Inheriting Mary Ainsworth and the Strange Situation: Questions of Legacy,Authority,and Methodology for Contemporary Developmental Attachment Researchers

Abstract

Mary Ainsworth’s legacy continues to shape the social and developmental sciences well after her death. The Ainsworth Strange Situation Procedure has, for decades, not only provided the underpinning methodology of attachment research, but also the frame of reference for theory. This has produced conditions where, as in psychoanalysis, debates about the future of the paradigm also entail a struggle to claim and negotiate the legacy of a founding figure. To date, historians have only looked at attachment research up to the 1980s. Interviews with 15 leading contemporary attachment researchers revealed Ainsworth’s importance to later research, but also laid bare the challenges of claiming her inheritance in responding to the current challenges facing this area of research.

Introduction

Attachment research has its basis in the work of John Bowlby and Mary Ainsworth. In the 1950s, a research group led by Bowlby pursued a study of children separated from their caregivers by hospitalization or by foster care. On return home, the researchers found that the children could be “over-dependent” and “ambivalent” toward the caregiver on reunion, or “rejecting” and avoidant (Bowlby et al., 1956, p. 238). Bowlby came to theorize that children are predisposed to develop an attachment behavioral system, which directs them to seek the availability of their familiar caregiver or caregivers when alarmed or separated. Among children who had been chronically unable to achieve this availability due to hospitalization, the over-dependent and ambivalent response made sense since the continual pining for the attachment figure expressed an intensification of the attachment behavioral system, together with attendant frustration. In turn, the rejecting and avoidant response could make sense as an attempt to suppress the attachment behavioral system and its associated pining for the caregiver or caregivers. Bowlby conceptualized the two responses as opposite ways of dealing with the problem of the lack of assuagement of the attachment behavioral system.

A member of this research group in the 1950s, Mary Ainsworth drew on Bowlby’s theory to develop a laboratory-based procedure, the Strange Situation (SSP), as a means of studying differences between infant-caregiver dyads in the functioning of the attachment behavioral system. In the SSP, infants undergo two brief separations and reunions with their caregivers. This is intended to ratchet up the child’s stress by increments from mild to moderate, allowing the observer to examine the child’s changing responses. The behavior displayed is initially rated using four rating scales: proximity-seeking, contact maintenance, avoidance, and resistance. Together with a general qualitative assessment, these four scales then inform placement of the dyad in a category. Dyads are classified as secure when the child uses the caregiver as a secure base from which to explore the environment when calm, and as a haven of safety when alarmed by the separations. The secure base concept suggests that children are more willing to move away from their caregiver to play when they trust they can readily return as needed. Dyads are classified as insecure and avoidant when the child directs their attention away from the caregiver as a means of minimizing displayed distress. Dyads are classified as insecure and resistant when infants display distress and frustration as a means of maximizing the caregiver’s attention to them (Ainsworth et al., [1978] 2015). Main and Solomon (1990, p. 122) later identified a fourth category, in which children “exhibiting a diverse array of inexplicable, odd, disorganized, disoriented, or overtly conflicting behaviors in the parent’s presence.” These dyads were classified as disorganized. Ainsworth also developed a scale for assessing caregiver sensitivity to children. Her longitudinal study demonstrated that sensitivity observed in parents over a child’s first year of life was associated with their behavior in the SSP. This association has been affirmed by subsequent research, though the importance of other factors in predicting the SSP has also been highlighted, such as parental social support and life stress (e.g., Sroufe et al., 2009). Subsequent researchers have also documented associations between infant attachment and later developmental outcomes, such as social competence and externalizing behaviors (Groh et al., 2017).

Attachment has subsequently become one of the most popular theories of human socioemotional development, with a global research community and widespread interest from clinicians, child welfare professionals, educationalists, and parents. A striking feature of attachment research is that it has remained a distinct theoretical paradigm with an active research community over a period that has seen the rapid decline of psychological theory (Beller & Bender, 2017). For instance, in a survey conducted by the British government of organizations working with children in need of help and protection, attachment theory was, by a large margin, cited as the most frequently used underpinning perspective (Department for Education, 2018).

Social and cultural anthropologists have offered commentary on attachment research, including for instance criticism of Ainsworth’s Strange Situation and sensitivity scale. These researchers have been concerned about the implicit normative judgments contained within methodologies that attachment researchers have used to draw cross-cultural conclusions about childcare and child development (e.g., Keller, 2018; LeVine, 2014). As we have documented elsewhere, a deep gulf remains between the perspectives and goals of attachment researchers based in developmental psychology and social and cultural anthropologists interested in the same phenomena (Duschinsky, 2020).

An important commentary on the work of Bowlby and Ainsworth has also been offered by historians. Researchers have placed these developments within the broader history of psychology and of 20th century social science. Dutch historians of science have been particularly interested to examine the intellectual context of and influences on the work of Bowlby and Ainsworth (e.g., Van der Horst, 2011; Van Dijken, 1998; Van Rosmalen et al., 2015). British historians have focused more on how attachment ideas and their reception occurred in the context of the development in the welfare state (e.g., Lewis, 2013; Riley, 1983; Thomson, 2013).

Some historians such as Vicedo (2017) have attempted to debunk Ainsworth’s science as a means of undermining the contemporary influence of attachment ideas on policy, professional practice, and parenting. This has included extensive study and critique of the reception of the founders of attachment research within popular culture. Yet the legacy of Ainsworth for contemporary attachment researchers—in terms of methodology, theory, and research culture—has not been examined. In fact, a weakness of this historical literature, somewhat paradoxically, has been a tendency to assume that attachment research remains an expression of the ideas of Bowlby and Ainsworth, without attention to how the tradition has been renegotiated or changed over the past 30 years: the histories almost always stop in the 1980s.

