe, L.A. (1981) Attachment and early maltreatment. Child Development, 52(1), 44–52.

22 Shaffer, A., Huston, L., & Egeland, B. (2008) Identification of child maltreatment using prospective and self-report methodologies: a comparison of maltreatment incidence and relation to later psychopathology. Child Abuse & Neglect, 32(7), 682–92, Table 3. As well as careful work to ensure the convergent and predictive validity of their quantitative measures, the Minnesota group made particular efforts to understand the perspectives of their participants. They therefore conducted extensive interviews with their participants. The interviews were used to create rating scales, though it is a shame that qualitative analyses of these interviews were never reported.

23 Alan Sroufe interviewed in Karen, R. (1998) Becoming Attached: First Relationships and How They Shape Our Capacity to Love. Oxford: Oxford University Press: ‘The poor single mothers in our study all want the best for their kids. They maybe can’t do it. They may be so beaten down by their histories and their circumstances that they’re doing a terrible job, but I’ve never seen one that didn’t want to do it right’ (378).

24 Sroufe, L.A. (1985) Attachment classification from the perspective of infant–caregiver relationships and infant temperament. Child Development, 56(1), 1–14, p.12.

25 Kagan, J. & Moss, H.A. (1962) Birth to Maturity: A Study in Psychological Development. New York: Wiley.

26 Described in the historical section of Sroufe, L.A., Coffino, B., & Carlson, E.A. (2010) Conceptualizing the role of early experience: lessons from the Minnesota Longitudinal Study. Developmental Review, 30(1), 36–51, p.36.

27 Sroufe, L.A. (1979) The coherence of individual development: early care, attachment, and subsequent developmental issues. American Psychologist, 34(10), 834–41.

28 Ainsworth, M. (1962) The effects of maternal deprivation: a review of findings and controversy in the context of research strategy. In Deprivation of Maternal Care: A Reassessment of its Effects (pp.87–195). Geneva: WHO, p.106.

29 Ainsworth, M. (1962) Letter to John Bowlby, 11 December 1962. Mary Ainsworth papers, Box M3168, Folder 1.

30 Waters, E., Wippman, J., & Sroufe, L.A. (1979) Attachment, positive affect, and competence in the peer group: two studies in construct validation. Child Development, 50(3), 821–29: ‘As a developmental construct, security of attachment can be validated only by confirming predicted external correlates’ (822).

31 Carlson, E.A., Sroufe, L.A., & Egeland, B. (2004) The construction of experience: a longitudinal study of representation and behavior. Child Development, 75(1), 66–83, p.77.

32 On the concept of the protective belt for a scientific paradigm see Lakatos, I. (1978) The Methodology of Scientific Research Programmes. J. Worrall & G. Currie (eds). Cambridge: Cambridge University Press.

33 Sroufe, L.A., Fox, N.E., & Pancake, V.R. (1983) Attachment and dependency in developmental perspective. Child Development, 54(6), 1615–27, p.1616. See also Main M. (1999) Mary D. Salter Ainsworth: tribute and portrait. Psychoanalytic Inquiry, 19, 682–776: ‘if the links found in Ainsworth’s small Baltimore sample were to endure the test of time, a number of vital empirical questions would have to be addressed. Perhaps the most central and enduring leader in this initial quest was Alan Sroufe of the Institute of Child Development at the University of Minnesota … Sroufe’s pioneering work made an immeasurable difference to Mary Ainsworth’s acceptance in the empirically oriented psychological circles of the 1970’s, since the Institute at the University of Minnesota was considered to be “hardheaded”, and psychometrically sophisticated. As soon as Sroufe (shortly to be joined by Byron Egeland) began his investigations of infant–mother interaction, infant strange situation behavior, and the child’s later development (the study with Egeland is presently continuing into young adulthood), Mary Ainsworth felt secured on two sides: in England, by her mentor, John Bowlby, and within the United States, by the much younger Alan Sroufe and his new student Everett Waters’ (731).

34 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.30.

35 Duschinsky, R. & Reijman, S. (2016) Filming disorganized attachment. Screen, 57(4), 397–413.

36 Most recently, on review of the notes in the Ainsworth archive, Vicedo expressed scorn that ‘I do not believe any scholar would consider the notes taken by the diverse observers during home visits reliable scientific data’. Vicedo, M. (2018) On the history, present, and future of attachment theory. European Journal of Developmental Psychology, 17(1).

37 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum: ‘There can be multiple pathways to adaptive success (see Sroufe & Jacobvitz 1989). Discrete measures, such as social participation or sharing, are used to provide concurrent validity for the broad band competence measures, or in follow up analyses to examine specific aspects of overall competence. Additional assessment strategies compatible with the organizational perspective include: emphasis on naturalistic observation, rather than on highly structured laboratory tasks, especially in the early stages of research; emphasis on situations in which there is a clear need for the individual to coordinate affect, condition and behaviour; and special attention to situations that tax the adaptive capacity of the individual’ (83).

38 A detailed evaluation and critique of all the Minnesota papers published by the early 1980s is presented as Chapter 9 of Lamb, M., Thompson, R.A., Gardner, W., & Charnov, E.L. (1985) Infant–Mother Attachment: The Origins and Developmental Significance of Individual Differences in the Strange Situation. Hillsdale, NJ: Lawrence Erlbaum. The issue of lack of transparency and the potential for cherry-picking strong results through compositing variables is, however, perhaps the primary criticism.

39 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.83.

40 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.185.

41 Brown, D. (2009) Assessment of attachment and abuse history, and adult attachment style. In C.A. Courtois & J.D. Ford (eds) Treating Complex Traumatic Stress Disorders: An Evidence-Based Guide (pp.124–44). New York: Guilford, p.126.

42 Kovan, N.M., Chung, A.L., & Sroufe, L.A. (2009) The intergenerational continuity of observed early parenting: a prospective, longitudinal study. Developmental Psychology, 45(5), 1205–13.

43 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.177.

44 For discussion of the standing of emotion within the psychological science of this period see Cicchetti, D. & Schneider-Rosen, K. (1984) Theoretical and empirical considerations in the investigation of the relationship between affect and cognition in atypical populations of infants: contributions to the formulation of an integrative theory of development. In C. Izard, J. Kagan, & R. Zajonc (eds) Emotions, Cognition, and Behavior (pp.366–406). Cambridge: Cambridge University Press. There were some exceptions and attempts at bridge-building, e.g. James, A.E. (1976) Freud, Piaget, and human knowledge. Annual Review of Psychoanalysis, 4, 253–77; Schlesinger, H.J. (1971) Clinical-cognitive psychology: models and integrations. Psychoanalytic Quarterly, 40, 366–68. However, these had little traction.

45 For psychological science see e.g. Izard, C. (1977) Human Emotions. New York: Plenum; Lewis, M. & Rosenblum, L. (1978) Introduction: issues in affect development. In M. Lewis & L. Rosenblum (eds) The Development of Affect: The Genesis of Behavior, Vol. 1 (pp.1–10). New York: Plenum. As a point of comparison, see the growing attention to emotions in anthropology. Rosaldo, M.Z. (1980) Knowledge and Passion: Ilongot Notions of Self and Social Life. Cambridge: Cambridge University Press. On the wider context of emotional citizenship in the late 1970s and early 1980s see Berlant, L. (2008) The Female Complaint. Durham, NC: Duke University Press.

46 Sroufe, L.A. (1979) Socioemotional development. In J. Osofsky (ed.) Handbook of Infant Development (pp.462–515). New York: John Wiley, p.462.

47 Ibid. p.502; Sroufe, L.A. & Wunsch, J.P. (1972) The development of laughter in the first year of life. Child Development, 43(4), 1326–44; Sroufe, L.A. & Waters, E. (1976) The ontogenesis of smiling and laughter: a perspective on the organization of development in infancy. Psychological Review, 83(3), 173–89.

48 In general, though without full consistency, Sroufe used the term ‘affect’ to refer to the embodied reaction to a salient event, and ‘emotion’ to refer to the subjective experience of this reaction, including awareness of its source. An affect seems to have more of the early, embodied prototype in play, whereas an emotion seems to be more inflected by contemporaneous cognitive resources. However, this contrast is only elaborated implicitly by Sroufe.

49 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.144.

50 Ibid. p.206.

51 Sroufe, L.A. (1979) Socioemotional development. In J. Osofsky (ed.) Handbook of Infant Development (pp.462–515). New York: John Wiley, p.491.

52 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.199.

53 Ibid. p.126.

54 Sroufe, L.A. (1979) Socioemotional development. In J. Osofsky (ed.) Handbook of Infant Development (pp.462–515). New York: John Wiley, p.478; Escalona, S.K. (1969) The Roots of Individuality: Normal Patterns of Development In Infancy. Oxford: Aldine.

55 Sroufe, L.A. (1989) Relationships, self, and individual adaptation. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood: A Developmental Approach (pp.70–94). New York: Basic Books, p.76. The term ‘emotion regulation’ remains undefined in the writings of Sroufe and Egeland. However, in practice their usage resembles Thompson, R.A., Flood, M.F., & Lundquist, L. (1995) Emotional regulation and developmental psychopathology. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology, Volume 6: Emotion, Cognition, and Representation (pp.261–99). Rochester, NY: University of Rochester Press: ‘emotional regulation consists of the extrinsic and intrinsic processes responsible for monitoring, evaluating and modifying emotional reactions, especially their intensive and temporal features, to accomplish ones goals … this definition of emotion regulation includes maintaining and enhancing emotional arousal as well as inhibiting it’ (265).

56 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.64–5.

57 Sroufe, L.A. (1979) Socioemotional development. In J. Osofsky (ed.) Handbook of Infant Development, (pp.462–515). New York: John Wiley, p.488; see also Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.55.

58 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.11.

59 See also Sroufe, L.A., Egeland, B., & Kreutzer, T. (1990) The fate of early experience following developmental change: longitudinal approaches to individual adaptation in childhood. Child Development, 61(5), 1363–73.

60 Sroufe, L.A. (2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development, 7(4), 349–67, p.353.

61 Sroufe, L.A. (1979) Socioemotional development. In J. Osofsky (ed.), Handbook of Infant Development (pp.462–515). New York: John Wiley, p.488.

62 Sroufe, L.A. (1989) Relationships, self, and individual adaptation. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood: A Developmental Approach (pp.70–94). New York: Basic Books, p.83.

63 Kagan, J. (1982) Psychological Research on the Human Infant: An Evaluative Summary. New York: W.T. Grant Foundation. See also Kagan, J. (1995) On attachment. Harvard Review of Psychiatry, 3(2), 104–106; Kagan, J. (1996) Three pleasing ideas. American Psychologist, 51(9), 901–908. On Kagan’s role as a ‘spokesperson’ for critics of attachment research see Karen, R. (1998) Becoming Attached: First Relationships and How They Shape Our Capacity to Love. Oxford: Oxford University Press, p.253.

64 Sroufe, L.A. (1985) Attachment classification from the perspective of infant–caregiver relationships and infant temperament. Child Development, 56(1), 1–14.

65 Behrens, K.Y. (2016) Reconsidering attachment in context of culture: review of attachment studies in Japan. Online Readings in Psychology and Culture, 6(1), 7, p.25. See Vaughn, B.E. & Bost, K.K. (2016) Attachment and temperament as intersecting developmental products and interacting developmental contexts throughout infancy and childhood. In J. Cassidy & P.R. Shaver (eds) Handbook of Attachment: Theory, Research, and Clinical Applications, 3rd edn (pp.202–22). New York: Guilford.

66 Van IJzendoorn, M.H. & Bakermans-Kranenburg, M.J. (2012) Integrating temperament and attachment. The differential susceptibility paradigm. In M. Zentner & R.L. Shiner (eds) Handbook of Temperament (pp.403–24). New York: Guilford, p.409.

67 Karen, R. (1998) Becoming Attached: First Relationships and How They Shape Our Capacity to Love. Oxford: Oxford University Press, p.256.

68 Warren, S.L., Emde, R.N., & Sroufe, L.A. (2000) Internal representations: predicting anxiety from children’s play narratives. Journal of the American Academy of Child & Adolescent Psychiatry, 39(1), 100–107: ‘The variance accounted for by our variables in the prediction from 5-year-old narratives to 6-year-old internalizing and anxiety symptoms was substantial for this kind of research (20%), [and] there is a strong likelihood that genetic predisposition and temperament account for much of the remaining variance’ (106).

69 Sroufe, L.A. (2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development, 7(4), 349–67. These results were earlier reported in Susman-Stillman, A., Kalkoske, M., Egeland, B., & Waldman, I. (1996) Infant temperament and maternal sensitivity as predictors of attachment security. Infant Behavior and Development, 19(1), 33–47.

70 Groh, A.M., Narayan, A.J., Bakermans-Kranenburg, M.J., et al. (2017) Attachment and temperament in the early life course: a meta-analytic review. Child Development, 88(3), 770–95.

71 Spangler, G. (2013) Individual dispositions as precursors of differences in attachment quality: why maternal sensitivity is nevertheless important. Attachment & Human Development, 15(5–6), 657–72.

72 Bowlby, J. (1960) Separation anxiety. International Journal of Psycho-Analysis, 41, 89–113, pp.93–4. In fact, use of the term ‘comfort’ to describe the set-goal of the attachment system in infancy led a subterranean existence over decades. For example, the concept of comfort can be seen dodging in and out of view in Separation: ‘accessibility in itself is not enough. Not only must an attachment figure be accessible but he, or she, must be willing to respond in an appropriate way; in regard to someone who is afraid this means willingness to act as comforter and protector. Only when an attachment figure is both accessible and potentially responsive can he, or she, be said to be truly available. In what follows, therefore, the word “available” is to be understood as implying that an attachment figure is both accessible and responsive.’ Bowlby, J. (1973) Separation. New York: Basic Books, p.234. Following Bowlby, Ainsworth tended to use ‘availability’ in descriptions of the set-goal in infancy in written presentations. However, in oral presentations she tended to use the term ‘comfort’.

73 Bischof, N. (1975) A systems approach toward the functional connections of attachment and fear. Child Development, 46(4), 801–17, p.802.

74 Sroufe, L.A. & Waters, E. (1977) Attachment as an organizational construct. Child Development, 48(4), 1184–99.

75 Ibid. p.1191.

76 Sroufe, L.A. (1979) The coherence of individual development: early care, attachment, and subsequent developmental issues. American Psychologist, 34(10), 834–41, p.837.

77 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.147.

78 E.g. Cicchetti, D. & Pogge-Hesse, P. (1981) The relation between emotion and cognition in infant development. In M. Lamb & L. Sherrod (eds) Infant Social Cognition (pp.205–72). Hillsdale, NJ: Lawrence Erlbaum; Erickson, M.F., Korfmacher, J., & Egeland, B. (1992) Attachments past and present: implications for therapeutic intervention with mother–infant dyads. Development & Psychopathology, 4(4), 495–507, p.496. In fact, Erickson and colleagues appear to have regarded Bowlby’s behavioural system model as superseded already by his own notion of internal working models.

79 Cummings, M.E. & El-Sheikh, M. (1986) An Organizational Scheme for the Classification of Attachments on a Continuum of Felt-Security. Morgantown: West Virginia University. https://files.eric.ed.gov/fulltext/ED288653.pdf.

80 Ainsworth, M. (1987) Letter to John Bowlby, 28 November 1987. PP/BOW/B.3/8.

81 Ainsworth, M. (1987) Letter to John Bowlby, 30 December 1987. PP/BOW/B.3/8.

82 Ainsworth, M.D.S. (1990) Some considerations regarding theory and assessment relevant to attachments beyond infancy. In M.T. Greenberg, D. Cicchetti, & E.M. Cummings (eds) Attachment in the Preschool Years: Theory, Research, and Intervention (pp.463–88). Chicago: University of Chicago Press.

83 See e.g. Kerns, K.A. & Brumariu, L.E. (2016) Attachment in middle childhood. In J. Cassidy & P. Shaver (eds) Handbook of Attachment, 3rd edn (pp.349–65). New York: Guilford.