In fact, contemporary attachment research has been facing an important period of transition, occasioned particularly by two developments. A first is that whilst attachment ideas retain a widespread audience, and perhaps even have grown in popularity, the empirical study of attachment has lost ground within developmental psychology. Developmental science in recent years has become preoccupied with big samples, clinical application, genes, and the brain. Researchers in the United States who apply for grants need to emphasize these themes with the Strange Situation on the periphery. A defining moment was the removal of the Ainsworth Strange Situation as a recommendation for publicly funded mental health research, by the National Institute of Mental Health (2016). It was stated that attachment theory “reified theoretical claims” made by psychoanalysis, therefore becoming redundant. In contrast to judgments made about other procedures in the document, no reference was made to the SSP’s psychometric properties, predictive validity, longitudinal stability, cultural specificity, or standardized administration. One can conclude that the disfavor of attachment research, more than established criteria of the NIMH motivated the rejection from public funding. Furthermore, the debt to psychoanalysis is characteristic of the founders of attachment research much more than subsequent empirical research. This aligns with other indications that the image of attachment research in circulation remains tied to Bowlby and Ainsworth, much more than subsequent empirical researchers, and their adaptations of the research program (Beckwith, 2019).

A second development has been the retirement of the second generation of attachment researchers, many of whom were personally trained by Ainsworth. The concept of a “generation” can be used to characterize members of a cohort who, facilitated by structural factors that suggest commonalities, regard themselves as facing a bundle of common challenges, including delimitation and appraisal of the legacy of an earlier generation. In the case of attachment research, a new set of researchers have taken over the nodal laboratories that have been central to the life of the research program over the previous decades (Duschinsky, 2020; Waters et al., 2013, 2021). The third generation is characterized as contemporary research leaders who did not have direct contact with Ainsworth. Yet despite this lack of personal contact, this generation must nonetheless negotiate Ainsworth’s inheritance, in judging what must be preserved, altered, or rejected from the tradition in responding to the field’s current challenges and opportunities. In the context of this changing of the guard, and the rise of a “third generation” of leaders of attachment research, established orthodoxies about the nature and future priorities of the paradigm have come into question, and with them the legacy of Ainsworth herself. Therefore, this study asks: “How do leaders of the second and third generation of attachment research negotiate the inheritance of Mary Ainsworth for the field today?” This necessitates first some consideration of theories of inheritance.

Theories of Inheritance

Inheritance has long been a concern for researchers across the social sciences and humanities (e.g., de Tocqueville, [1838] 2003; Smith, [1776] 1963). No attempt will be made at a general survey of scholarship on inheritance here. Work on the topic is quite fragmented both within and between disciplines. And what cohesion there has been, for instance in studies in inheritance law or economic history, have not developed a tradition of theory-building that would allow for application to problems in other domains.

There are three notable exceptions, however, of special relevance to questions of inherence within academic research. A strength of all three areas is that they have considered both successful and unsuccessful successions, rather than treating succession as inevitable.

Psychoanalysis

First, the earliest coherent body of theory was in psychoanalysis. In Totem and Taboo (Freud, [1913] 1960) Sigmund Freud presented a mythic tale of the origins of human society: a father exiled his male children in order to retain control over the family’s women, but the children returned to murder and eat the father. As a response to their guilt at this act, the children came to subsequently venerate the father, which became the origins of religion. Freud’s tale was written during and immediately following the breakdown of his relationships with Alfred Adler and Carl Jung, who had been considered as potential successors. Indeed, Freud (1911) stated explicitly to Jung that the work was an attempt to encompass a “larger synthesis” than Jung’s own work on myth.

Subsequent psychoanalytic theory has continued attention to the theme of succession, with particular attention to the topic over the past 15 years. Over the years, Freud’s story has been taken less a proposition about actual historical events, and rather as a parable for conceptualizing the work of differentiation and incorporation, and the potential for rivalry and love between generations, both in the family and in collective organizations. Psychoanalytic theorists have characterized Freud’s concern with veneration as a defense against guilt as a local case of a broader phenomenon: the need for a new generation to consider and enact change, whilst attempting to avoid the social costs of transgression against the values that organized the activities of the older generation (e.g., Paul, 2014). There is the risk of an “unsuccessful succession” if the younger generation fail to step up and squarely face the potential for change, as the result is stasis and a lack of responsiveness to circumstances (Marill & Siegel, 2004). Psychoanalytic theorists have also drawn out the need for the “band of brothers” to manage potential rivalries of their own: both in achieving sufficient coordination among the new generation for a fruitful renegotiation of the inheritance of the older generation, and in achieving a peaceable distribution of loyalties and possessions as part of this inheritance. Psychoanalytic reflections on succession have often been used reflexively in thinking about issues of leadership in psychoanalytic institutions, such as consideration of how these organizations can best retain fidelity to their inheritance whilst also responding to the changing position and declining status of psychoanalysis in contemporary society (e.g., Khaleelee, 2008).

Executive succession in business

A second cohesive body of theory regarding succession has been in studies of executive leadership in business. This work gained critical mass in the 1990s, concerned above all in the question of what makes a successful succession and the consequences for performance (Kesner & Sebora, 1994). Various aspects of the problem have been explored, such as whether internal or external appointment of the new CEO tends to be more successful, and the role of organizational inertia or potential for learning (for reviews, see Berns & Klarner, 2017; Cragun et al., 2016). For instance, the ideology of the founder of the organization has been identified as a significant factor. For example, Haveman and Khaire (2004) found that a founder with a strong ideology or vision contributed to the longevity of a firm, but increased the likelihood that a firm would not survive subsequent successions.