84 In the first and second editions of Cassidy and Shaver’s Handbook of Attachment, a chapter by Kobak attempted to adjudicate the issue in favour of ‘availability’ as the set-goal of the attachment behavioural system. This section was removed from the chapter for the third edition. Besides Kobak’s chapter, there has been little attempt to adjudicate the issue, except where it has caused problems on the borders between the developmental and the social psychology attachment traditions (Chapter 5). An early exception is Greenberg, M.T., Siegel, J.M., & Leitch, C.J. (1983) The nature and importance of attachment relationships to parents and peers during adolescence. Journal of Youth and Aolescence, 12(5), 373–86. Greenberg and colleagues retain proximity and felt security as two different set-goals of the attachment system, one or both of which may be met. For instance, the ambivalent/resistant infant was conceptualised as achieving proximity but not felt security.

85 This criticism of Sroufe and Waters has recently been made again in a book chapter collectively written by a group of anthropologists and developmental psychologists: Gaskins, S., Beeghly, M., Bard, K.A., et al. (2017) Meaning and methods in the study and assessment of attachment. In H. Keller & K.A. Bard (eds) Contextualizing Attachment: The Cultural Nature of Attachment (pp.321–33). Cambridge, MA: MIT Press.

86 Kobak, R. (1994) Adult attachment: a personality or relationship construct? Psychological Inquiry, 5(1), 42–4; Kobak, R. (1999) The emotional dynamics of disruptions in attachment relationships: implications for theory, research, and clinical intervention In J. Cassidy & P.R. Shaver (eds) Handbook of Attachment: Theory, Research, and Application (pp.21 – 43). New York: Guilford, p.31.

87 Kobak, R., Zajac, K., & Madsen, S.D. (2016) Attachment disruptions, reparative processes, and psychopathology: theoretical and clinical implications. In P.R. Shaver & J. Cassidy (eds) Handbook of Attachment, 3rd edn (pp.25–39). New York: Guilford.

88 Sroufe, L.A., ???Schork, E., Motti, F., Lawroski, N., & LaFreniere, P. (1984) The role of affect in social competence. In C. E. Izard, J. Kagan, & R. B. Zajonc (Eds.), Emotion, cognition, and behavior (pp. 289-319). Cambridge: Cambridge University Press. Mary Jo Ward later recalled this period: ‘My fondest memories of Alan are from the year of the preschool project. He was in his element: choosing just the right subjects to compose a class, working daily with some very gifted teachers, supervising grad students, planning assessments and coming up with methods to capture the richness of what we were observing each day. But he was most happy about getting to know each of the precious children. His insights into each child’s coping, his joy in their accomplishments, his concern for their trials gave us an amazing opportunity to experience the human side of developmental science. It’s an example that we’ll never forget.’ Ward, M.J. (2009) Tribute to L. Alan Sroufe on the Occasion of his Retirement, 17 October 2009, Minneapolis. Unpublished manuscript shared by the author.

89 Sroufe, L.A., Schork, E., Motti, E., Lawroski, N., & LaFreniere, P. (1984) The role of affect in social competence. In C. Izard, J. Kagan, & R. Zajonc (eds) Emotion, Cognition and Behavior (pp.289–319). New York: Plenum, p.290.

90 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press: ‘Emotion is part of all critical transactions with the environment. It guides, directs, and sometimes disrupts action. And it is the currency of personal relationships’ (12).

91 Sroufe, L.A., Schork, E., Motti, E., Lawroski, N., & LaFreniere, P. (1984) The role of affect in social competence. In C. Izard, J. Kagan, & R. Zajonc (eds) Emotion, Cognition and Behavior (pp.289–319). New York: Plenum, p.298–9.

92 Ibid. p.305.

93 Sroufe, L.A. & Fleeson, J. (1986) Attachment and the construction of relationships. In W. Hartup & Z. Rubin (eds) Relationships and Development (pp.51–71). Hillsdale, NJ: Lawrence Erlbaum.

94 Sroufe, L.A. & Fleeson, J. (1988) The coherence of family relationships. In R.A. Hinde & J. Stevenson-Hinde (eds) Relationships within Families: Mutual Influences (pp.27–47). Oxford: Oxford University Press.

95 See also Sroufe, L.A., Cooper, R.G., DeHart, G.B., & Marshall, M.E. (1996) Child Development: Its Nature and Course, 3rd edn. New York: McGraw-Hill, p.381.

96 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.221.

97 Sroufe, L.A. (1989) Relationships and relationship disturbances. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood (pp.97–124). New York: Basic Books, p.107.

98 Sroufe, J.W. (1991) Assessment of parent–adolescent relationships: implications for adolescent development. Journal of Family Psychology, 5(1), 21–45.

99 Ibid. p.34. Reflecting on Sroufe and Fleeson’s work in his private notes, Bowlby considered it a brilliant contribution, showing that ‘it looks as though “self” is one pole of a relationship and so cannot be conceptualised except in terms of a relationship’. Bowlby, J. (1983) Concept of self . PP/Bow/H.8.

100 Sroufe, L.A. & Ward, M.J. (1980) Seductive behavior of mothers of toddlers: occurrence, correlates, and family origins. Child Development, 51(4), 1222–9.

101 Ibid. p.225.

102 At age 13, the relationship between mother and daughter was also more peer-like than the rest of the sample, and the mothers displayed more child-like and needy behaviours towards their daughters: Sroufe, L.A. (1989) Relationships and relationship disturbances. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood (pp.97–124). New York: Basic Books, pp.77–8.

103 Sroufe, L.A., Jacobvitz, D., Mangelsdorf, S., DeAngelo, E., & Ward, M.J. (1985) Generational boundary dissolution between mothers and their preschool children: a relationship systems approach. Child Development, 56(2), 317–25, p.322.

104 Sroufe, L.A. (1989) Relationships and relationship disturbances. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood (pp.97–124). New York: Basic Books, p.103.

105 Sroufe, L.A., Jacobvitz, D., Mangelsdorf, S., DeAngelo, E., & Ward, M.J. (1985) Generational boundary dissolution between mothers and their preschool children: a relationship systems approach. Child Development, 56(2), 317–25, p.324.

106 Additionally, mothers with substance abusing partners were more likely to show peer-like or spouse-like behaviour with their 13-year-old sons; they were also more likely to have daughters who showed caregiving behaviour towards them. Hiester, M. (1993) Generational boundary dissolution between mothers and children in early childhood and early adolescence. Unpublished doctoral dissertation, University of Minnesota.

107 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.194.

108 Nelson, N.N. (1994) Predicting adolescent behavior problems in late adolescence from parent–child interactions in early adolescence. Unpublished doctoral dissertation, University of Minnesota; Shaffer, A. & Sroufe, L.A. (2005) The developmental and adaptational implications of generational boundary dissolution: findings from a prospective, longitudinal study. Journal of Emotional Abuse, 5(2–3), 67–84.

109 If intimately intrusive behaviours by parents towards their child had been framed as child abuse, then researchers would have been obliged to report the families to child protective services. This may have been a disincentive to include measure of such behaviours within a research study, given the ethical and administrative issues it would have raised.

110 The research on boundary-dissolving behaviour at Minnesota also likely influenced the development of a scale for role-reversing behaviour in Main and Hesse’s work on the Adult Attachment Interview (AAI).

111 Egeland, B. & Sroufe, L.A. (1981) Attachment and early maltreatment. Child Development, 52(1), 44–52. According to a later report, using slightly different criteria to Egeland and Sroufe’s 1981 paper, 47 children were abused and/or neglected by their caregivers in infancy (of 211 with adequate data; 22%), 66 in early childhood and in middle childhood (of 185 in early childhood and 190 in middle childhood; 35–36%), and 21 in adolescence (of 179; 12%). Johnson, W.F., Huelsnitz, C.O., Carlson, E.A., et al. (2017) Childhood abuse and neglect and physical health at midlife: prospective, longitudinal evidence. Development & Psychopathology, 29(5), 1935–46. The reasons for the change in definition of infant maltreatment from those used by Egeland and Sroufe are not fully provided by these later researchers.

112 Egeland, B. & Sroufe, L.A. (1981) Attachment and early maltreatment. Child Development, 52(1), 44–52. Of studies using the Ainsworth classifications, there has been a very substantial association between maltreatment and insecure attachment. Baer, J.C. & Martinez, C.D. (2006) Child maltreatment and insecure attachment: a meta-analysis, Journal of Reproductive and Infant Psychology, 24(3), 187–97.

113 This is only somewhat higher than the 17% reported in a recent meta-analysis: Verhage, M.L., Schuengel, C., Madigan, S., et al. (2016) Narrowing the transmission gap: a synthesis of three decades of research on intergenerational transmission of attachment. Psychological Bulletin, 142(4), 337–66, Table 4.

114 Warren, S.L., Huston, L., Egeland, B., & Sroufe, L.A. (1997) Child and adolescent anxiety disorders and early attachment. Journal of the American Academy of Child & Adolescent Psychiatry, 36(5), 637–44, p.640. See also Waters, E., Vaughn, B.E., & Egeland, B.R. (1980) Individual differences in infant–mother attachment relationships at age one: antecedents in neonatal behavior in an urban, economically disadvantaged sample. Child Development, 51(1), 208–16.

115 Groh, A.M., Narayan, A.J., Bakermans-Kranenburg, M.J., et al. (2017) Attachment and temperament in the early life course: a meta-analytic review. Child Development, 88(3), 770–95.

116 Egeland, B. & Sroufe, L.A. (1981) Attachment and early maltreatment. Child Development, 52(1), 44–52, p.50. The same factors would also, however, have reduced the caregivers’ attentional availability to their infant, increasing the relevance of a strategy that increased attachment signals in order to have their caregiver heed and respond to their desire for felt security. See Chapter 2 for criticism of the attribution of ‘inconsistent care’ as the ultimate cause of ambivalent/resistant attachment behaviour.

117 A relationship between parental drug use and ambivalent/resistant attachment was also later documented by Seifer, R., LaGasse, L.L., Lester, B., et al. (2004) Attachment status in children prenatally exposed to cocaine and other substances. Child Development, 75(3), 850–68.

118 Alan Sroufe and Byron Egeland, personal communication, July 2012.

119 Egeland, B. & Sroufe, L.A. (1981) Developmental sequelae of maltreatment in infancy. New Directions for Child and Adolescent Development, 1981(11), 77–92, p.84.

120 Egeland, B. & Farber, E.A. (1984) Infant–mother attachment: factors related to its development and changes over time. Child Development, 55(3), 753–71.

121 Sroufe, L.A. & Waters, E. (1982) Issues of temperament and attachment. American Journal of Orthopsychiatry, 52(4), 743–6, p.744.

122 Egeland, B. & Farber, E.A. (1984) Infant–mother attachment: factors related to its development and changes over time. Child Development, 55(3), 753–71, p.767.

123 Egeland, B. & Brunnquell, D. (1979) An at-risk approach to the study of child abuse: some preliminary findings. Journal of the American Academy of Child Psychiatry, 18(2), 219–35.

124 Bowlby, J. (1971) Evidence presented on 6/4/1971 to the Committee on the Adoption of Children, Home Office and Department of Health and Social Security, The National Archives, Kew, BN 29/2340.

125 Rohner, R.P. & Rohner, E.C. (1980) Antecedents and consequences of parental rejection: a theory of emotional abuse. Child Abuse & Neglect, 4(3), 189–98; Trowell, J. (1983) Emotional abuse of children. Health Visitor, 56(7), 252–5.

126 Cicchetti, D. & Manly, J.T. (1990) A personal perspective on conducting research with maltreating families: problems and solutions. In G. Brody & I. Sigel (eds) Methods of Family Research: Families at Risk, Vol. 2 (pp.87–133). Hillsdale, NJ: Lawrence Erlbaum. More recently, see Bilson, A., Featherstone, B., & Martin, K. (2017) How child protection’s ‘investigative turn’ impacts on poor and deprived communities. Family Law, 47, 316–19.

127 Sroufe, L.A. (1985) Attachment classification from the perspective of infant–caregiver relationships and infant temperament. Child Development 56(1), 1–14. An early report on findings from this analysis was covered in the New York Times:Brody, J.E. (1983) Emotional deprivation seen as devastating form of child abuse. New York Times, 20 December 1983. This was unusual: over the decades there has been relatively little press coverage of the Minnesota Longitudinal Study of Risk and Adaptation, as compared with other large longitudinal studies. Egeland and Sroufe did not much seek out media appearances.

128 Sroufe, L.A. (1985) Attachment classification from the perspective of infant–caregiver relationships and infant temperament. Child Development 56(1), 1–14, p.9.

129 Egeland, B. (1997) Mediators of the effects of child maltreatment on developmental adaptation in adolescence. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology, Volume VIII: The Effects of Trauma on the Developmental Process (pp.403–34). Rochester, NY: University of Rochester Press, Table 5.

130 Carlson, E.A. (1998) A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69(4), 1107–28, Table 4.

131 Cyr, C., Euser, E.M., Bakermans-Kranenburg, M.J., & Van Ijzendoorn, M.H. (2010) Attachment security and disorganization in maltreating and high-risk families: a series of meta-analyses. Development & Psychopathology, 22(1), 87–108, Table 4.

132 E.g. Enlow, M.B., Egeland, B., Carlson, E., Blood, E., & Wright, R.J. (2014) Mother–infant attachment and the intergenerational transmission of posttraumatic stress disorder. Development & Psychopathology, 26(1), 41–65.

133 Waters, E. & Sroufe, L.A. (1983) Social competence as a developmental construct. Developmental Review, 3, 79–97, p.81.

134 Baumrind, D. (1978) Parental disciplinary patterns and social competence in children. Youth & Society, 9(3), 239–67, p.249. See also White, R. (1959) Motivation reconsidered: the concept of competence. Psychological Review, 66, 297–333.

135 Erickson, M., Sroufe, L.A., & Egeland, B. (1985) The relationship of quality of attachment and behavior problems in preschool in a high risk sample. Monographs of the Society for Research in Child Development, 50(1–2), 147–86, p.157, 164.

136 Meuwissen, A.S. & Englund, M.M. (2016) Executive function in at-risk children: importance of father-figure support and mother parenting. Journal of Applied Developmental Psychology, 44, 72–80.

137 Sroufe, L.A. (1983) Infant–caregiver attachment and patterns of adaptation in preschool: the roots of maladaptation and competence. In M. Perlmutter (ed.) Minnesota Symposium in Child Psychology, Vol. 16 (pp.41–83). Hillsdale, NJ: Lawrence Erlbaum, p.76.

138 Sroufe, L.A., Fox, N.E., & Pancake, V.R. (1983) Attachment and dependency in developmental perspective. Child Development, 54(6), 1615–27.

139 Sroufe, L.A. (1989) Relationships, self, and individual adaptation. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood: A Developmental Approach (pp.70–94). New York: Basic Books.

140 Troy, M. & Sroufe, L.A. (1987) Victimization among preschoolers: the role of attachment relationship history. Journal of the American Academy of Child and Adolescent Psychiatry, 26(2), 166–72.

141 Fearon, R.P., Bakermans-Kranenburg, M.J., Van IJzendoorn, M.H., Lapsley, A.M., & Roisman, G.I. (2010) The significance of insecure attachment and disorganization in the development of children’s externalizing behavior: a meta-analytic study. Child Development, 81(2), 435–56. Questionnaires of parents revealed an association of d = .22. Questionnaires of teachers revealed an association of d = .30.

142 The hypothesis put forward later by Sroufe regarding the link between disorganised attachment and conduct problems could equally be applied as an alternative hypothesis in the case of avoidant attachment, given that both showed substantial prospective links to dissociative tendencies in Carlson’s 1998 report. Sroufe, L.A. (2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development, 7(4), 349–67: ‘Disorganized attachment also predicts conduct disorder, we believe, because of the dissociative tendencies and attendant problems with impulse control’ (361).