The challenges of periods of succession have been repeatedly highlighted: only 12% of family firms survive transition to a third generation of leaders (PwC, 2016). Fear of what happens when a third generation need to inherit has been described as a common worry, since the environment that initially made the business a success will likely have long since altered, necessitating a delicate work of preservation and change (Yanagisako, 2002). New leaders must gain the trust of employees and stakeholders and defuse potential rivalries, whilst simultaneously adapting the firm to challenges that were deferred or not registered by the previous regime (De Massis et al., 2008). For instance, potential rivalries with other potential leaders, or with the leaders of the previous generation, may be mitigated if the achievements of these others are not perceived as under threat from the new regime—but this places potential limits on what changes can be implemented within the organization in responding to the environment (Osnes, 2014).

Cultural anthropology

A third area of cohesion in the study and theorization of succession has been work in kinship and reproduction that formed part of the basis for social anthropology as a discipline. Researchers have been interested in the way that succession preserves arrangements on inequality in the distribution of resources. Studies have examined, in particular, the role of symbolic processes in consecrating heirs and reducing or managing threats of splits in the patrimony (e.g., Goody, 1973), and the importance of enculturating heirs for succession. This latter point has been developed in the theoretical work of Bourdieu ([1989–1992] 2014, [2002] 2008, p. 238), who has criticized the “naively critical view of the social order” in which “heirs are only too happy to inherit.” In fact, heirs need to learn the forms of valuation and judgment that make them able to inherit, including registering the importance and relevance of the patrimony itself: “the role of educational strategies is thus absolutely capital, as a real work of inculcation is needed to produce a king who wants to inherit and is qualified to do so.” This perspective underlines the potential fragility of succession, and the way that communities that had otherwise been stable for some time may rapidly face extinction if conditions change and a younger generation perceive insufficient rewards to make inheritance worthwhile, compared to other opportunities (Bourdieu, [2002] 2008). Similarly, the extinction of a paradigm may result if it loses utility for its heirs.

In sum, work on inheritance within psychoanalysis, business studies and cultural anthropology has documented the importance of the topic for shedding light on social structures and their transformation. Questions of inheritance spotlight the construction, distribution, and renewal of power and ownership; the need for and the threat posed by change to a cultural tradition in the context of changing environmental demands; the role of imagined pasts and futures in binding collectives; remaining static in the face of change and the training and investment of individuals to act for collectives. In studying issues of inheritance in the tradition of attachment research, this inquiry has the potential to shed light on fundamental contemporary dynamics of this influential area of scientific practice, as well as offer findings suggestive for understanding stability and change in other areas of academic inquiry.

Methods

Sampling

The population for this study comprises leaders of research groups whose work is predominantly concerned with child attachment. This group was purposively sampled, and comprises both members of the second and of the third generation of attachment researchers. All are affiliated to academic institutions, though some also pursue clinical practice and commercial training of practitioners. About 39 participants were invited to interview; 15 accepted the invitation. All the participants were developmental psychologists of some form; though social psychologist attachment researchers were invited to take part, none agreed. This may suggest differential access to a claim to Ainsworth’s legacy, as will be discussed later. Participants represented five nationalities and all interviews were conducted in English via online platforms (Skype, Zoom, and Facetime). The NWU Institutional Research and Ethics Regulatory Committee provided permission for the research to be conducted and all participants gave consent for recording and use of the interviews.

Data Collection and Analysis

A qualitative descriptive research design was utilized to understand participants’ perceptions and experiences of the legacy of Ainsworth for contemporary attachment research. Semi-structured interviews were used, focused by questions about participants’ encounters with Ainsworth’s work as well as her contributions to attachment research and the meaning her work has for researchers today. These were informed by developments in the methodology of oral history, which have come to emphasize how the words of participants are embedded within broader structures of meaning shaped by speaker’s practical concerns and the social and institutional structures that organize these concerns (Shopes, 2014). This includes consideration of convergence and divergence between interview data and published sources in seeking to understand the speaker’s perspective and context. In this study, the interviewees were very senior researchers, much higher status than the interviewer. However participants seemed enthusiastic about taking the opportunity presented by the interview to think about shifts in direction in attachment research in historical perspective. Indeed, as others have observed (e.g., Portelli, 2018), oral history is a methodology well suited to asking about the meanings participants give to the culture they have inherited, and the problems they and their community face in the present.

Clarke and Braun’s (2013) model for thematic analysis was used to analyze the interview data. Both major and minor themes were examined, in order to understand both consensus and dissensus in perceptions of Ainsworth and her legacy. Analysis of the interviews was performed with the aid of Atlas.ti software. In line with Clarke and Braun’s (2013) model, codes were generated using an open coding method. Similar codes were then grouped into code groups. Code groups facilitated the construction of networks, enabling the researchers to attribute relationships between the codes and code groups. These networks were then utilized to generate themes and subthemes. Identifying information was removed from data drawn upon in the paper to protect participant anonymity. To support trustworthiness of interpretations of the data, member checking was performed during two phases. During the first phase, when there was ambiguity regarding participants’ meaning in the interviews, these participants were contacted to gain clarity. In the second phase, all participants were provided with an opportunity to scrutinize and offer feedback on a draft of the paper. We did not implement all feedback provided in the second phase; in any case many of the comments disagreed with one another. Our primary goal in member checking was to attempt to ensure that we had understood the accounts of our participants, not to align our account with theirs. For instance we sought to make sense of the key lines of convergence and divergence between attachment researchers of the older and the younger generation, rather than align our paper with either stance.