143 Renken, B., Egeland, B., Marvinney, D., Mangelsdorf, S., & Sroufe, L.A. (1989) Early childhood antecedents of aggression and passive-withdrawal in early elementary school. Journal of Personality, 57(2), 257–81.

144 Sroufe, L.A. (2007) The place of development in developmental psychopathology. In A. Masten (ed.) Multilevel Dynamics in Developmental Psychopathology: Pathways to the future. The Minnesota Symposia on Child Psychology, Vol. 34 (pp.285–99). Mahwah, NJ: Lawrence Erlbaum, p.287.

145 The matter has been examined in a meta-analytic treatment by Fearon, R.P., Bakermans-Kranenburg, M.J., Van IJzendoorn, M.H., Lapsley, A.M., & Roisman, G.I. (2010) The significance of insecure attachment and disorganization in the development of children’s externalizing behavior: a meta-analytic study. Child Development, 81(2), 435–56. An alternative/additional hypotheses for the association between avoidant attachment and externalising behaviours is the idea that when an avoidant strategy fails or is disrupted, anger may be evoked. This may be as the intrusion of a segregated system, as proposed for example by Crittenden, P.M. (2016) Raising Parents: Attachment, Representation, and Treatment, 2nd edn. London: Routledge. Mikulincer and Shaver also suggested the idea of a dominance behavioural system, which may be engaged to supplement or replace a conditional strategy (Chapter 5).

146 Rutter, M. & Sroufe, L.A. (2000) Developmental psychopathology: concepts and challenges. Development & Psychopathology, 12(3), 265–96, p.271.

147 Fearon, R.P., Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., Lapsley, A.M., & Roisman, G.I. (2010) The significance of insecure attachment and disorganization in the development of children’s externalizing behavior: a meta-analytic study. Child Development, 81(2), 435–56.

148 Sroufe, L.A. (1989) Relationships, self, and individual adaptation. In A.J. Sameroff & R.N. Emde (eds) Relationship Disturbances in Early Childhood: A Developmental Approach (pp.70–94). New York: Basic Books, p.89.

149 Fury, G., Carlson, E.A., & Sroufe, L.A. (1997) Children’s representations of attachment relationships in family drawings. Child Development, 68(6), 1154–64.

150 Marvinney, D. (1988) Sibling relationships in middle childhood: implications for social-emotional development. Unpublished doctoral dissertation, University of Minnesota. This relationship did not, however, hold when the elder sibling was themselves secondborn.

151 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.170.

152 Slade, A. & Holmes, J. (eds) (2014) Attachment Theory. London: Sage.

153 Ziv, Y. & Hotam, Y. (2015) Theory and measure in the psychological field: the case of attachment theory and the strange situation procedure. Theory & Psychology, 25(3), 274–91.

154 See Steinberg, L. & Lerner, R.M. (2004) The scientific study of adolescence: a brief history. Journal of Early Adolescence, 24(1), 45–54.

155 See e.g. Petersen, A.C. (1988) Adolescent development. Annual Review of Psychology, 39, 583–607.

156 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.92.

157 Sroufe, L.A., Bennett, C., Englund, M., Urban, J., & Shulman, S. (1993) The significance of gender boundaries in preadolescence: contemporary correlates and antecedents of boundary violation and maintenance. Child Development, 64(2), 455–66.

158 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.97.

159 Sroufe, L.A., Bennett, C., Englund, M., Urban, J., & Shulman, S. (1993) The significance of gender boundaries in preadolescence: contemporary correlates and antecedents of boundary violation and maintenance. Child Development, 64, 455–66.

160 Sroufe, L.A., Egeland, B., & Carlson, E.A. (1999) One social world: the integrated development of parent–child and peer relationships. In W. . Collins & B. Laursen (eds) Relationships as Developmental Contexts: The Minnesota Symposia on Child Psychology, Vol. 30 (pp.241–61). Mahwah, NJ: Lawrence Erlbaum, Table 11.4.

161 Collins, W.A., Hennighausen, K.H., Schmit, D.T., & Sroufe, L.A. (1997) Developmental precursors of romantic relationships: a longitudinal analysis. In S. Shulman (ed.) New Directions for Child Development. San Francisco: Jossey-Bass (pp.69–84), p.78; Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.186.

162 Ibid. p.194. Despite these intriguing results, in the two decades since the study took place, no further research has been done using the gender violations scale. It might be suspected that the amalgamation of cross-gender sociality and sexual harassment folded together phenomena that were of interest to two rather different communities of researchers; both have subsequently been considered in the literature on gender and education, but generally not together. An exception is research in qualitative feminist sociology, e.g. Renold, E. (2002) Presumed innocence: (hetero) sexual, heterosexist and homophobic harassment among primary school girls and boys. Childhood, 9(4), 415–34.

163 Sroufe, L.A. & Egeland, B. (1991) Illustrations of person and environment interaction from a longitudinal study. In T. Wachs & R. Plomin (eds) Conceptualization and Measurement of Organism–Environment Interaction (pp.68–84). Washington, DC: American Psychological Association, p.76.

164 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.97.

165 Kagan, J. (1984) The Nature of the Child. New York: Basic Books; Jerome Kagan interviewed by Robert Karen, 1 January 1989, cited in Karen, R. (1998) Becoming Attached: First Relationships and How They Shape Our Capacity to Love. Oxford: Oxford University Press: ‘My view is, if you’re attached, you are motivated to adopt the values of your parents. If your parent values autonomy, you’ll be autonomous; if your parent values dependency, you’ll be dependent’ (151).

166 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family–Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.100.

167 Englund, M.M., Levy, A.K., Hyson, D.M., & Sroufe, L.A. (2000) Adolescent social competence: effectiveness in a group setting. Child Development, 71(4), 1049–60.

168 Sroufe, L.A., Egeland, B., & Carlson, E. (1999) One social world: the integrated development of parent–child and peer relationships. In W.A. Collins & B. Laursen (eds) Relationships as Developmental Context: The 30th Minnesota Symposium on Child Psychology (pp.241–62). Hillsdale, NJ: Lawrence Erlbaum, 241–62, p.256.

169 Ibid. p.258.

170 Englund, M.M., Levy, A.K., Hyson, D.M., & Sroufe, L.A. (2000) Adolescent social competence: effectiveness in a group setting. Child Development, 71(4), 1049–60, p.1056.

171 Collins, A. & Sroufe, L.A. (1999) Capacity for intimate relationships: a developmental construction. In W. Furman, B.B. Brown, & C. Feiring (eds) (1999) The Development of Romantic Relationships in Adolescence (pp.125–47). New York: Cambridge University Press, p.135.

172 Roisman, G.I., Madsen, S.D., Hennighausen, K.H., Sroufe, L.A., & Andrew Collins, W. (2001) The coherence of dyadic behavior across parent–child and romantic relationships as mediated by the internalized representation of experience. Attachment & Human Development, 3(2), 156–72.

173 In historical perspective, Simpson’s appointment can be regarded as an important moment for the social psychological tradition (Chapter 5) which has held substantial sway at Minnesota since the retirement of Egeland and Sroufe. Subsequently, the leaders of the Minnesota Longitudinal Study of Risk and Adaptation have been Jeff Simpson—a social psychologist—and Glenn Roisman, whose time with Chris Fraley at Illinois was influential for his attitude towards attachment methodology and theory (Chapter 3).

174 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.203.

175 Salvatore, J.E., Kuo, S.I.C., Steele, R.D., Simpson, J.A., & Collins, W.A. (2011) Recovering from conflict in romantic relationships: a developmental perspective. Psychological Science, 22(3), 376–83.

176 Hennighausen, K.H. (1999) Developmental antecedents of young adult romantic relationships. Unpublished doctoral dissertation, University of Minnesota.

177 See e.g. Madsen, S.D. & Collins, W.A. (2011) The salience of adolescent romantic experiences for romantic relationship qualities in young adulthood. Journal of Research on Adolescence, 21(4), 789–801; Labella, M.H., Johnson, W.F., Martin, J., et al. (2018) Multiple dimensions of childhood abuse and neglect prospectively predict poorer adult romantic functioning. Personality and Social Psychology Bulletin, 44(2), 238–51.

178 Simpson, J.A., Collins, W.A., Tran, S., & Haydon, K.C. (2007) Attachment and the experience and expression of emotions in romantic relationships: a developmental perspective. Journal of Personality and Social Psychology, 92(2), 355–67. The mediational account has, however, later been criticised by Gillath and colleagues. Reviewing findings from other cohort studies, they argued that the model presented by Simpson and colleagues underestimates the extent to which early attachment experiences provide scaffolding for the quality of peer relationships. The quality of peer relationships is therefore not an independent intervening variable. Gillath, O., Karantzas, G.C., & Fraley, R.C. (2016) Adult Attachment: A Concise Introduction to Theory and Research. London: Academic Press, p.69.

179 Van Ryzin, M.J., Carlson, E.A., & Sroufe, L.A (2011) Attachment discontinuity in a high-risk sample. Attachment & Human Development, 13(4), 381–401, p.397.

180 In a later meta-analysis based on 80 independent samples (N = 4441), Groh and colleagues found a robust association between attachment security with mother and peer competence (d = .39). The association was of the same strength no matter how long after the attachment assessment peer competence was measured: there was no moderation by age, suggesting effects were enduring. Groh, A.M., Fearon, R.P., Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., Steele, R.D., & Roisman, G.I. (2014) The significance of attachment security for children’s social competence with peers: a meta-analytic study. Attachment & Human Development, 16(2), 103–136, p.128.

182 Carlson, E.A. (1998) A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69(4), 1107–28.

183 Kobak, R. & Bosmans, G. (2018) Attachment and psychopathology: a dynamic model of the insecure cycle, Current Opinion in Psychology, 25, 76–80, p.76.

184 Groh, A.M., Fearon, R.P., van IJzendoorn, M.H., Bakermans-Kranenburg, M.J., & Roisman, G.I. (2017) Attachment in the early life course: meta-analytic evidence for its role in socioemotional development. Child Development Perspectives, 11(1), 70–76.

185 Betty Carlson, personal communication, May 2019.

186 Though not discussed in the paper, in Table 5 Carlson presented findings from teacher report of participants suggesting that disorganised attachment was associated weakly with internalising problems in elementary and high school (r = .19, .18), and had no association with externalising problems. Carlson, E.A. (1998) A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69(4), 1107–28, Table 5. The findings may be the product of chance. However, it is conceivable that the distinctive predominance of child neglect in the Minnesota sample compared to other studies may have played a role in these results running contrary to later meta-analyses of samples.

187 See e.g. Jacobvitz, D., Hazen, N., Zaccagnino, M., Messina, S., & Beverung, L. (2011) Frightening maternal behavior, infant disorganization, and risks for psychopathology. In D. Cicchetti & G.I. Roisman (eds) The Origins and Organization of Adaptation and Maladaptation: The Minnesota Symposia on Child Psychology, Vol. 36 (pp.283–322). Hoboken, NJ: Wiley.

188 Carlson, E.A., Egeland, B., & Sroufe, L.A. (2009) A prospective investigation of the development of borderline personality symptoms. Development & Psychopathology, 21(4), 1311–34. A later study by Lyons-Ruth and colleagues found only a marginal association between infant disorganised attachment and later self-injurious behaviours, but there was a marked relationship between maternal withdrawing behaviours in response to child distress, forms of disorganised attachment that entailed approaching the caregiver, and suicidality in early adulthood. Lyons-Ruth, K., Bureau, J.F., Holmes, B., Easterbrooks, A., & Brooks, N.H. (2013) Borderline symptoms and suicidality/self-injury in late adolescence: prospectively observed relationship correlates in infancy and childhood. Psychiatry Research, 206(2–3), 273–81.

189 Warren, S.L., Huston, L., Egeland, B., & Sroufe, L.A. (1997) Child and adolescent anxiety disorders and early attachment. Journal of the American Academy of Child & Adolescent Psychiatry, 36(5), 637–44.

190 Sroufe, L.A. (2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development, 7(4), 349–67, p.361.

191 Groh, A.M., Roisman, G.I., van IJzendoorn, M.H., Bakermans-Kranenburg, M.J., & Fearon, R.P. (2012) The significance of insecure and disorganized attachment for children’s internalizing symptoms: a meta-analytic study. Child Development, 83(2), 591–610; Madigan, S., Atkinson, L., Laurin, K., & Benoit, D. (2013) Attachment and internalizing behavior in early childhood: a meta-analysis. Developmental Psychology, 49(4), 672–89. In contrast, a positive association (r = .37) between resistant attachment and anxiety has been reported when self-report measures of anxiety are included in the meta-analysis: Colonnesi, C., Draijer, E.M., Jan, J.M., et al. (2011) The relation between insecure attachment and child anxiety: a meta-analytic review. Journal of Clinical Child and Adolescent Psychology, 40, 630–45. This would suggest that one of the distinct sequelae of resistant attachment is participant report of anxiety symptoms on self-report measures, perhaps as a consequence of a maximising strategy (Chapter 5).

192 E.g. Cicchetti, D. & Schneider-Rosen, K. (1986) An organizational approach to childhood depression. In M. Rutter, C. Izard, & P. Read (eds) Depression in Young People: Clinical and Developmental Perspectives (pp.71–134). New York: Guilford.

193 Duggal, S., Carlson, E.A., Sroufe, L.A., & Egeland, B. (2001) Depressive symptomatology in childhood and adolescence. Development & Psychopathology, 13(1), 143–64. The finding that insecure attachment does not predict later depression after taking into account maternal social support ran too far contrary to expectations, and has generally been ignored by subsequent researchers who have continued to treat insecure childhood attachment as a cause of depression in adult caregiers. See e.g. Toth, S.L., Rogosch, F.A., & Cicchetti, D. (2008) Attachment-theory informed intervention and reflective functioning in depressed mothers. In H. Steele & M. Steele (eds) The Adult Attachment Interview in Clinical Context (pp.154–72). New York: Guilford: ‘Retrospective studies have found that depressed adults report histories replete with inadequate or abusive care. Thus insecure childhood attachment relationships may contribute to mothers’ depression’ (155). It is possible, however, that there are two different diagnostic entities in play. Duggal and colleagues were reporting on depressive symptoms; Toth and colleagues were discussing major depressive disorder.

194 Sroufe, L.A., Coffino, B., & Carlson, E.A. (2010) Conceptualizing the role of early experience: lessons from the Minnesota longitudinal study. Developmental Review, 30(1), 36–51, p.45.

195 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W. A. (2005) The Development of the Person. New York: Guilford, p.249.

196 Enlow, M.B., Egeland, B., Carlson, E., Blood, E., & Wright, R.J. (2014) Mother–infant attachment and the intergenerational transmission of posttraumatic stress disorder. Development & Psychopathology, 26(01), 41–65.

197 Ibid. p.54.

198 Sroufe, L.A., Carlson, E.A., Levy, A.K., & Egeland, B. (1999) Implications of attachment theory for developmental psychopathology. Development & Psychopathology, 11(1), 1–13, p.8.

199 Jimerson, S., Egeland, B., Sroufe, L.A., & Carlson, B. (2000) A prospective longitudinal study of high school dropouts examining multiple predictors across development. Journal of School Psychology, 38(6), 525–49, p.544.

200 Raby, K.L., Roisman, G.I., Fraley, R.C., & Simpson, J.A. (2015) The enduring predictive significance of early maternal sensitivity: social and academic competence through age 32 years. Child Development, 86(3), 695–708.

201 Englund, M.M., Egeland, B., & Collins, W.A. (2008) Exceptions to high school dropout predictions in a low-income sample: do adults make a difference? Journal of Social Issues, 64(1), 77–94.

202 Bowlby, J. (1969, 1982) Attachment. London: Penguin, p.207. The potential for school to serve as a secure base has been elaborated by Geddes, H. (2006) Attachment in the Classroom: A Practical Guide for Schools. London: Worth Publishing.