Reflexivity is a key part of thematic analysis as a methodology. RS is a clinical psychologist and academic based in South Africa. His doctoral study, undertaken in the Netherlands, used the Strange Situation to study the attachment relationships of preschool-aged children of mothers with HIV. RS understood the interviews, conducted the analysis and drafted the paper. He had no prior relationship with any of the interviewees, though he knew of their work. His own doctoral supervisor was the former doctoral student of one of the interviewees 25 years earlier, but there had been no prior contact between RS and this interviewee. Reflexivity was helped by being based in South Africa, away from the centers of gravity of attachment research and their institutional influence (Hume & Wainwright, 2018). RS was supported by regular meetings and discussions with RD, a historian based in the United Kingdom, who provided feedback on the analysis and paper draft. RD knew most of the interviewees from his own prior research, which examined differences in perspectives between attachment research groups, and between researchers and practitioners. He has written elsewhere about his in-but-out status among attachment researchers in his pursuit of historical inquiry and critique of attachment research (Duschinsky, 2019).

The present research was founded on a belief shared by RS and RD that attachment research has potential value as a project for understanding and contributing to the wellbeing of children and families, but that it has become complacent about some of its underpinning assumptions, hindering the realization of this value. One goal for us in asking about the inheritance of Mary Ainsworth’s ideas and methodology has been to understand the process and potential for renewal within attachment research. Some social scientific and historical commentary on attachment research has sought to “debunk” the paradigm as a whole and as it exists today through criticism of Bowlby and Ainsworth as founding figures (e.g., Vicedo, 2017; Walsh et al., 2014). Our view is that precisely consideration of differences among attachment researchers, and especially change in perspectives over time, offers a more accurate, less “flattened” picture (Duschinsky, 2020). It is also one that offers, we hope, more suggestive for understanding change processes in psychological research more generally.

Findings

The Leadership of Mary Ainsworth

Ainsworth won the inaugural winner of the Mentor Award at the 106th American Psychological Association convention in 1998, and has been repeatedly described by the international community in the published literature as an academic secure base from which researchers could explore (Stevenson-Hinde, 1999). This was echoed strongly by the participants in interview. A third generation attachment researcher from America stated, “[Ainsworth] is a fantastic secure base.” He wished that he could speak to Ainsworth herself, to say: “thank you for your career, otherwise I would have none.”

Yet several participants also expressed concern that Ainsworth had subsequently been allocated the role of “prophet, god and the queen of attachment.” As a result, they felt that images of Ainsworth and her legacy had sometimes unhelpfully shaped contemporary priorities and perceptions of legitimate research. A third generation researcher alleged that the second generation had been too concerned with remaining loyal to Ainsworth in their research, rather than following the signals present in the scientific work they were conducting: “She was doing work in her time and others should have been doing work in their time as opposed to just drawing on her template of twenty years prior.” A second generation researcher acknowledged the point, stating that, “I read all her work and just took it as gospel.”

Perhaps some of the intense loyalty shown by her students toward Ainsworth is encapsulated in the repeated description of her as a “mother figure” for her students (e.g., Stevenson-Hinde, 1999). Ainsworth (2013) herself had referred to her students as her “academic family.” One participant stated that “she had all her graduate students that were her children. Then she had her grandchildren who were people like Mark Greenberg, Jude Cassidy, and a lot of other students of her students. Then she had a lot of adopted children,” offering examples such as Alan Sroufe and Klaus and Karin Grossmann.

Constructions of Ainsworth as the “mother” of developmental attachment research and attachment researchers aligned with participants’ emphasis on her labor to keep the family connected, both with her and with one another. Ainsworth was described as intensely concerned to facilitate communication between researchers so that they could maintain contact and learn from each other. Leaving the nest to establish their own laboratories, the small group of Ainsworth’s immediate students and collaborators

had gone their separate ways. Mary was the essential. . . I would say the modern word would be ‘networker’. She gave me access to her library and was constantly reminding me to contact X, and she would give me unpublished papers by any of the group. (Second generation researcher).

This tradition of close networking and personal contact between attachment researchers in the developmental tradition, established by Ainsworth, was regarded by participants as having continued into the present. As reported by a third-generation researcher: “We need regular contact with one another that goes beyond simply reading each other’s work. Some of these debates ought to be worked out together in person, or at least over Skype.”

Ainsworth was also described by participants as shaping and actively directing what could count as orthodox attachment research and theory. Participants acknowledged that there have been some limitations to this directedness. “She has provided direction. Perhaps a little bit overly directed in some sense.” (third generation researcher).

Limitations of the Scientific Paradigm Modeled on Ainsworth

The Strange Situation Procedure (SSP) has been widely regarded as Ainsworth’s most important legacy. Ainsworth is quoted as saying, “the Strange Situation is a tool to be used while it is needed; the better use we make of it, the sooner we will put it aside” (Waters et al., 2013, p. 673). Yet participants’ reflections on the SSP acknowledged its continued and even hegemonic importance. Nonetheless most participants were also critical of how an excessive focus on the SSP had contributed to a reification of the construct of attachment, and of the way that Ainsworth’s focus on sensitivity had directed attention away from other important aspects of caregiving.

The strange situation procedure as “gold standard.”

The most widespread and basic concern among participants was regarding the way that the SSP had tended to circumscribe the limits of imagination in attachment research. One of Ainsworth’s graduate students described “a hardening of commitment to the Strange Situation as though it was a gold standard of assessment that captured everything interesting about relationships and about infants. Regrettably that resulted to some extent in a closing of minds and an unwillingness to look for other ways of exploring individual differences.” The SSP was described as having come to play this role because it was successful in predicting expected correlates:

An enduring problem in psychology is that when you find something statistically significant, people click their heels with excitement. A number of meta-analyses have showed a statistical significant difference between babies who behaved in one fashion and those who behaved in another fashion in the strange situation, but what has been forgotten is the issue of how much of the variance is adequately explained. Of course, the meta-analyses do refer to it, but the headlines don’t say, ‘Strange situation behavior explains a reliable three to ten percent of the variance adequately.’ I don’t blame Ainsworth for the problem. Although I do think in the 1980s, she became much more bound up with zealously protecting and promoting the Strange Situation.