203 For a review see Pietromonaco, P.R. & Beck, L.A. (2019) Adult attachment and physical health. Current Opinion in Psychology, 25, 115–20.

204 Cassidy, J. (2015) Early relationships, later functioning: why and how a secure base matters. Keynote address presented at 7th International Attachment Conference, New York; Ehrlich, K.B. & Cassidy, J. (2019) Attachment and physical health: introduction to the special issue. Attachment & Human Development, 21(1), 1–4.

205 Puig, J., Englund, M.M., Simpson, J.A., & Collins, W.A. (2013) Predicting adult physical illness from infant attachment: a prospective longitudinal study. Health Psychology, 32(4), 409–417.

206 Farrell, A.K., Simpson, J.A., Carlson, E.A., Englund, M.M., & Sung, S. (2017) The impact of stress at different life stages on physical health and the buffering effects of maternal sensitivity. Health Psychology, 36(1), 35–44; see also Farrell, A.K., Waters, T.E.A., Young, E.S., et al. (2019) Early maternal sensitivity, attachment security in young adulthood, and cardiometabolic risk at midlife. Attachment & Human Development, 21(1), 70–86.

207 Johnson, W.F., Huelsnitz, C.O., Carlson, E.A., et al. (2017) Childhood abuse and neglect and physical health at midlife: prospective, longitudinal evidence. Development & Psychopathology, 29(5), 1935–46.

208 Pietromonaco, P.R. & Beck, L.A. (2019) Adult attachment and physical health. Current Opinion in Psychology, 25, 115–20.

209 Ehrlich, K.B. & Cassidy, J. (2019) Attachment and physical health: introduction to the special issue. Attachment & Human Development, 21(1), 1–4.

210 E.g. Windhorst, D.A., Mileva-Seitz, V.R., Linting, M., et al. (2015) Differential susceptibility in a developmental perspective: DRD4 and maternal sensitivity predicting externalizing behavior. Developmental Psychobiology, 57(1), 35–49.

211 Raby, K.L., Steele, R.D., Carlson, E.A., & Sroufe, L.A. (2015) Continuities and changes in infant attachment patterns across two generations. Attachment & Human Development, 17(4), 414–28.

212 Sroufe, L.A. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 17–29, p.20. Holism is listed top of the list of principles governing any true developmental perspective; Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.8. On the history of holism within psychology see Shelley, C. (2008) Jan Smuts and personality theory: the problem of holism in psychology. In R. Diriwächter & J. Valsiner (eds) Striving for the Whole: Creating Theoretical Syntheses (pp.89–109). New Brunswick: Transaction Publishers.

213 Sroufe, L.A. (1990) An organizational perspective on the self. In D. Cicchetti & M. Beeghly (eds) The Self in Transition: Infancy to Childhood (pp.281–307). Chicago: University of Chicago Press: ‘The self should be conceived as an inner organisation of attitudes, feelings, expectations and meanings, which arises itself from an organised caregiving matrix (a dyadic organisation that exists prior to the emergence of the self) and which has organisational significance for ongoing adaptation and experience. The self is organisation. It arises from organisation. It influences ongoing organisation of experience’ (281).

214 Ibid. p.303.

215 Bowlby, J. (1962) Defences that follow loss: causation and function. PP/Bow/D.3/78.

216 Egeland, B. & Susman-Stillman, A. (1996) Dissociation as a mediator of child abuse across generations. Child Abuse & Neglect, 20(11), 1123–32, p.1129. The work by Main and Hesse offering the most sustained attention to these issues remained published only in Italian (Chapter 3).

217 Carlson, E.B. & Putnam, F.W. (1993) An update on the Dissociative Experiences Scale. Dissociation, 6, 16–27.

218 Ogawa, J.R., Sroufe, L.A., Weinfield, N.S., Carlson, E.A., & Egeland, B. (1997) Development and the fragmented self: longitudinal study of dissociative symptomatology in a nonclinical sample. Development & Psychopathology, 9(4), 855–79, p.874.

220 Ibid. p.868.

221 Ibid. p.855.

222 Liotti, G. (1992) Disorganized/disoriented attachment in the etiology of the dissociative disorders. Dissociation, 5, 196–204.

223 Carlson, E.A., Yates, T.M., & Sroufe, L.A. (2009) Development of dissociation and development of the self. In P.F. Dell & J.A. O’Neil (eds) Dissociation and the Dissociative Disorders: DSM-V and Beyond (pp.39–52). New York: Routledge: ‘The process of dissociation, which begins as a protective mechanism to promote the integrity of the self in the face of trauma, may directly threaten optimal functioning when employed routinely or pervasively’ (47).

224 Ibid. p.43. For aligned reflections by other attachment researchers see also Schuder, M. & Lyons-Ruth, K. (2004) ‘Hidden trauma’ in infancy: attachment, fearful arousal, and early dysfunction of the stress response system. In J. Osofsky (ed.) Young Children and Trauma (pp.69–104). New York: Guilford; Liotti, G. (2004) Trauma, dissociation, and disorganized attachment: three strands of a single braid. Psychotherapy: Theory, Research, Practice, Training, 41(4), 472.

225 Haltigan, J.D. & Roisman, G.I. (2015) Infant attachment insecurity and dissociative symptomatology: findings from the NICHD Study of Early Child Care and Youth Development. Infant Mental Health Journal, 36(1), 30–41.

226 Ibid. p.8.

227 Main, M. (1995) Recent studies in attachment: overview, with selected implications for clinical work. In S. Goldberg, R. Muir, & J. Kerr (eds) Attachment Theory: Social, Developmental and Clinical Perspectives (pp.407–470). Hillsdale, NJ: Analytic Press: ‘So long as direct maltreatment is not involved, many, perhaps most, are expected to have become “organised” by adulthood, being either secure, dismissing or preoccupied’ (454).

228 Masten, A.S. (2006) Developmental psychopathology: pathways to the future. International Journal of Behavioral Development, 30(1), 47–54; Cicchetti, D. (2013) An overview of developmental psychopathology. In P. Zelazo (ed.) Oxford Handbook of Developmental Psychology (pp.455–80). Oxford: Oxford University Press.

229 The movement took its name from Achenbach’s 1974 book Developmental Psychopathology (New York: Ronald Press). A former trainee of Norman Garmezy at Minnesota, in his book Achenbach argued that classifications of adult mental disorder should not be imposed backward on childhood. Achenbach called for a new paradigm thinking about mental health in a way commensurate with developmental processes. Cicchetti reviewed other precursors in Cicchetti, D. (1990) A historical perspective on the discipline of developmental psychopathology. In J. Rolf, A. Masten, D. Cicchetti, K. Nuechterlein, & S. Weintraub (eds) Risk and Protective Factors in the Development of Psychopathology (pp.2–28). New York: Cambridge University Press.

230 There were a number of other relevant meetings at the time, such as the High Risk Consortium, San Juan, Puerto Rico, 11–13 March 1980; and the 15th Annual Minnesota Symposium on Child Development, Minneapolis, 30 October–1 November 1980. See W.A. Collins (ed.) The Concept of Development: The Minnesota Symposia on Child Psychology, Vol. 15. Hillsdale, NJ: Lawrence Erlbaum. In-person meetings of the movement would be supported by subsequent annual Minnesota Symposia, and also the Rochester Symposia on Developmental Psychopathology which were initiated in 1987 by Cicchetti.

231 Other key sources of funding that supported the emergence of developmental psychopathology included the National Center on Child Abuse and Neglect, which was founded in 1974, and the National Institute of Mental Health in the 1980s.

232 Garmezy, N. (1971) Vulnerability research and the issue of primary prevention. American Journal of Orthopsychiatry, 41, 101–116; Garmezy, N. (1974) Children at risk: the search for the antecedents of schizophrenia. Part I. Conceptual models and research methods. Schizophrenia Bulletin, 1(8), 14–90; Rutter, M. & Quinton, D. (1984) Parental psychiatric disorder: effects on children. Psychological Medicine, 14(4), 853–80.

233 Rutter, M. (1979) Protective factors in children’s responses to stress and disadvantage. In M.W. Kent & J.E. Roif (eds) Primary Prevention of Psychopathology, Vol. 3: Social Competence in Children. Hanover: University Press of New England.

234 This term was introduced to psychology from ecology by Jack Block in the 1950s to refer to a quality of individual children that allows some to flexibly return to adaptive functioning after a disturbance; ‘resilience’ was then transferred to the wider literature on adversity and development by Emmy Werner in the early 1970s. Block J. & Thomas, H. (1955) Is satisfaction with self a measure of adjustment? Journal of Abnormal and Social Psychology, 51(2), 254–9; Werner, E.E., Bierman, J.M., & French, F.E. (1971) The Children of Kauai Honolulu. Hawaii: University of Hawaii Press; Werner, E. & Smith, R. (1988) Vulnerable but Invincible: A Longitudinal Study of Resilient Children and Youth. New York, Adams, Bannister & Cox; Masten, A.S. & Garmezy, N. (1985) Risk, vulnerability, and protective factors in the developmental psychopathology. In B.B. Lahey & A.E. Kazdin (eds) Advances in Clinical Child Psychology, Vol. 8 (pp.1–51). New York: Plenum.

235 Garmezy, N. & Rutter, M. (eds) (1983) Stress, Coping, and Development in Children. Baltimore: Johns Hopkins University Press. The influence of Garmezy on the emergence of developmental psychopathology in the 1980s is addressed in a festschrift from 1990: Rolf, J., Masten, A.S., Cicchetti, D., Nuechterlein, K., & Weintraub, S. (eds) (1990) Risk and Protective Factors in the Development of Psychopathology. Cambridge: Cambridge University Press. Meanwhile, another important development was the Social Science Research Council’s Committee on Social and Affective Development during Childhood, which held a symposium in 1982. Members included Michael Rutter, Carroll Izard, and Carol Dweck and the committee was chaired by Martin Hoffman. A further development in 1983 was the first meeting of the MacArthur Working Group on the study of Attachment in the Transition Period, which was concerned with the application of attachment theory and methods beyond infancy (Chapter 3). The Network met twice yearly between 1983 and 1987. Members of the network included Mary Ainsworth and several of her former students including Inge Bretherton, Jude Cassidy, Mark Cummings, Mark Greenberg, Mary Main, and Robert Marvin. The group also included other researchers who would play an important role in the subsequent development of attachment theory including Jay Belsky, Dante Cicchetti, and Susan Spieker.

236 Sameroff, A.J. (1983) Developmental systems: contexts and evolution. In P. Mussen (ed.) Handbook of Child Psychology, Vol. 1 (pp.237–94). New York: Wiley.

237 The concept of ‘promotive factors’ to describe processes that enhance the likelihood of adaptation (in contrast to protective factors, which buffer the likelihood that adversities will have an effect) was introduced by Sameroff, A.J. (1999) Ecological perspectives on developmental risk. In J.D. Osofsky & H.E. Fitzgerald (eds) WAIMH Handbook of Infant Mental Health: Vol. 4. Infant Mental Health Groups at Risk (pp.223–48). New York: Wiley. The terms ‘protective-stabilising’ and ‘protective-enhancing’ were also introduced by Luthar in making this distinction: Luthar, S.S. (1993) Annotation: methodological and conceptual issues in the study of resilience. Journal of Child Psychology and Psychiatry, 34, 441–53.

238 Garmezy later attributed the influence of both Sroufe and Rutter to his reorientation away from research on adult schizophrenia and towards the study of child and adolescent development. Masten, A.S. (1995) Interview with Norman Garmezy. SRCD oral history interview. http://srcd.org/sites/default/files/documents/garmezy_norman_interviewweb.pdf.

239 Sroufe, L.A. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.18.

240 Ibid. p.23.

241 Sroufe, L.A. (1990) Considering normal and abnormal together: the essence of developmental psychopathology. Development & Psychopathology, 2(4), 335–47: ‘Stereotypy increased in frequency and intensity during demanding tasks, was preceded by heart rate acceleration, and was followed by a pronounced heart rate deceleration … We suggested that both stereotypies and behavioural negativism were serving the purpose of modulating arousal’ (337).

242 Sroufe, L.A. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.21.

243 Caspi, A., Houts, R.M., Belsky, D.W., et al. (2014) The p factor: one general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science, 2(2), 119–37. The debt to developmental psychopathology is clear in this paper: Rutter’s work is cited as an ‘initial motivation’, and ‘developmental psychopathology’ is one of the article’s three keywords.

244 Elicker, J., Englund, M., & Sroufe, L.A. (1992) Predicting peer competence and peer relationships in childhood from early parent–child relationships. In R. Parke & G. Ladd (eds) Family-Peer Relationships: Modes of Linkage (pp.77–106). Hillsdale, NJ: Lawrence Erlbaum, p.80.

245 Sroufe, L. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.23.

246 Egeland, B., Carlson, E., & Sroufe, L.A. (1993) Resilience as process. Development & Psychopathology, 5(4), 517–28: ‘Any constitutional or environmental factors may serve as vulnerability, protective, or risk variables’ (517).

247 Rutter, M. (1983) Stress, coping and development: some issues and some questions. In N. Garmezy & M. Rutter (eds) Stress, Coping, and Development in Children (pp.1–42). Baltimore: Johns Hopkins University Press, p.23.

248 Rutter, M. (1985) Resilience in the face of adversity: protective factors and resistance to psychiatric disorder. British Journal of Psychiatry, 147(6), 598–611, p.598.

249 Rutter, M. (1986) Child psychiatry: looking 30 years ahead. Journal of Child Psychology and Psychiatry, 27(6), 803–840, p.803. See also Rutter, M.L. (1999) Psychosocial adversity and child psychopathology. British Journal of Psychiatry, 174(6), 480–93: ‘Bowlby’s 1951 monograph, although overstating the case in some respects, led to a major change in the approach to experiences during the preschool years’ (482).

250 Sroufe, L.A. (1986) Appraisal: Bowlby’s contribution to psychoanalytic theory and developmental psychology; attachment: separation: loss. Journal of Child Psychology and Psychiatry, 27(6), 841–9. Regarding the debate between Bowlby and Rutter over what priority to assign to early experience, the Minnesota group argued that no global position was acceptable, or in line with evidence from embryology. Sroufe, L.A., Coffino, B., & Carlson, E.A. (2010) Conceptualizing the role of early experience: lessons from the Minnesota longitudinal study. Developmental Review, 30(1), 36–51: ‘It is clearly not always the case that early experience is most critical … Maternal nutrition during pregnancy is a dramatic contrary example. While certain nutrients like folic acid are critically important in the early embryonic period, poor general maternal nutrition has almost no demonstrable effects in the first trimester, because the tiny developing organism can simply take many of the nutrients it needs from maternal stores. However, in the third trimester, when rapid fetal size and weight gain is occurring, adequate maternal nutrition is crucial’ (37–8).

251 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford: ‘We may think of early experience as the foundation for the building. The foundation cannot be more important than solid supporting beams or a sturdy roof; without these, the house will not last. But at the same time, a house cannot be stronger than its foundations’ (11). For a recent and comprehensive report see Raby, K. L., Roisman, G.I., Labella, M.H., Martin, J., Fraley, R.C., & Simpson, J.A. (2019) The legacy of early abuse and neglect for social and academic competence from childhood to adulthood. Child Development, 90(5), 1684–701.

252 Sroufe, L.A., Coffino, B., & Carlson, E.A. (2010) Conceptualizing the role of early experience: lessons from the Minnesota longitudinal study. Developmental Review, 30(1), 36–51, p.38.

253 Sroufe, L.A. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.19.

254 Spaulding, J. & Balch, P. (1983) A brief history of primary prevention in the twentieth century: 1908 to 1980. American Journal of Community Psychology, 11, 59–80; Cowen, E. L. (1983) Primary prevention in mental health: past present and future. In R.D. Felner, L.A. Jason, J.N. Montsugu, & S.S. Farber (eds) Prevention Psychology: Theory, Research and Practice (pp.290–97). New York: Pergammon Press.

255 E.g. Rose, N. & Lentzos, F. (2017) Making us resilient: responsible citizens for uncertain times. In S. Trnka & C. Trundle (eds) Competing Responsibilities: The Ethics and Politics of Contemporary Life (pp.27–48). Durham, NC: Duke University Press.

256 Bourbeau, P. (2018) A genealogy of resilience. International Political Sociology, 12(1), 19–35. See e.g. Howard, D.E. (1996) Searching for resilience among African-American youth exposed to community violence. Journal of Adolescent Health, 18(4), 254–62.

257 Farber, E. & Egeland, B. (1987) Abused children: can they be invulnerable? In J. Anthony & B. Cohler (eds) The Invulnerable Child. New York: Guilford, 253–88, p.286.

258 These positions can already be seen in comparing Sroufe’s chapter to others in Ciccheti, D. (1989) (ed.) The Emergence of a Discipline: Rochester Symposium on Developmental Psychopathology, Vol. 1. Hillsdale, NJ: Lawrence Erlbaum. Some of these differences would also later be remarked upon by Michael Rutter, looking back on the period: Rutter, M. (2008) Developing concepts in developmental psychopathology. In J.J. Hudziak (ed.) Developmental Psychopathology and Wellness: Genetic and Environmental Influences (pp.3–22). New York: American Psychiatric Publications.

259 Sroufe, L.A. & Egeland, B. (1991) Illustrations of person and environment interaction from a longitudinal study. In T. Wachs & R. Plomin (eds) Conceptualization and Measurement of Organism-Environment Interaction (pp.68–84). Washington, DC: American Psychological Association, p.69.

260 Kaufman, J. & Zigler, E. (1989) The intergenerational transmission of child abuse. In D. Cicchetti & V. Carlson (eds) Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect (pp.129–50). Cambridge: Cambridge University Press.

261 Egeland, B. (1997) Mediators of the effects of child maltreatment on developmental adaptation in adolescence. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology: Vol. VIII. The Effects of Trauma on the Developmental Process (pp.403–434). Rochester, NY: University of Rochester Press.

262 Berzenski, S.R., Yates, T.M., & Egeland, B. (2014) A multidimensional view of continuity in intergenerational transmission of child maltreatment. In J.E. Korbin & R.D. Krugman (eds) Handbook of Child Maltreatment, Vol. 2 (pp.115–29). New York: Springer. Another example from the Minnesota group was work by Jeff Simpson and colleagues examining the implications of early environmental unpredictability—operationalised as changes in residence, cohabitatio, and parental occupation that occur before a child is five years old. These early experiences have been found to predict later aggressive behaviour, more substance use at age 16, and more sexually risky behaviour at age 23. This is in contrast to changes in residence, cohabitation, and parental occupation after the age of five, which have barely any effect. Furthermore, the combination of these factors and harsh maternal caregiving additionally predicted adolescent risk-taking behaviours. See Simpson, J.A., Griskevicius, V., & Kuo, S.I. (2012) Evolution, stress, and sensitive periods: the influence of unpredictability in early versus late childhood on sex and risky behavior. Developmental Psychology, 48(3), 674–86; Doom, J.R., Vanzomeren-Dohm, A.A., & Simpson, J.A. (2016) Early unpredictability predicts increased adolescent externalizing behaviors and substance use: a life history perspective. Development & Psychopathology, 28(4), 1505–16.

263 Berzenski, S.R., Yates, T.M., & Egeland, B. (2014) A multidimensional view of continuity in intergenerational transmission of child maltreatment. In J.E. Korbin & R.D. Krugman (eds) Handbook of Child Maltreatment, Vol. 2 (pp.115–29). Dordrecht: Springer. For parental depression reducing the likelihood of physical abuse see Pears, K.C. & Capaldi, D.M. (2001) Intergenerational transmission of abuse: a two-generational prospective study of an at-risk sample. Child Abuse & Neglect, 25(11), 1439–61. For parental depression increasing the likelihood of sexual abuse see Leifer, M., Kilbane, T., & Kalick, S. (2004) Vulnerability or resilience to intergenerational sexual abuse: the role of maternal factors. Child Maltreatment, 9(1), 78–91.

264 Sroufe, L.A., Carlson, E.A., Levy, A.K., & Egeland, B. (1999) Implications of attachment theory for developmental psychopathology. Development & Psychopathology, 11(1), 1–13, p.6; see also Sroufe, L. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.20.

265 Yates, T.M., Dodds, M.F., Sroufe, L.A., & Egeland, B. (2003) Exposure to partner violence and child behavior problems: a prospective study controlling for child physical abuse and neglect, child cognitive ability, socioeconomic status, and life stress. Development & Psychopathology, 15(1), 199–218, p.209; Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, Table 4.2.

266 Narayan, A.J., Labella, M.H., Englund, M.M., Carlson, E.A., & Egeland, B. (2017) The legacy of early childhood violence exposure to adulthood intimate partner violence: variable- and person-oriented evidence. Journal of Family Psychology, 31(7), 833–43. On the direct and indirect pathways between exposure to domestic violence in childhood and perpetration or victimisation in adulthood see also Narayan, A.J., Englund, M.M., Carlson, E.A., & Egeland, B. (2014) Adolescent conflict as a developmental process in the prospective pathway from exposure to interparental violence to dating violence. Journal of Abnormal Child Psychology, 42(2), 239–50; Zamir, O., Szepsenwol, O., Englund, M.M., & Simpson, J.A. (2018) The role of dissociation in revictimization across the lifespan: a 32-year prospective study. Child Abuse & Neglect, 79, 144–53.

267 Collins, W.A., LaFreniere, P., & Simpson, J.A. (2011) Relationships across the lifespan: the benefits of a theoretically based longitudinal-developmental perspective. In D. Cicchetti & G.I. Roisman (eds) The Origins and Organization of Adaptation and Maladaption: The Minnesota Symposium on Child Psychology, Vol. 36 (pp.155–83). New York: Wiley, p.177.

268 Englund, M.M., Kuo, S.I.C., Puig, J., & Collins, W.A. (2011) Early roots of adult competence: the significance of close relationships from infancy to early adulthood. International Journal of Behavioral Development, 35(6), 490–96.

269 Cicchetti, D. (2013) The legacy of development and psychopathology. Development & Psychopathology, 25(4.2), 1199–200, p.1200.

270 Sroufe, L.A. (1990) Considering normal and abnormal together: the essence of developmental psychopathology. Development & Psychopathology, 2(4), 335–47, p.335.

271 Egeland, B. (1991) From data to definition. Development & Psychopathology, 3(1), 37–43.

272 Sroufe, L.A. (1997) Psychopathology as an outcome of development. Development & Psychopathology, 9(2), 251–68, p.257. See also Rutter, M. & Sroufe, L.A. (2000) Developmental psychopathology: concepts and challenges. Development & Psychopathology, 12(3), 265–96.

273 Sroufe, L.A., Carlson, E.A., Levy, A.K., & Egeland, B. (1999) Implications of attachment theory for developmental psychopathology. Development & Psychopathology, 11(1), 1–13, p.3.

274 Sroufe, L.A., Duggal, S., Weinfield, N.S., & Carlson, E. (2000) Relationships, development, and psychopathology. In M. Lewis & A. Sameroff (eds) Handbook of Developmental Psychopathology, 2nd edn (pp.75–92). New York: Kluwer Academic/Plenum Press, p.84.

275 Carlson, E.A., Jacobvitz, D., & Sroufe, L.A. (1995) A developmental investigation of inattentiveness and hyperactivity. Child Development, 66(1), 37–54, p.52.

276 E.g. Sroufe, L.A. (2007) The place of development in developmental psychopathology. In A. Masten (ed.) Multilevel Dynamics in Developmental Psychopathology: Pathways to the Future: The Minnesota Symposia on Child Psychology, Vol. 34 (pp.285–99). Mahwah, NJ: Lawrence Erlbaum, p.297.

277 Aguilar, B., Sroufe, L.A., Egeland, B., & Carlson, E. (2000) Distinguishing the early-onset/persistent and adolescence-onset antisocial behavior types: from birth to 16 years. Development & Psychopathology, 12(2), 109–132.

278 Rutter, M. & Sroufe, L.A. (2000) Developmental psychopathology: concepts and challenges. Development & Psychopathology, 12(3), 265–96, pp.273–4; Sroufe, L.A., Duggal, S., Weinfield, N., & Carlson, E. (2000) Relationships, development, and psychopathology. In A.J. Sameroff, M. Lewis, & S.M. Miller (eds) Handbook of Developmental Psychopathology (pp.75–91). New York: Kluwer Academic/Plenum Publishers, p.83.

279 Cicchetti, D. & Sroufe, L.A. (2000) Editorial: the past as prologue to the future: the times, they’ve been a’changin’. Development & Psychopathology, 12, 255–64.

280 See also Sroufe, L.A. & Jacobvitz, D. (1989) Diverging pathways, developmental transformations, multiple etiologies and the problem of continuity in development. Human Development, 32(3–4), 196–203.

281 Appleyard, K., Egeland, B., Dulmen, M.H., & Sroufe, L.A. (2005) When more is not better: the role of cumulative risk in child behavior outcomes. Journal of Child Psychology and Psychiatry, 46(3), 235–45, p.242–3. This stance would be contested by researchers such as Bakermans-Kranenburg, as we shall see.

282 Carlson, E.A., Jacobvitz, D., & Sroufe, L.A. (1995) A developmental investigation of inattentiveness and hyperactivity. Child Development, 66, 37–54; Egeland, B. (1997) Mediators of the effects of child maltreatment on developmental adaptation in adolescence. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology: Vol. 8. The Effects of Trauma on the Developmental Process (pp.403–434). Rochester, NY: University of Rochester Press. Similarly socioeconomic status has been used by other researchers conducting analyses of the dataset, e.g. Raby, K.L., Roisman, G.I., Fraley, R.C., & Simpson, J.A. (2015) The enduring predictive significance of early maternal sensitivity: social and academic competence through age 32 years. Child Development, 86(3), 695–708.

283 Yates, T.M., Egeland, B., & Sroufe, A. (2003) Rethinking resilience. In S. Luthar (ed.) Resilience and Vulnerability: Adaptation in the Context of Childhood Adversities (pp.243–66). Cambridge: Cambridge University Press: ‘Poverty is a distal risk factor whose effects are mediated by proximal risk factors such as parenting behaviors, family structure, community variables, and the broader social networks within which the child and her or his family are embedded’ (245).

284 Egeland, B. (1997) Mediators of the effects of child maltreatment on developmental adaptation in adolescence. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology: Vol. 8. The Effects of Trauma on the Developmental Process (pp.403–434). Rochester, NY: University of Rochester Press, p.428.

285 Fraley, R.C. & Roisman, G.I. (2015) Do early caregiving experiences leave an enduring or transient mark on developmental adaptation? Current Opinion in Psychology, 1, 101–106.

286 Lyons-Ruth and colleagues reported findings from a high-risk sample of 76 families, comprising 41 referred to child protective services and 35 matched comparisons. Data were available on 56 families who remained in the sample 20 years later. Lyons-Ruth and colleagues reported that observed lack of responsiveness of parents at home and in the laboratory to their infants accounted for half of the variance in participants’ self-report of dissociative symptoms at age 19 on the Carlson and Putnam Dissociative Experiences Scale. The researchers did not find that either disorganised attachment or trauma predicted later dissociative symptoms; no analysis appears to have been conducted for avoidant attachment. However, verbal abuse by parents, which was not assessed by the Minnesota study, made an additional contribution to dissociative symptoms in adolescence, over and above caregiver emotional neglect/psychological unavailability. Dutra, L., Bureau, J.F., Holmes, B., Lyubchik, A., & Lyons-Ruth, K. (2009) Quality of early care and childhood trauma: a prospective study of developmental pathways to dissociation. Journal of Nervous and Mental Disease, 197(6), 383–90.

287 Garmezy, N., Masten, A.S., & Tellegen, A. (1984) The study of stress and competence in children: a building block for developmental psychopathology. Child Development, 55(1), 97–111.

288 E.g. Anthony, E.J. (1974) Introduction: the syndrome of the psychologically vulnerable child. In E.J. Anthony & C. Koupernik (eds) The Child in his Family: Children at Psychiatric Risk, Vol. 3 (pp.3–10). New York: Wiley.

289 Harper, D. & Speed, E. (2012) Uncovering recovery: the resistible rise of recovery and resilience. Studies in Social Justice, 6, 9–25; Joseph, J. (2013) Resilience as embedded neoliberalism: a governmentality approach. Resilience, 1(1), 38–52.

290 On later regret among developmental psychopathologists regarding the widespread popular interpretation of the concept of resilience, and the role of some of their members in contributing to this interpretation see Luthar, S.S., Cicchetti, D., & Becker, B. (2000) The construct of resilience: a critical evaluation and guidelines for future work. Child Development, 71(3), 543–62.

291 Werner, E.E. & Smith, R.S. (1982) Vulnerable but Invincible: A Study of Resilient Children. New York: McGraw-Hill; Werner, E.E. & Smith, R.S. (1992) Overcoming the Odds: High Risk Children from Birth to Adulthood. Ithaca, NY: Cornell University Press.

292 Sroufe, L.A. (2007) The place of development in developmental psychopathology. In A. Masten (ed.) Multilevel Dynamics in Developmental Psychopathology: Pathways to the Future: The Minnesota Symposia on Child Psychology, Vol. 34 (pp.285–99). Mahwah, NJ: Lawrence Erlbaum, p.293.

293 Yates, T.M., Egeland, B., & Sroufe, A. (2003) Rethinking resilience. In S. Luthar (ed.) Resilience and Vulnerability: Adaptation in the Context of Childhood Adversities (pp.243–66). Cambridge: Cambridge University Press, pp.249–50.

294 Supkoff, L., Puig, J., & Sroufe, L.A. (2013) Situating resilience in developmental context. In M. Ungar (ed.) The Social Ecology of Resilience (pp.127–42). New York: Springer, p.128.

295 Egeland, B., Carlson, E., & Sroufe, L.A. (1993) Resilience as process. Development & Psychopathology, 5(4), 517–28. See also Luthar, S.S., Doernberger, C.H., & Zigler, E. (1993) Resilience is not a unidimensional construct: insights from a prospective study of inner-city adolescents. Development & Psychopathology, 5(4), 703–717.

296 Sroufe, L.A., Coffino, B., & Carlson, E.A. (2010) Conceptualizing the role of early experience: lessons from the Minnesota longitudinal study. Developmental Review, 30(1), 36–51, p.39.

297 See also Causadias, J.M., Salvatore, J.E., & Sroufe, L.A. (2012) Early patterns of self-regulation as risk and promotive factors in development: a longitudinal study from childhood to adulthood in a high-risk sample. International Journal of Behavioral Development, 36(4), 293–302.

298 Yates, T.M., Egeland, B., & Sroufe, A. (2003) Rethinking resilience. In S. Luthar (ed.) Resilience and Vulnerability: Adaptation in the Context of Childhood Adversities (pp.243–66). Cambridge: Cambridge University Press, p.251.

299 Another illustration is Supkoff, L., Puig, J., & Sroufe, L.A. (2013) Situating resilience in developmental context. In M. Ungar (ed.) The Social Ecology of Resilience (pp.127–42). New York: Springer: ‘We found that we could distinguish between those who recovered from depression in adulthood following formation of a romantic partnership from those who did not recover after partnering on the basis of different histories of infant attachment (Sroufe, Corffino & Carlson 2010) ’ (138). In fact, however, these findings are not reported in Sroufe, Corffino, and Carlson (2010). Some analogous results are reported in Salvatore, J.E. (2011) Moderating processes in the link between early caregiving and adult individual and romantic functioning: the distinctive contributions of early adult romantic relationships. Unpublished doctoral dissertation, University of Minnesota. However the interaction is between the early childhood composite variable, not attachment, and adult relationships, in predicting depressive symptoms at age 23.