However, another participant recalls a conversation from this period with Ainsworth in which “she said, ‘I made a mistake in developing the Strange Situation. . . people grabbed on to it.’” It was recalled that, in her final interview (Ainsworth & Marvin, 1995), Ainsworth explained that she wanted researchers to use the SSP together with naturalistic observations sensitive to cultural context, but that the SSP came to dominate. In this, Ainsworth felt that the second generation of attachment researchers had neglected an important part of her legacy. An exception was her enthusiasm for the Attachment Q-sort developed by Waters and Deane (1985), which assesses attachment as a dimension of secure versus insecure and can be used in naturalistic observations, such as in the home, and may have greater benefits in application cross-culturally. Our interviewees acknowledged the strengths of the Attachment Q-sort, but felt it had not achieved anything like the hold over the imagination of attachment researchers of the Strange Situation.

In the interviews, many participants also expressed concern with misapplications of the SSP to clinical assessment, without adequate validation of the tool for this use. This was recently the subject of a major consensus statement by 40 leading attachment researchers (Granqvist et al., 2017), who urged the need for caution in applications of the SSP. Participants worried that clinicians assumed that perceived insecurity represented pathology:

In the minds of the social workers this has amounted to something quite serious, like ‘Insecure!! That can’t be good!’ So, in some cases they have actually removed kids from an intended permanent foster home or even from biological parents, even when there have been no additional signs of adversity in the home, because insecure or disorganized attachment is believed to be such a big deal.

It was felt that a selective interpretation of Ainsworth and the Strange Situation had entered into public circulation, focused on the prediction of development from the categorization of infants and assimilating Ainsworth’s work to common-sense assumptions about harm to children and hegemonic notions of diagnosis and pathology. Though the Strange Situation is an assessment of dyadic attachment relationships, what is measured is the behavior of individual infants. Participants felt that the impression that the Strange Situation measured differences between infants had contributed to reification of attachment in uses of the instrument, in the research program in general, and above all its reception by practitioners:

People came to think of attachment as being a characteristic or quality of the infant and not a descriptor of the relationship. It’s a simple thing that had a tremendous impact.

Over-focus on sensitivity

Non-American and third generation attachment researchers especially criticized Ainsworth’s focus on sensitivity as the main predictor for attachment security. Meta-analysis of decades of research has revealed that the relationship is much weaker than anticipated by Ainsworth (Verhage et al., 2016), and Fearon and Roisman (2017) have described a variety of important moderating variables in the sensitivity-attachment relationship. A second generation researcher expressed regret that Ainsworth’s

work has not been taken sufficiently enough as exploratory. It has been taken as more or less definite in terms of substantiation of the basic correlation between sensitivity and attachment. Repeated attempts to replicate the strong link she found between sensitivity and attachment may have delayed the search for alternative complementary determinant predictors of attachment security.

He stated:

The effect size in that study is impossibly large. It’s really an outlier, effect-wise. That is because the development of the measurement of both the strange situation and sensitivity measures took place in the same subset of families in which this idea of the connection between the two was tested. So, you get confusion between development and testing, and between context of discovery and context of justification. . . and I think that is a pity. . . .people have been busy to try and come closer to the effect size that Ainsworth found. It may have delayed the progress a bit, because if you want to have an effect size of about .8 and you get .2. . .and the field is sure that it should be .8, then you’re going to search for mistakes, blame yourself or your assistants, or the type of population, or measure, or the training. . . all that, whereas it [the effect] might just in reality be a little bit less and maybe it should be.

A second-generation participant described the greatest flaw in Ainsworth’s theory as her assumption that “all children are more or less equally susceptible to the effect of sensitivity and attachment security.” This was also a particular concern for all the third-generation attachment researchers in the sample. One expressed both his respect for and frustration with the data that have been compiled in this area:

People have been looking at sensitivity, and we can synthesize that data, but you know, this is 30 years of work, and we only have that piece of the puzzle figured out, and not so many resources spent on other ways. We have other instruments, but if you describe attachment theory it is always about sensitivity and always about strange situations and that is because people have perhaps been trying so hard to replicate her strong initial findings. In “trying to replicate Ainsworth’s findings, we have these many repetitions of very similar studies, which is great for your replication of science, but not so great for your progression of theory.”

Another third-generation attachment researcher echoed that “all bets are placed on sensitivity as the transmission mechanism and it’s pretty simplistic if you really think about it.” Ainsworth “didn’t pay sufficient attention to parents’ many different responsibilities. A mother is not only a mother, she’s also often a woman who is working and she doesn’t only have one child, so she needs to be flexible in how she prioritizes things in her life. If you look at the strict operationalization of sensitivity and of high scores in particular, it looks as though this parent does nothing but perfectly attend to the child whenever he or she needs the parent. That’s a very idealistic version of parenting, most certainly so from an evolutionary perspective and also in light of the demands placed on parents in the contemporary Western world.”

Orthodoxies in Attachment Methodology

In considering Ainsworth’s legacy, participants described Ainsworth’s contribution to certain in-group/out-group dynamics within the field of attachment research. Some participants only hinted at the idea of this phenomenon, while others made it very clear:

She built on Bowlby’s foundation and extended it in important ways and that was the foundation for me and others to go forth and intellectually multiply in a Biblical and metaphorical sense, but the downside has been that there was a merger, and it may be less evident today than it was in the seventies and sixties, where we ended up with what I coined the ‘Attachment Mafia.’ There was this dogmatism in the articulation of attachment theory and efforts to deviate, which means differentiate, modify, revise, rethink the dogma, was often treated as heresy.

Participants perceived that a critical factor predisposing the formation of such orthodoxy was the demands of training in the intensive observational assessments of developmental attachment research, above all the Strange Situation—as well as the Adult Attachment Interview introduced by Ainsworth’s student Mary Main (Hesse, 2016). These trainings are in principle open to everyone, but require significant investment of time and resources. A second-generation researcher stated, “If attachment theory and research is a club, then it is clearly a club that anyone can join. Many of the major contributors to attachment, such as Belsky, Kochanska, Sagi, and Thompson, have no direct connection to Ainsworth.” Nonetheless, the sheer time and resource requirement to gain fluency in the method and theory of attachment research made it difficult to dabble: participants characterized developmental attachment research as more all-or-nothing than other cognate areas of inquiry, and as dependent on a good deal of tacit knowledge.

One third generation participant reported that being part of the in-group offered credibility and a buffer against being ignored or rejected:

I was trained by [major in-group figure of the second generation] and had the good fortune of not just being trained, but frankly being launched into a career with some of his credibility rubbing off. If you are not fully embedded in a particular intellectual context or an insider, breaking in can be incredibly difficult. Being an insider, it’s been very difficult for me to be expelled. The fact is if you’re part of the family, there’s a very high threshold for getting expelled.

While being part of the in-group offered advantages and protection, being an outsider meant a lower threshold for rejection. Another participant reported:

So, the bottom-line is I think it’s challenging for those who are outside the tradition. If you’re not trained by the right people at the right institutions, it’s hard for your work to be taken seriously and the way I think some folks who have really been excited about attachment research have managed that is by attending those trainings. That means that they are not well positioned to advance beyond the status quo, as they sort of cathect to the canonical view.

One figure discussed by participants as outside the tradition was Ainsworth’s graduate student Patricia Crittenden, who developed a system with many additional classifications beyond those proposed by Ainsworth. However, in part due to personal and in part due to scientific controversies, since the 1990s lines of communication between Crittenden and her followers and mainstream attachment research have been poor (see Duschinsky, 2015; Van Ijzendoorn et al., 2018). A third generation researcher in our sample identified the response to Crittenden’s ideas as reflecting a broader dynamic:

My sense is that there was a way in which some of the efforts, for example Crittenden’s efforts, were sort of poo-pooed in a way that maybe was correct, but there was nonetheless this dynamic of ‘we have a gold standard, so stick to it’ – even if it was unintended.

Participants were clear that the existence of an orthodoxy did not stop the potential for disagreements within the group. An example of this is Everett Waters’ criticisms of the disorganized attachment classification in the 1990s, which were controversial but tolerated (Waters & Valenzuela, 1999), mostly by being ignored, though these criticisms of limitations to the construct have become more accepted over time (Granqvist et al., 2017). There was concern among some participants that in-group/out-group dynamics and the potential for rejection had inadvertently functioned as a social threat, disrupting by limiting the work of generating and empirically testing hypotheses, as well as the pool of people with a stake in pursuing this work.

Developmental and social psychology

One issue of orthodoxy raised by many participants was her dislike of self-report methods. Ainsworth had attempted to develop self-report measures of security in the 1950s and 1960s but had ultimately given them up as too vulnerable to participant biases. In a paper written in 1984, Ainsworth urged future researchers: “do not take at its face value a person’s self-reports of security.” In the 1980s, the consensus among Ainsworth’s immediate colleagues was that self-report assessments could not assess attachment, that this path had been attempted by Ainsworth and was found to be blocked. It was anticipated that an ethological focus on observational measures was preferable. Whether through interpretation of Ainsworth’s published remarks or from personal conversations, there was a general attitude among participants that Ainsworth had disapproved of self-report assessments of attachment. There were occasional efforts that were not published (e.g., Hesse & Van Ijzendoorn, 1991).

Yet a strong tradition of self-report research on adult attachment emerged from within social psychology in the 1980s, led by researchers such as Phillip Shaver (Ravitz et al., 2010). Social psychology aims to understand individuals’ self-concept and attitudes toward others, so self-report approaches are more popular than in developmental psychology, where the reporting subject may often not be assumed. In the 1990s and early 2000s, relations between the social psychological and developmental traditions of attachment research were far from cordial. One close colleague of Ainsworth’s from this period dismissed the self-report tradition of attachment research as a deviation from the theory as established by Ainsworth. This participant argued that the self-report measures represented the assessment of some other aspect of personal relationships, and that these measures often had correlates that would not be expected by Ainsworth’s theory: “These self-report measures clearly tap into personality given their range of correlates with other self-report measures, but it is not clear that they are specifically measures of attachment, which would require anchoring to at least some observable attachment behavior.” It is possible that the sense of alienation from Ainsworth and her legacy contributed to the lack of willingness of social psychologist attachment researchers to take part in the present study.

Over the past 15 years, relations between the two traditions have grown warmer. A few third generation researchers, such as Pehr Granqvist and Glen Roisman, have drawn from both traditions in their work. Indeed a distinctive characteristic of third generation participants was their construction of Ainsworth in interview as a pioneer and radical, interested above all in scientific advancement over maintenance of orthodoxy. Fidelity to Ainsworth meant fidelity to her science, not her beliefs. One third generation researcher felt that, though he never met her, the Ainsworth he imagined would have been encouraging to both traditions of attachment research: “I would hope that Ainsworth wouldn’t act like some of my developmental psychologist colleagues. They sort of close a frontier against social psychologists” and “they say ‘this is not real attachment research’. They turn a blind eye to the fact that self-report measures do predict a lot of what the theory would expect them to predict.”