300 E.g. Widom, C.S., Czaja, C.J., & DuMont, K.A. (2015) Intergenerational transmission of child abuse and neglect Science, 347(6229), 1480–85; Ben-David, V., Jonson-Reid, M., Drake, B., & Kohl, P.O. (2015) The association between childhood maltreatment experiences and the onset of maltreatment perpetration in young adulthood controlling for proximal and distal risk factors. Child Abuse & Neglect, 46, 132–41.

301 Egeland, B., Jacobvitz, D., & Sroufe, L.A. (1988) Breaking the cycle of abuse. Child Development, 59(4), 1080–1088, p.1087.

302 Ibid. p.1082.

303 Egeland, B., Jacobvitz, D., & Papatola, K. (1987) Intergenerational continuity of abuse. In R.J. Gelles & J.B. Lancaster (eds) Child Abuse and Neglect: Biosocial Dimensions (pp.255–76). New York: Aldine De Gruyter, p.269.

304 Subsequent research has suggested that it is concrete ecological and financial supports that especially predict the interruption of the transmission of abusive care, more than individual psychological factors. St-Laurent, D., Dubois-Comtois, K., Milot, T., & Cantinotti, M. (2019) Intergenerational continuity/discontinuity of child maltreatment among low-income mother–child dyads: the roles of childhood maltreatment characteristics, maternal psychological functioning, and family ecology. Development & Psychopathology, 31(1), 189–202.

305 Bosquet Enlow, M., Englund, M.M., & Egeland, B. (2018) Maternal childhood maltreatment history and child mental health: mechanisms in intergenerational effects. Journal of Clinical Child & Adolescent Psychology, 47(1), 47–62. Later research with the sample also indicated competence in romantic relationships as a potent mediator of the association between experiences of abuse and neglect in childhood and parenting identified as maltreating. Labella, M.H., Raby, K.L., Martin, J., & Roisman, G.I. (2019) Romantic functioning mediates prospective associations between childhood abuse and neglect and parenting outcomes in adulthood. Development & Psychopathology, 31(1), 95–111.

306 Berzenski, S.R., Yates, T.M., & Egeland, B. (2014) A multidimensional view of continuity in intergenerational transmission of child maltreatment. In J.E. Korbin & R.D. Krugman (eds) Handbook of Child Maltreatment, Vol. 2 (pp.115–29). New York: Springer.

307 Thompson, R. (2006) Exploring the link between maternal history of childhood victimization and child risk of maltreatment. Journal of Trauma Practice, 5(2), 57–72. A recent meta-anlysis of 142 studies showed modest intergenerational transmission of abuse (k = 80; d = .45), and of kind of abuse (neglect: d = .24; physical abuse: d = .41; emotional abuse: d = .57; sexual abuse: d = .39). See Madigan, S., Cyr, C., Eirich, R., et al. (2019) Testing the cycle of maltreatment hypothesis: meta-analytic evidence of the intergenerational transmission of child maltreatment. Development & Psychopathology, 31(1), 23–51.

308 Supkoff, L., Puig, J., & Sroufe, L.A. (2013) Situating resilience in developmental context. In M. Ungar (ed.) The Social Ecology of Resilience (pp.127–42). New York: Springer, p.134. It is not clear, however, what outcome measures were used.

309 Sroufe, L. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29.

310 Cao, C., Rijlaarsdam, J., van der Voort, A., Ji, L., Zhang, W., & Bakermans-Kranenburg, M.J. (2017) Associations between dopamine D2 receptor (DRD2) gene, maternal positive parenting and trajectories of depressive symptoms from early to mid-adolescence. Journal of Abnormal Child Psychology, 46(2), 365–79.

311 Sroufe, L. & Rutter, M. (1984) The domain of developmental psychopathology. Child Development, 55(1), 17–29, p.26.

312 Pianta, R.C., Sroufe, L.A., & Egeland, B. (1989) Continuity and discontinuity in maternal sensitivity at 6, 24, and 42 months in a high-risk sample. Child Development, 60(2), 481–7, p.486.

313 Pianta, R.C., Egeland, B., & Sroufe, L.A. (1990) Maternal stress in childrens’ development: predictions of school outcomes and identification of protective factors. In J.E. Rolf, A. Masten, D. Cicchetti, K. Neuchterlen, & S. Weintraub (eds) Risk and Protective Factors in the Development of Psychopathology (pp.215–35). Cambridge, MA: Cambridge University Press.

314 Sroufe, L.A., Cooper, R.G., DeHart, G.B., & Marshall, M.E. (1996) Child Development: Its Nature and Course, 3rd edn. New York: McGraw-Hill, p.373.

315 Sroufe, L.A. & Egeland, B. (1991) Illustrations of person and environment interaction from a longitudinal study. In T. Wachs & R. Plomin (eds) Conceptualization and Measurement of Organism–Environment Interaction (pp.68–84). Washington, DC: American Psychological Association.

316 Sroufe, L.A., Cooper, R.G., DeHart, G.B., & Marshall, M.E. (1996) Child Development: Its Nature and Course, 3rd edn. New York: McGraw-Hill: ‘Gender is a central organising theme in development. It plays a key role in the way people define and experience their worlds. In all societies, parents and others treat boys and girls differently and expect different things from them. Because of this, children learn cultural stereotypes regarding male and female behaviours and characteristics. This learning begins early and is pervasive. It manifests itself in children’s activities, preferences, and social styles. Even among preschoolers, gender is so salient that a child’s most advanced thinking is often applied to it. Preschoolers label and categorise different activities in terms of gender (Fagot et al. 1992); they remember modelled behaviours better when they are “gender appropriate” (Bauer 1993); and in general they use gender as a basis for organising information (Serbin et al. 1993)’ (372–3).

317 Ainsworth, M. (1989) Letter to John Bowlby, 16 November 1989. PP/BOW/B.3/8.

318 Bakermans-Kranenburg, M.J. & van IJzendoorn, M.H. (2009) No reliable gender differences in attachment across the lifespan. Behavioral and Brain Sciences, 32(1), 22–3.

319 Raby, K.L., Labella, M.H., Martin, J., Carlson, E.A., & Roisman, G.I. (2017) Childhood abuse and neglect and insecure attachment states of mind in adulthood: prospective, longitudinal evidence from a high-risk sample. Development & Psychopathology, 29(2), 347–63, Table 4.

320 E.g. Sroufe, L.A. (1983) Infant–caregiver attachment and patterns of adaptation in preschool: the roots of maladaptation and competence. In M. Perlmutter (ed.) Development and Policy Concerning Children with Special Needs: The Minnesota Symposium in Child Psychology, Vol. 16 (pp.41–83). Hillsdale, NJ: Lawrence Erlbaum.

321 Egeland, B. (1997) Mediators of the effects of child maltreatment on developmental adaptation in adolescence. In D. Cicchetti & S. Toth (eds) Rochester Symposium on Developmental Psychopathology: Vol. VIII. The Effects of Trauma on the Developmental Process (pp.403–434). Rochester, NY: University of Rochester Press, p.417.

322 E.g. Herman, J.L. (1992) Trauma and Recovery. New York: Basic Books; Ceci, S.J. & Loftus, E.F. (1994) ‘Memory work’: a royal road to false memories? Applied Cognitive Psychology, 8, 351–64. For professionals’ views on recovered memories see Ost, J., Wright, D.B., Easton, S., Hope, L., & French, C.C. (2013) Recovered memories, satanic abuse, dissociative identity disorder and false memories in the UK: a survey of clinical psychologists and hypnotherapists. Psychology, Crime & Law, 19(1), 1–19. For discussions of this debate in historical context see e.g. Hult, J. (2005) The re-emergence of memory recovery: return of seduction theory and birth of survivorship. History of the Human Sciences, 18(1), 127–42.

323 Toth, S. L. & Cicchetti, D. (1998) Remembering, forgetting, and the effects of trauma on memory: a developmental psychopathology perspective. Development & Psychopathology, 10(4), 589–605.

324 Duggal, S. & Sroufe, L.A. (1998) Recovered memory of childhood sexual trauma: a documented case from a longitudinal study. Journal of Traumatic Stress, 11(2), 301–321, p.306.

325 Ibid. p.311.

326 Ibid. p.313.

327 Ibid. p.315–16.

331 Cicchetti, D. (2004) Biography. American Psychologist, 59(8), 1–4 p.728–31.

332 Cicchetti, D. (2015) Reflections on Carroll Izard’s contributions: influences on diverse scientific disciplines and personal recollections. Emotion Review, 7(2), 104–109.

333 Cicchetti, D. & Sroufe, L.A. (1976) The relationship between affective and cognitive development in Down’s syndrome infants. Child Development, 47, 920–29.

334 Several other later researchers in developmental psychopathology situated themselves as under the banner of Sroufe’s ‘organisational perspective’. See for instance Wyman, P.A., Cowen, E.L., Work, W.C., & Kerley, J.H. (1993) The role of children’s future expectations in self-system functioning and adjustment to life stress: a prospective study of urban at-risk children. Development & Psychopathology, 5(4), 649–61. However, Cicchetti has been the most consistent advocate for this as an explicit approach to and model of developmental psychopathology.

335 Cicchetti, D. & Sroufe, L.A. (1978) An organizational view of affect: illustration from the study of Down’s syndrome infants. In M. Lewis & L. Rosenblum (eds) The Development of Affect (pp.309–350). New York: Plenum Press; Cicchetti, D. & Schneider-Rosen, K. (1986) An organisational approach to childhood depression. In M. Rutter, C. Izard, & P. Read (eds) Depression in Young People: Clinical and Developmental Perspectives. New York: Guilford.

336 Cicchetti, D. & Greenberg, M.T. (1991) The legacy of John Bowlby. Development & Psychopathology, 3(4), 347–50.

337 Cicchetti, D., Taraldson, B., & Egeland, B. (1978) Perspectives in the treatment and understanding of child abuse. In A. Goldstein (ed.) Prescriptions for Child Mental Health and Education (pp.301–378). New York: Pergamon.

338 Cicchetti, D., Cummings, E.M., Greenberg, M.T., & Marvin, R.S. (1990) An organisational perspective on attachment beyond infancy: implications for theory, measurement and research. In M.T. Greenberg, D. Cicchetti, & E.M. Cummings (eds) Attachment in the Preschool Years (pp.1–50). Chicago: University of Chicago Press, p.35; Cicchetti, D. & Rogosch, F.A. (1996) Equifinality and multifinality in developmental psychopathology. Development & Psychopathology, 8, 597–600. These terms were drawn from general systems theory and embryology: see von Bertalanffy, L. (1968) General System Theory. New York: Braziller; Weiss, P. (1969) Principles of Development. New York: Hafner.

339 National Academy of Sciences (1989) Research on Children and Adolescents with Mental, Behavioral, and Developmental Disorders. Washington, DC: National Academy Press.

340 US Department of Health and Human Services (1990) National Plan for Research on Child and Adolescent Mental Disorders. Rockville, MD: US Department of Health and Human Services, National Institute on Mental Health.

341 Emde, R. (1990) Preface. In M.T. Greenberg, D. Cicchetti, & E.M. Cummings (eds) Attachment in the Preschool Years (pp.ix–xii). Chicago: University of Chicago Press, p.x.

342 Coll, C.G., Akerman, A., & Cicchetti, D. (2000) Cultural influences on developmental processes and outcomes: implications for the study of development and psychopathology. Development & Psychopathology, 12(3), 333–56. On the role of cultural components within the organisation of the attachment behavioural system see also Morelli, G., Chaudhary, N., Gottlieb, A., et al. (2017) Taking culture seriously: a pluralistic approach to attachment. In H. Keller & K.A. Bard (eds) The Cultural Nature of Attachment (pp.139–69). Cambridge, MA: MIT Press.

343 Cicchetti, D. (2002) How a child builds a brain: insights from normality and psychopathology. In W. Hartup & R. Weinberg (eds) Child Psychology in Retrospect and Prospect: The Minnesota Symposia on Child Psychology, Vol. 32 (pp.23–71). Mahwah, NJ: Lawrence Erlbaum.

344 Curtis, W.J. & Cicchetti, D. (2003) Moving research on resilience into the 21st century: theoretical and methodological considerations in examining the biological contributors to resilience. Development & Psychopathology, 15(3), 773–810.

345 Cicchetti, D. & White, J. (1988) Emotional development and the affective disorders. In W. Damon (ed.) Child Development: Today and Tomorrow (pp.177–98). San Francisco: Jossey-Bass.

346 Cicchetti, D. & Rogosch, F.A. (2001) The impact of child maltreatment and psychopathology upon neuroendocrine functioning. Development & Psychopathology, 13, 783–804.

347 Egeland, B. (1983) Book review: the second year: the emergence of self awareness by J. Kagan. Journal of Orthopsychiatry, 53, 365–7: ‘Most of us tend to ignore the issue of inheritance and the biological basis of behavior’ (366).

348 Sroufe had a career-long concern with medications given to children, and this concern was generally a sense of dismay at how social and family relationships got sidelined by the hunt for neuroendocrinal solutions to developmental problems. See e.g. Sroufe, L.A. & Stewart, M.A. (1973) Treating problem children with stimulant drugs. New England Journal of Medicine, 209, 407–413; Sroufe, L.A. (2012) The ‘other’ drug dilemma. USA Today, May 2012, p.21.

349 Siebenbruner, J., Zimmer-Gembeck, M.J., & Egeland, B. (2007) Sexual partners and contraceptive use: a 16-year prospective study predicting abstinence and risk behavior. Journal of Research on Adolescence, 17(1), 179–206, p.184.

350 Egeland, B., Weinfield, N., Bosquet, M., & Cheng, V. (2000) Remembering, repeating, and working through: lessons from attachment-based interventions. In J. Osofsky (ed.) WHIMH Handbook of Infant Mental Health, Vol. 4 (pp.35–89). New York: Wiley: ‘Suboptimal social environments tend to promote physiological systems that are dominated by the parasympathetic branch of the autonomic nervous system (avoidant histories) or the sympathetic branch of the autonomic nervous system (resistant histories)’ (42).

351 Raby, K.L., Cicchetti, D., Carlson, E.A., Cutuli, J.J., Englund, M.M., & Egeland, B. (2012) Genetic and caregiving-based contributions to infant attachment: unique associations with distress reactivity and attachment security. Psychological Science, 23(9), 1016–23.

352 Ibid. p.1017.

353 Raby, K.L., Roisman, G.I., & Booth-LaForce, C. (2015) Genetic moderation of stability in attachment security from early childhood to age 18 years: a replication study. Developmental Psychology, 51(11), 1645–9.

354 Hinshaw, S.P. & Cicchetti, D. (2000) Stigma and mental disorder: conceptions of illness, public attitudes, personal disclosure, and social policy. Development & Psychopathology, 12(4), 555–98.

355 E.g. Richers, J.E. & Cicchetti, D. (1993) Mark Twain meets DSM-III-R: conduct disorder, development, and the concept of harmful dysfunction. Development & Psychopathology, 5(1–2), 5–29.

356 Cicchetti, D. & Toth, S.L. (1997) Preface. In D. Cicchetti & S.L. Toth (eds) Rochester Symposium on Developmental Psychopathology: Trauma: Perspectives on Theory, Research, and Intervention, Vol. 8 (pp.xi–xvi). Rochester, NY: University of Rochester Press.

357 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.84.

358 Cicchetti, D., Toth, S.L., & Lynch, M. (1995) Bowlby’s dream comes full circle. In Advances in Clinical Child Psychology (pp.1–75). Boston: Springer. Cf. Bowlby, J. (1988) A Secure Base. London: Routledge: It was ‘unexpected that, whereas attachment theory was formulated by a clinician for use in the diagnosis and treatment of emotionally disturbed patients and families, its usage … has been mainly to promote research in developmental psychology’ (ix).