Four Categories or Two Dimensions

Ainsworth developed scales for coding aspects of attachment behavior, but ultimately reported her data in terms of categories of attachment. The second generation of attachment researchers almost always followed Ainsworth’s lead on this, though a few contrary positions began to emerge from the late 1980s (e.g., Van Ijzendoorn & Kroonenberg, 1990). From the 2000s, Ainsworth’s scales and all the unpublished data accumulated through their use in coding have been foregrounded by third generation researchers, such as Fraley and Roisman, who have advocated a two-dimensional model of individual differences in attachment, influenced by developments in the social psychological tradition (Brennan et al., 1998), in place of Ainsworth’s categories. A first dimension is of proximity seeking + contact maintenance versus avoidance. The second dimension is of proximity seeking + contact maintenance versus resistance + disorganization (Fearon & Roisman, 2017).

The second-generation researchers, with direct links to Ainsworth, have traditionally been defenders of the category-based system for coding the SSP (e.g., Sroufe, 2003). In interview, the majority of participants were convinced that Ainsworth would not have been open to the two-dimensional approach. A graduate student of Ainsworth’s stated, “When we first started talking about another way of extracting information from the Strange Situation behavior, she was relatively noncommittal. She felt that most of the meaningful information was captured by the categories.” Another former student suggested that she would have been open to dimensions depending on what the dimensions were, “. . .if you’re dimensionalizing security, you lose avoidance and resistance as different forms of insecurity. I think Ainsworth would have been against it.”

Discussion

Ziv and Hotam (2015) have argued that part of how attachment research has circumvented the decline of theory in academic psychology has been by embodying theory within methodology, protecting theoretical propositions whilst supplanting the need to explicitly acknowledge or discuss them. Our findings concur that, when designing studies and interpreting findings, attachment researchers, particularly those of the second generation, have drawn on the frame of the Ainsworth SSP as an important interpretive framework. This has given Ainsworth and her ideas an important symbolic role within subsequent attachment research: she is not merely one of the founders, now gone. To achieve initiation and recognition within the developmental tradition of attachment research it has generally been necessary for researchers to learn to code her measures at lengthy training institutes, permitting the development and trained exercise of complex (and often implicit) skills of perception, thought, appreciation, and valuation (Waters et al., 2021). Through training in the SSP, and mentorship by existing leaders, researchers become members of a community socialized in tacit skills of observation, conceptualization and judgment that offer access to the practical intricacies of relevant theory, such as the secure base concept (Duschinsky, 2020). Our interviews have also highlighted that such researchers can also gain a greater degree of license in what theoretical positions they can hold, whilst still being treated as able to contribute to the cumulative tradition. The commonly-used metaphor of the developmental community as the “family” “descended” from Ainsworth reflected, and likely partially reinforced, these structural dynamics of the study of attachment in developmental science (Atkinson, 2014).

In this, there are analogies to psychoanalysis, where ownership of the perceived legacy of Freud has served as a source of legitimacy, such that new developments must renegotiate the founding father’s ideas and clinical method, and membership of the community depends on a lengthy process of tacit enculturation and mentorship (see e.g., Eisold, 2008; Young-Bruehl, 2008). For instance, Lunbeck (2014) has documented that Kohut’s ascendance as a leader and innovator in psychoanalytic theory in the 1970s was dependent on his earlier credibility and acceptance within the American psychoanalytic community, including as an exegete of Freud’s ideas. Despite their differences, what links Freud and Ainsworth is that for new entrants to the communities they established, tacit enculturation in method is a key means of gaining accreditation and acceptance.

The social psychological tradition in attachment research offers a clear counterpoint: anyone can pick up and conduct a study using the available self-report measures without training or any complicated understanding of theory. Facing accusations by developmental attachment researchers that their work is not “real” attachment research, Shaver and Mikulincer (2002) have claimed Bowlby as a forebear for research in social psychology. However little equivalent attempt has been made to claim Ainsworth in published writings or undertake trainings in her measures and receive enculturation in related tacit knowledge; social psychologists also did not respond to invitations to take part in the present study. The exception that proves the rule is a statement by Granqvist (2014), one of the few developmental researchers to also use measures from the social psychological tradition, who has described the achievements of social psychological attachment research as “surpass[ing] nearly all other research programs in psychological science, both in terms of originality and sheer quantity. I’ll stand on Mary Ainsworth’s coffee table in my cowboy boots and say that!” Both the gendered and spatial aspects of the metaphor are interesting, with Granqvist asserting that he would be willing to trespass Ainsworth’s domestic space with masculine outdoor boots, in asserting that the contributions of social psychological research on attachment should be recognized, including by the developmental tradition.

The social psychological tradition of attachment research is like many other research programs: they do not raise issues of inheritance to the same degree as the developmental tradition, since they do not require such extensive enculturation in order to produce a subject whose claims are considered credible. Of course, both psychoanalytic theory and developmental attachment research certainly circulate in simplified versions, which require little shift in the values or judgments of those who draw on them. These simplified versions are “non-formative” cultural forms (Wood, 2009). By contrast, for those initiated into the tradition, both psychoanalytic theory and the measures of developmental attachment research have been claimed by their adherents to offer precisely such a generalized shift in the perception of social relationships. For many other areas of psychological science, researchers’ eyes are turned primarily toward the next experiment; what is relevant from the past can largely be extracted through meta-analysis. By contrast, the subtle and complex theory embedded in the unpublished scales used for coding the SSP and Ainsworth’s sensitivity scale, and the trainings in how to use them passed down from Ainsworth, mean that the legacy of this founding figure is a special concern during a period of succession for attachment as a research program. Our interviews showed that new leaders of the attachment community experience the challenge of holding together a community anchored by training in measures that are becoming rather antiquated precisely in the necessity of such extensive training, and the complexity and ambiguous psychometric properties of the measures that prompt the need for training. They are also reflecting on the potential to reactivate aspects of Ainsworth’s contribution that faded out of view among her immediate students, most notably her scales for coding the Strange Situation. Our interviews also suggest growing interest among third generation researchers in reviving Ainsworth’s use of qualitative observation of caregiving practices, though use of mixed-methods research and direct collaboration with qualitative researchers is still rare (though see e.g., Longhi et al., 2020).