359 Cicchetti, D. & Toth, S.L. (1995) Child maltreatment and attachment organization: implications for intervention. In S. Goldberg, R. Muir, & J. Kerr (eds) Attachment Theory: Social, Developmental, and Clinical Perspectives (pp.279–308). Hillsdale, NJ: The Analytic Press, p.280.

360 Cicchetti and colleagues termed this ‘hierarchic motility’: the way that early components may be incorporated into later behavioural systems and remain accessible to be activated, especially during periods of extreme stress. This can be an asset, if earlier resources can be made available and repurposed for new tasks. However, it can also be a liability if these earlier components are called upon but not well suited to contemporary challenges. See Cicchetti, D. (1994) Integrating developmental risk factors: perspectives from developmental psychopathology. In C. Nelson (ed.) Threats to Optimal Development: Integrating Biological, Psychological, and Social Risk Factors: The Minnesota Symposia on Child Psychology, Vol. 27 (pp.229–72). Hillsdale, NJ: Lawrence Erlbaum.

361 Cicchetti, D. & Toth, S.L. (eds) (1996) Rochester Symposium on Developmental Psychopathology: Adolescence: Opportunities and Challenges, Vol. 7. Rochester, NY: University of Rochester Press.

362 Cicchetti, D. & Hinshaw, S.P. (2002) Prevention and intervention science: contributions to developmental theory. Development & Psychopathology, 14(4), 667–71; Toth, S.L., Sturge-Apple, M.L., Rogosch, F.A., & Cicchetti, D. (2015) Mechanisms of change: testing how preventative interventions impact psychological and physiological stress functioning in mothers in neglectful families. Development & Psychopathology, 27(4), 1661–74.

363 Cicchetti, D., Toth, S.L., & Rogosch, F.A. (1999) The efficacy of toddler–parent psychotherapy to increase attachment security in offspring of depressed mothers. Attachment & Human Development, 1(1), 34–66; Toth, S.L., Rogosch, F.A., & Cicchetti, D. (2008) Attachment-theory informed intervention and reflective functioning in depressed mothers. In H. Steele & M. Steele (eds) The Adult Attachment Interview in Clinical Context (pp.154–72). New York: Guilford.

364 Cicchetti, D. & Doyle, C. (2016) Child maltreatment, attachment and psychopathology: mediating relations. World Psychiatry, 15(2), 89–90, p.89.

365 Ibid. p.90.

366 Guild, D.J., Toth, S.L., Handley, E.D., Rogosch, F.A., & Cicchetti, D. (2017) Attachment security mediates the longitudinal association between child–parent psychotherapy and peer relations for toddlers of depressed mothers. Development & Psychopathology, 29(2), 587–600.

367 Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989) Finding order in disorganization: lessons from research on maltreated infants’ attachments to their caregivers. In D. Cicchetti & V. Carlson (eds) Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect (pp.494–528). Cambridge: Cambridge University Press.

368 Another early, important study of a sample with substantial rates of maltreatment was conducted by Mary Jo Ward, also a former student of Sroufe: Ward, M.J. & Carlson, E.A. (1995) Associations among adult attachment representations, maternal sensitivity, and infant–mother attachment in a sample of adolescent mothers. Child Development, 66(1), 69–79.

369 Schneider-Rosen, K., Braunwald, K.G., Carlson, V., & Cicchetti, D. (1985) Current perspectives in attachment theory: illustration from the study of maltreated infants. Monographs of the Society for Research in Child Development, 50(1–2), 194–210.

370 Crittenden, P.M. (1988) Relationships at risk. In J. Belsky & T. Nezworski (eds) Clinical Implications of Attachment Theory (pp.136–74). Hillsdale, NJ: Lawrence Erlbaum.

371 Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989) Finding order in disorganization: lessons from research on maltreated infants’ attachments to their caregivers. In D. Cicchetti & V. Carlson (eds) Child Maltreatment: Theory and Research on the Causes and Consequences of Child Abuse and Neglect (pp.494–528). Cambridge: Cambridge University Press, p.507.

372 Barnett, D., Ganiban, J., & Cicchetti, D. (1999) Maltreatment, negative expressivity, and the development of type D attachments from 12 to 24 months of age. Monographs of the Society for Research in Child Development, 64(3), 97–118.

373 Cicchetti, D., Rogosch, F.A., & Toth, S.L. (2006) Fostering secure attachment in infants in maltreating families through preventive interventions. Development & Psychopathology, 18(3), 623–49, p.646. The research was funded by the Administration of Children, Youth, and Families, the National Institute of Mental Health, and the Spunk Fund, Inc.

374 Fraiberg, S., Adelson, E., & Shapiro, V. (1975) Ghosts in the nursery: a psychoanalytic approach to impaired infant–mother relationships. Journal of the American Academy of Child Psychiatry, 14, 387–421.

375 Lieberman, A.F. (1992) Infant–parent psychotherapy with toddlers. Development & Psychopathology, 4, 559–74. See also Lieberman, A.F. & Van Horn, P. (2008) Psychotherapy with Infants and Young Children. New York: Guilford.

376 Cicchetti, D., Rogosch, F.A., & Toth, S.L. (2006) Fostering secure attachment in infants in maltreating families through preventive interventions. Development & Psychopathology, 18(3), 623–49, p.646.

377 Cicchetti, D., Rogosch, F.A., & Toth, S.L. (2011) The effects of child maltreatment and polymorphisms of the serotonin transporter and dopamine D4 receptor genes on infant attachment and intervention efficacy. Development & Psychopathology, 23(2), 357–72.

378 Cicchetti, D. & White, J. (1988) Emotional development and the affective disorders. In W. Damon (ed.) Child Development: Today and Tomorrow (pp.177–98). San Francisco: Jossey-Bass, p.185.

379 Cicchetti, D., Rogosch, F.A., & Toth, S.L. (2011) The effects of child maltreatment and polymorphisms of the serotonin transporter and dopamine D4 receptor genes on infant attachment and intervention efficacy. Development & Psychopathology, 23(2), 357–72.

380 Stronach, E.P., Toth, S.L., Rogosch, F., & Cicchetti, D. (2013) Preventive interventions and sustained attachment security in maltreated children. Development & Psychopathology, 25(4.1), 919–30.

381 Toth, S.L., Cicchetti, D., MacFie, J., Maughan, A., & Vanmeenen, K. (2000) Narrative representations of caregivers and self in maltreated preschoolers. Attachment & Human Development, 2(3), 271–305; Toth, S.L., Maughan, A., Manly, J.T., Spagnola, M., & Cicchetti, D. (2002) The relative efficacy of two interventions in altering maltreated preschool children’s representational models: implications for attachment theory. Development & Psychopathology, 14(4), 877–908.

382 Goossens, F.A., van IJzendoorn, M.H., Tavecchio, L.W.C., & Kroonenberg, P.M. (1986) Stability of attachment across time and context in a Dutch sample. Psychological Reports, 58(1), 23–32; Bakermans-Kranenburg, M.J. & van IJzendoorn, M.H. (1993) A psychometric study of the Adult Attachment Interview: reliability and discriminant validity. Developmental Psychology, 29(5), 870–79.

383 Cf. Collins, R. (2002) On the acrimoniousness of intellectual disputes. Common Knowledge, 8(1), 47–70.

384 See van IJzendoorn, M.H. & Bakermans-Kranenburg, M.J. (2014) Confined quest for continuity: the categorical versus continuous nature of attachment. Monographs of the Society for Research in Child Development, 79(3), 157–67.

385 On the exclusion of a disorganised dimension in Fraley and Spieker see Fraley, C. (2002) Response to Reviewers, Letter to Douglas M. Teti, Associate Editor of Developmental Psychology, 30 January 2002. Unpublished text shared by the author: ‘The reviewer states that “the way the authors address the issue of the disorganized group is not convincing. The D’s were coded in the NICHD sample. Why were they not included?” The MAXCOV analyses require at least three indicators, and there are only two indicators of D in the study: the overall D ratings of the two independent coders per case. The analyses simply cannot be performed. The other analyses were conducted with and without children classified as D. As noted in the manuscript, the results are the same either way. We would have liked to have had the opportunity to study the taxonicity of D, and we’re hoping that this article will be a catalyst for getting other researchers who might be pursue the matter further (e.g., researchers who originally developed the D classification may have access to a larger number of indicators of disorganization).’

386 The Working Model of the Child Interview, developed in the 1990s on the model of the AAI, also amalgamated the preoccupied and unresolved classifications. Benoit, D., Parker, K.C., & Zeanah, C.H. (1997) Mothers’ representations of their infants assessed prenatally: stability and association with infants’ attachment classifications. Journal of Child Psychology and Psychiatry, 38(3), 307–313: ‘the caregiver may seem preoccupied or distracted by other concerns, confused and anxiously overwhelmed by the infant, self-involved or may expect the infant to please or be excessively compliant’ (308).

387 Fraley, R.C, Roisman, G.L, Booth-LaForce, C., Owen, M.T., & 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), 817–38; web-based supplement, Part C.

388 Anticipated differences between clinical and non-clinical samples in this regard have been discussed by van IJzendoorn, M.H. & Bakermans-Kranenburg, M.J. (2014) Confined quest for continuity: the categorical versus continuous nature of attachment. Monographs of the Society for Research in Child Development, 79(3), 157–67.

389 Roisman, G.I., Fraley, R.C., & Belsky, J. (2007) A taxometric study of the Adult Attachment Interview. Developmental Psychology, 43, 675–86; Haltigan, J.D., Roisman, G.I., & Haydon, K.C. (2014) The latent structure of the Adult Attachment Interview: exploratory and confirmatory evidence. Monographs of the Society for Research in Child Development, 79(3), 15–35; Raby, K.L., Labella, M.H., Martin, J., Carlson, E.A., & Roisman, G.I. (2017) Childhood abuse and neglect and insecure attachment states of mind in adulthood: prospective, longitudinal evidence from a high-risk sample. Development & Psychopathology, 29, 347–63.

390 van IJzendoorn, M.H. & Bakermans-Kranenburg, M.J. (2014) Confined quest for continuity: the categorical versus continuous nature of attachment. Monographs of the Society for Research in Child Development, 79(3), 157–67. In support see Stovall-McClough, K. & Cloitre, M. (2006) Unresolved attachment, PTSD, and dissociation in women with childhood abuse histories. Journal of Consulting and Clinical Psychology, 74(2), 219–28.

391 Raby, K.L., Labella, M.H., Martin, J., Carlson, E.A., & Roisman, G.I. (2017) Childhood abuse and neglect and insecure attachment states of mind in adulthood: prospective, longitudinal evidence from a high-risk sample. Development & Psychopathology, 29(2), 347–63; Raby, K.L., Yarger, H.A., Lind, T., Fraley, R.C., Leerkes, E., & Dozier, M. (2017) Attachment states of mind among internationally adoptive and foster parents. Development & Psychopathology, 29(2), 365–78.

392 Haltigan, J.D., Roisman, G.I., & Haydon, K.C. (2014) The latent structure of the Adult Attachment Interview: exploratory and confirmatory evidence. Monographs of the Society for Research in Child Development, 79(3), 15–35.

393 Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford, p.35.

394 Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press, p.8.

395 Ainsworth, M. (1969) Object relations, dependency and attachment. Child Development, 40, 969–1025: ‘Psychoanalytically and ethologically oriented theories imply that the attachment—the relationship—resides in the inner structure, which has both cognitive and affective aspects, and which affects behaviour’ (1003).

396 Sroufe, L.A., Fox, N.E., & Pancake, V.R. (1983) Attachment and dependency in developmental perspective. Child Development, 54(6), 1615–27: ‘In the past 15 years, a major advance in the study of early social development has been the conceptual distinction between attachment (the relationship between infant and caregiver) and dependency (the reliance of the child on adults for nurturance, attention, or assistance)’ (1615).

397 Pederson, D.R. & Moran, G., (1996) Expressions of the attachment relationship outside of the strange situation. Child Development, 67(3), 915–27, p.915–16.

398 Rutter, M. & Sroufe, L.A. (2000) Developmental psychopathology: concepts and challenges. Development & Psychopathology, 12(3), 265–96, p.274.

399 E.g. Sroufe, L.A., Egeland, B., Carlson, E.A., & Collins, W.A. (2005) The Development of the Person. New York: Guilford: ‘Attachment is the “dyadic regulation of emotion” (Sroufe 1996). The infant must learn to draw effectively on external resources in the service of emotion regulation’ (42).

400 Sroufe, L.A. (2016) The place of attachment in development. In J. Cassidy & P.R. Shaver (eds) Handbook of Attachment: Theory, Research, and Clinical Applications, 3rd edn (pp.997–1011). New York: Guilford, p.1004.

401 Sroufe, L.A. (2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development, 7(4), 349–67, p.353.

402 Appleyard, K., Egeland, B., Dulmen, M.H., & Sroufe, L.A. (2005) When more is not better: the role of cumulative risk in child behavior outcomes. Journal of Child Psychology and Psychiatry, 46(3), 235–45, p.242–3.

403 Egeland, B., Carlson, E., & Sroufe, L.A. (1993) Resilience as process. Development & Psychopathology, 5(4), 517–28, p.526.

404 Korfmacher, J. (2001) Harder than You Think: Determining What Works, for Whom, and Why in Early Childhood Interventions. Chicago: Herr Research Center, Erikson Institute, p.6.

405 Erickson, M.F., Korfmacher, J., & Egeland, B.R. (1992) Attachments past and present: implications for therapeutic intervention with mother–infant dyads. Development & Psychopathology, 4(4), 495–507, p.501.

406 For further characterisation of the sample see Egeland, B., Erickson, M.F., Butcher, J.N., & Ben-Porath, Y.S. (1991) MMPI-2 profiles of women at risk for child abuse. Journal of Personality Assessment, 57(2), 254–63.

407 Consider, as a point of contrast, Bowlby’s hesitancy in the 1950s regarding whether attachment theory could have specific implications for intervention: Association for Psychiatric Social Workers (1955) Presentation at the Annual General Meeting 1955: Dr John Bowlby on preventative activities. Modern Records Centre Warwick University, MSS.378/APSW/P/16/6/19-20: ‘Dr Bowlby drew a sharp distinction between helping such workers to understand the nature of the problems with which they were dealing and advising them on what action they should take. He thought nothing but good could come from helping such workers to understand the nature of the problems … he was more sceptical on the wisdom of advising action.’

408 Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson. Another pioneering study from the period was Alicia Lieberman’s child–parent psychotherapy. This ran contemporaneously with STEEP, and results from the use of the Strange Situation as an outcome measure were published earlier: Lieberman, A.F., Weston, D.R., & Pawl, J.H. (1991) Preventive intervention and outcome with anxiously attached dyads. Child Development, 62(1), 199–209. Whereas Lieberman’s intervention was largely psychoanalytic in inspiration, STEEP was the first attachment-based family support programme also evaluated using attachment measures. Other evaluations of interventions using attachment measures soon followed. For a review of this early work see van IJzendoorn, M.H., Juffer, F., & Duyvesteyn, M.G. (1995) Breaking the intergenerational cycle of insecure attachment: a review of the effects of attachment-based interventions on maternal sensitivity and infant security. Journal of Child Psychology and Psychiatry, 36(2), 225–48.

409 E.g. Martensen, P., Spector, S., & Erickson, M.F. (1999) Applying attachment theory and research in a public health home visiting program. Zero to Three, 20(2), 20–22: ‘We are in a position to encourage Anita on the journey of looking at how her own childhood experiences of loss and shame come to bear on her relationships with her children. Bright and capable of insight, Anita shows glimmers of thinking about how her mother’s rejection of her might be playing out in her own anger toward Lisa. Or how her own self-hatred might be projected specifically onto her daughter, in whom she sees more of herself than in her son. Or how her tendency to have her children meet her emotional needs is rooted in her own early relationship experience. Anita is clear in stating that she wants to be sure neither of her children suffers in the way she did. And that is a solid foundation for helping her work through these issues and learn to see more clearly through the eyes of her children’ (21–2).

410 The use of video as part of supportive parenting interventions was introduced, at least in print, by Beebe, B. & Stern, D. (1977) Engagement–disengagement and early object experience. In M. Freedman & S. Grenel (eds) Communicative Structures and Psychic Experiences (pp.33–55). NewYork: Plenum.

411 The video-feedback technique would later be segmented off as a distinct and trademarked intervention called Seeing Is Believing.

412 Korfmacher, J. (1994) The relationship between participant and treatment characteristics to outcome in an early intervention program. Unpublished doctoral thesis, University of Minnesota.

413 Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson, p.37. Interestingly, an observer-reported measure of group cohesion found no associations with caregiving or maternal measures at the end of the intervention or follow-up: Korfmacher, J. (1994) The relationship between participant and treatment characteristics to outcome in an early intervention program. Unpublished doctoral thesis, University of Minnesota.

414 Mothers also reported fewer depressive symptoms. However, this finding was not replicated by Suess, G.J., Bohlen, U., Carlson, E.A., Spangler, G., & Frumentia Maier, M. (2016) Effectiveness of attachment based STEEP™ intervention in a German high-risk sample. Attachment & Human Development, 18(5), 443–60. Weak or no effects were found for mothers in the intervention group who had lower levels of participation. Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson. To the researchers’ surprise, variables that had been anticipated to correlate with poor participation were not significant: mother’s level of depression, anxiety, low self-esteem, cynicism, or any personality features. An explanation was identified in the fact that these qualities had no association with a scale of Facilitator Positive Regard for the mother, suggesting that ‘the facilitators, despite having little clinical training in handling a therapeutic alliance, remained true to the intervention model and did not back off from “difficult” mothers, so that negative aspects of the mother’s character made less difference in whether they became committed or not’. Korfmacher, J. (1994) The relationship between participant and treatment characteristics to outcome in an early intervention program. Unpublished doctoral thesis, University of Minnesota.

415 Egeland, B. & Erickson, M. (1993) An Evaluation of STEEP: A Program for High Risk Mothers, Final Report. Washington, DC: US Department of Health and Human Services, Public Health Service, National Institute of Mental Health.

416 Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson.

417 A replication of the STEEP evaluation in a German sample suffered from the same problem: high levels of security in the comparison group. In this case, the researchers attributed their findings to the high quality of ordinary services in Germany for at-risk parents. See Suess, G.J., Bohlen, U., Carlson, E.A., Spangler, G., & Frumentia Maier, M. (2016) Effectiveness of attachment based STEEP™ intervention in a German high-risk sample. Attachment & Human Development, 18(5), 443–60, p.454.

418 Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson, p.42. The same conclusion was drawn by Lieberman, who found increases in both disorganisation and anger between 12 and 18 months in their intervention, but no differences from controls at 24 months: Lieberman, A.F., Weston, D.R., & Pawl, J.H. (1991) Preventive intervention and outcome with anxiously attached dyads. Child Development, 62(1), 199–209, p.207.

419 Suess, G.J., Bohlen, U., Carlson, E.A., Spangler, G., & Frumentia Maier, M. (2016) Effectiveness of attachment based STEEP™ intervention in a German high-risk sample. Attachment & Human Development, 18(5), 443–60, p.446.

420 Korfmacher, J. (1994) The relationship between participant and treatment characteristics to outcome in an early intervention program. Unpublished doctoral thesis, University of Minnesota.

421 Egeland, B. & Erickson, M. (1993) An Evaluation of STEEP: A Program for High Risk Mothers, Final Report. Washington, DC: US Department of Health and Human Services, Public Health Service, National Institute of Mental Health.

422 Egeland, B. & Erickson, M. (1993) Attachment theory and findings: implications for prevention and intervention. In S. Kramer & H. Parens (eds) Prevention in Mental Health: Now, Tomorrow, Ever? (pp.21–50). Northvale, NJ: Jason Aronson, p.45–46.

423 Ibid. p.38.

424 Ibid. p.43.

425 Korfmacher, J., Adam, E., Ogawa, J., & Egeland, B. (1997) Adult attachment: implications for the therapeutic process in a home visitation intervention. Applied Developmental Science, 1(1), 43–52.

426 Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., & Juffer, F. (2003) Less is more: meta-analyses of sensitivity and attachment interventions in early childhood. Psychological Bulletin, 129(2), 195–215; Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., & Juffer, F. (2005) Disorganized infant attachment and preventive interventions: a review and meta-analysis. Infant Mental Health Journal, 26(3), 191–216.

427 Egeland, B., Weinfield, N., Bosquet, M., & Cheng, V. (2000) Remembering, repeating, and working through: lessons from attachment-based interventions. In J. Osofsky (ed.) WHIMH Handbook of Infant Mental Health, Vol. 4 (pp.35–89). New York: Wiley, p.79.

428 Ibid. p.70–71.

429 Ibid. p.78.

430 In fact, a recent meta-analysis has shown that attachment-based interventions actually increase in effectiveness as a function of the age of the child, contrary to the classical focus of Bowlby, Sroufe, and Egeland on the priority of early intervention. Facompré, C.R., Bernard, K., & Waters, T.E. (2018) Effectiveness of interventions in preventing disorganized attachment: a meta-analysis. Development & Psychopathology, 30(1), 1–11: ‘If programs are implemented prenatally or when infants are very young, parents may not have opportunities to practice skills, such as following children’s lead in play or responding in nonfrightening ways when toddlers begin to cause trouble. In addition, it is possible that interventions are more effective at later stages of infancy or during toddlerhood due to shifts in children’s developmental readiness. The later developmental window may represent a time of increased plasticity, during which children are more susceptible to environmental changes’ (8).

431 Bakermans-Kranenburg, M.J., van Ijzendoorn, M.H., & Juffer, F. (2003) Less is more: meta-analyses of sensitivity and attachment interventions in early childhood. Psychological Bulletin, 129(2), 195–215, p.210.

432 This conclusion can be partially qualified by recent findings from a replication of STEEP in a German sample that at 12 and 24 months there were fewer dyads classified as showing disorganised attachment in the intervention sample than the comparison group: Suess, G.J., Bohlen, U., Carlson, E.A., Spangler, G., & Frumentia Maier, M. (2016) Effectiveness of attachment based STEEP™ intervention in a German high-risk sample. Attachment & Human Development, 18(5), 443–60. In general terms, however, the criticisms made by Bakermans-Kranenburg and colleagues of various forms of comprehensive intervention on the basis of their meta-analysis still stand.

433 Korfmacher, J. (1994) The relationship between participant and treatment characteristics to outcome in an early intervention program. Unpublished doctoral thesis, University of Minnesota.The analysis compared: 1) Insight-oriented therapy; 2) Problem-solving around parenting tasks; 3) Problem-solving concerning areas of life besides parenting; 4) Emotionally supportive therapy; 5) Secondary ego support, supportive therapy involving internalization of good object relations; 6) Crisis-oriented treatment; 7) Companionship, with treatment involving some concrete assistance for the mother (such as providing transportation) or superficial visiting.

434 Egeland, B., Weinfield, N., Bosquet, M., & Cheng, V. (2000) Remembering, repeating, and working through: lessons from attachment-based interventions. In J. Osofsky (ed.) WHIMH Handbook of Infant Mental Health, Vol. 4 (pp.35–89). New York: Wiley, p.70. This point has been made again more recently by Slade, A., Simpson, T.E., Webb, D., Albertson, J.G., Close, N., & Sadler, L.S. (2018) Minding the Baby®: developmental trauma and home visiting. In H. Steele & M. Steele (eds) Handbook of Attachment-Based Interventions (pp.151–73). New York: Guilford.

435 Sroufe, L.A. (2005) Researchers examine the impact of early experience on development. The Brown University Child and Adolescent Behavior Letter, 21(11), 1, 5–6, p.6.

436 Facompré, C.R., Bernard, K., & Waters, T.E. (2018) Effectiveness of interventions in preventing disorganized attachment: a meta-analysis. Development & Psychopathology, 30(1), 1–11, Table 2.

437 Suess, G., Erickson, M.F., Egeland, B. Scheuerer-Englisch, H., & Hartmann, H-P. (2018) Attachment-based preventive intervention: lessons from 30 years of implementing, adapting and evaluating the STEEP™ program. In H. Steele & M. Steele (eds) Handbook of Attachment-Based Interventions (pp.104–128). New York: Guilford.

438 E.g. Niccols, A. (2008) ‘Right from the Start’: randomized trial comparing an attachment group intervention to supportive home visiting. Journal of Child Psychology and Psychiatry, 49(7), 754–64.

439 Knafo, H., Murphy, A., Steele, H., & Steele, M. (2018) Treating disorganized attachment in the Group Attachment-Based Intervention (GABI©): a case study. Journal of Clinical Psychology, 74(8), 1370–82.

440 Murphy, A., Steele, M., & 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 J.E. Bettmann & D.D. Friedman (eds) Attachment-Based Clinical Work with Children and Adolescents (pp.237–57). New York: Springer.

441 Steele, M., Steele, H., Bate, J., et al. (2014) Looking from the outside in: the use of video in attachment-based interventions. Attachment & Human Development, 16(4), 402–415.

442 Steele, H., Murphy, A., Bonuck, K., Meissner, P., & Steele, M. (2019) Randomized control trial report on the effectiveness of Group Attachment-Based Intervention (GABI©): improvements in the parent–child relationship not seen in the control group. Development & Psychopathology, 31(1), 203–217.

443 The Adverse Childhood Experiences measure assesses reports of five forms of abuse and five forms of household dysfunction (parent mentally ill, incarcerated, drug addicted, domestic violence, and separation/divorce).

444 An inverse relationship between attrition rates and differential effect sizes in attachment-based interventions was reported by Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., & Juffer, F. (2003) Less is more: meta-analyses of sensitivity and attachment interventions in early childhood. Psychological Bulletin, 129(2), 195–215.

445 See Grossmann, K.E., Grossmann, K., & Waters, E. (eds) (2005) Attachment from Infancy to Adulthood: The Major Longitudinal Studies. New York: Guilford.

446 Beller, S. & Bender, A. (2017) Theory, the final frontier? A corpus-based analysis of the role of theory in psychological articles. Frontiers in Psychology, 8, 951.

447 See e.g. Ward, M.J. & Carlson, E.A. (1995) Associations among adult attachment representations, maternal sensitivity, and infant–mother attachment in a sample of adolescent mothers. Child Development, 66(1), 69–79; Hazen, N.L., Allen, S.D., Christopher, C.H., Umemura, T., & Jacobvitz, D.B. (2015) Very extensive nonmaternal care predicts mother–infant attachment disorganization: convergent evidence from two samples. Development & Dsychopathology, 27(3), 649–61.

448 The primary attempt to recruit father-figures in the study was at the 13-year follow-up, where 175 mothers and 44 fathers or father-figures were brought in to the laboratory. However, the findings regarding fathers from the 13-year follow-up were not published in a prominent location, and do not appear to have ever subsequently been cited except by the authors. Sroufe, L.A. & Pierce, S. (1999) Men in the family: associations with juvenile conduct. In G. Cunningham (ed.) Just in Time Research: Children, Youth and Families (pp.19–26). Minneapolis: University of Minnesota Extension Service.

449 Ibid. An additional major empirical report received unfavourable peer-review and remained unpublished: Pierce, S.L., Coffino, B.S., & Sroufe, L.A. (2005) The Role of Fathers and Father Figures in the Development of Lower SES Children: Predicting Adolescent Externalizing Behavior from Early Childhood. Unpublished manuscript, Alan Sroufe personal archive.

450 There were several indirect sources of information on father–child relationships in the Minnesota study. In particular, the AAI offered information relevant to paternal care. However, as the children of the Minnesota study themselves became parents, the significance of fathers has come further into view: Kovan, N.M., Chung, A.L., & Sroufe, L.A. (2009) The intergenerational continuity of observed early parenting: a prospective, longitudinal study. Developmental Psychology, 45(5), 1205–213.

451 There have been a relatively small number of studies in the USA that have measured paternal sensitivity and infant attachment. These have included: Easterbrooks, M.A. & Goldberg, W.A. (1984) Toddler development in the family: impact of father involvement and parenting characteristics. Child Development, 55(3), 740–52; Volling, B.L. & Belsky, J. (1992) Infant, father, and marital antecedents of infant father attachment security in dual-earner and single-earner families. International Journal of Behavioral Development, 15(1), 83–100; Braungart-Rieker, J.M., Garwood, M.M., Powers, B.P., & Wang, X. (2001) Parental sensitivity, infant affect, and affect regulation: predictors of later attachment. Child Development, 72, 252–70; Eiden, R.D., Edwards, E.P., & Leonard, K.E. (2002) Mother–infant and father–infant attachment among alcoholic families. Development & Psychopathology, 14(2), 253–78; Kochanska, G., Aksan, N., & Carlson, J.J. (2005) Temperament, relationships, and young children’s receptive cooperation with their parents. Developmental Psychology, 41, 648–60; Hazen, N.L., McFarland, L., Jacobvitz, D., & Boyd-Soisson, E. (2010) Fathers’ frightening behaviours and sensitivity with infants: relations with fathers’ attachment representations, father–infant attachment, and children’s later outcomes. Early Child Development and Care, 180, 51–69

453 Fraley, R.C. & Spieker, S.J. (2003) Are infant attachment patterns continuously or categorically distributed? A taxometric analysis of Strange Situation behavior. Developmental Psychology, 39(3), 387–404.

454 Sroufe, L.A. (2003) Attachment categories as reflections of multiple dimensions. Developmental Psychology, 39(3), 413–16, p.414.

455 Ibid. p.413.

456 Ibid. p.415.

457 Ibid. p.415.

458 The primary exceptions were researchers in the developmental tradition with a particular focus on adult attachment, such as the Cowans, e.g. Cowan, P.A. & Cowan, C.P. (2007) Attachment theory: seven unresolved issues and questions for future research. Research in Human Development, 4(3–4), 181–201.

459 E.g. Aviezer, O. & Sagi-Schwartz, A. (2008) Attachment and non-maternal care: towards contextualizing the quantity versus quality debate. Attachment & Human Development, 10(3), 275–85.

460 Bernard, K. & Dozier, M. (2010) Examining infants’ cortisol responses to laboratory tasks among children varying in attachment disorganization: stress reactivity or return to baseline? Developmental Psychology, 46(6), 1771–8.

461 Sroufe, L.A. (2016) The place of attachment in development. In J. Cassidy & P.R. Shaver (eds) Handbook of Attachment: Theory, Research, and Clinical Applications, 3rd edn (pp.997–1011). New York: Guilford, p.1008.

462 E.g. Sroufe, L.A. (1996) Emotional Development. Cambridge: Cambridge University Press: ‘The security of the attachment relationship is related to the regularity with which arousal has historically led or not led to infant behavioural disorganisation in the context of the caregiver’ (158).

463 Sroufe, L.A. (2003) Attachment categories as reflections of multiple dimensions. Developmental Psychology, 39(3), 413–16, p.415.

464 Padrón, E., Carlson, E.A., & Sroufe, L.A. (2014) Frightened versus not frightened disorganized infant attachment. American Journal of Orthopsychiatry, 84(2), 201–208.

465 Reijman, S., Foster, S., & Duschinsky, R. (2018) The infant disorganised attachment classification: ‘patterning within the disturbance of coherence’. Social Science & Medicine, 200, 52–8.

466 Sroufe, L.A. (2013) The promise of developmental psychopathology: past and present. Development & Psychopathology, 25(4.2), 1215–24, p.1216.

467 Kuhn, T.S. (1970) The Structure of Scientific Revolutions, 2nd edn. Chicago: University of Chicago Press. See also Mannheim, K. (1952) Essays on the Sociology of Knowledge. London: Routledge.

468 An exception is Puig, J., Englund, M.M., Simpson, J.A., & Collins, W.A. (2013) Predicting adult physical illness from infant attachment: a prospective longitudinal study. Health Psychology, 32(4), 409–417. However, this appears to have been written during the transition period in leadership of the Minnesota group.