Though generally invisible in the published record, our interviews revealed qualities of the affective relationship between the generations of researchers. In general, these relations were characterized quite warmly, with little rivalry. As well as the affection of former mentors and mentees, and the construction of the community as a “family,” this warmth appears to have been sustained because the third generation appear to show no signs of putting aside the main achievements of the second generation. Even if the effect sizes from meta-analytic research are smaller than from high-profile individual studies conducted by leaders of the second generation (Groh et al., 2017; Verhage et al., 2016), these meta-analytic studies still suggest important implications of early attachment, and in interview the second generation generally felt that their contribution has been retained.

A partial exception has been the proposal to replace Ainsworth’s categories with use of her scales in a two dimensional model, which was met with dismay by second generation researchers. However, curiously, no attempt has been made to demonstrate the relative predictive validity of the SSP categories or the two dimensional model. Data to address these theoretical and methodological debates is readily available to the third generation from the numerous cohort studies conducted by second generation researchers. There are remaining technical issues for pursuing these questions, such as variation among researchers in perspectives on how best to compute and integrate scale scores for the two dimensional model. However an additional reason such analyses have not been conducted, or at least not published, may have been to avoid direct antagonism between the third and second generation.

Even if rivalry has been avoided or circumvented, it is unquestionable that developmental science has shifted profoundly since Ainsworth put down her tools at the start of the 1990s and the second generation of attachment researchers assumed leadership of the research program. The third generation spoke of feeling the intense challenge of at once preserving and adapting their inheritance. The issue of the reception of attachment research among practitioners was incidental for the second generation; our interviews indicate practitioner understanding as now a particular worry, and the translation of attachment theory into interventions as a source of pride. The second generation worked hard to validate Ainsworth’s argument for the importance of caregiver sensitivity; this importance is now accepted, but third generation researchers suspected that too exclusive a focus on sensitivity has risked placing attachment research out of alignment with both contemporary social values and scientific values which emphasize multiple causality.

The second generation were much more concerned with fidelity in the use of Ainsworth’s subtle measures than in the scalability of their assessments; it was accepted that the number of active researchers in the developmental tradition would be limited by the costs of entry. By contrast, the third generation has seen attachment research become increasingly regarded as parochial. Responding to this situation, Pasco Fearon—chair of the Society for Emotion and Attachment Studies, and one of the leaders of the third generation—has worked to validate a brief version of the Attachment Q-sort, for use at scale in research and clinical practice. Yet a first attempt to establish convergence with the Strange Situation found no association between the two measures (Cadman et al., 2018). This finding confronts attachment researchers with a stark dilemma and a microcosm of their wider predicament: how much they are willing to compromise allegiance to the Strange Situation in adapting their tradition to the demands of contemporary developmental science.

Conclusion

This study set out to understand how the contributions of Mary Ainsworth are perceived by contemporary attachment researchers, and what must be preserved, altered, or rejected from her legacy in responding to the field’s current challenges and opportunities. Our sample included both second generation of attachment researchers, who with a few exceptions were direct students of Ainsworth, as well as a new third generation of research leaders who were trained by the second generation. Our findings revealed the ongoing importance of Ainsworth and her legacy for participants, and the challenges for attachment research in adapting this legacy to the challenges of contemporary developmental science. The SSP was acknowledged to have profoundly shaped the methodological and conceptual imagination of the second generation of developmental attachment researchers. Though this had contributed to a convergent research program and to cohesion among developmental attachment researchers, many participants identified drawbacks. These included obstacles to innovation, such as the refinement of Ainsworth’s scales or uptake of the Attachment Q-sort, and the firm identification of other aspects of children’s care that predict attachment besides sensitivity. The subtlety and complexity of Ainsworth’s measures was also perceived as having necessitated the extensive enculturation of developmental attachment researchers, contributing to a research culture with relatively high cost of entry, and a split with social psychological research on attachment. These were all issues that the third generation of attachment researchers are actively seeking to address in their current work, and as such possible images of Ainsworth loomed large for them. Imagined conversations with Ainsworth took place, as contemporary attachment researchers worked to sift her legacy for value, to preserve and to adapt it.

In general, traditions flourish when the inheritance on offer appears attractive and worth the costs of inheriting, and can help coordinate collective activities that are adapted to the present and its pressing demands. We might speculate that the continued vitality of attachment research going forward, and its continued ability to attract new researchers, will depend significantly on the extent to which efforts to adapt the “gold standard” measures to make them less resource-intensive and usable by practitioners are successful. These efforts will, we suspect, direct attention away from measurement of child behavior and concern with the Ainsworth categories, and toward multi-dimensional measurement of caregiver behavior (for indications of this see e.g., Forslund et al., 2021; Madigan et al., 2021; Schuengel et al., 2021). Limitations of the scientific paradigm modeled on Ainsworth and orthodoxies in the measurement of attachment among second generation of attachment researchers have been powerful contributors to the longevity and autonomy of attachment research as a field of scientific inquiry. However they have both also hindered the circulation of the field’s knowledge: potentially limiting the recruitment of potential researchers who must be willing to learn labor-intensive methods; thinning the cross-disciplinary movement of researchers within and between forms of attachment study and other forms of scientific inquiry; and obstructing communication between researchers and applied practitioners.

A limitation of the study is that our sample contained only attachment researchers from the developmental tradition; attachment researchers from social psychology declined to participate. Our interviewees were told that we were interested in the legacy of Mary Ainsworth; with this clear to participants from the start, and shaping their interpretation of the goals of the interview, we were not in a position to compare how they spoke about Ainsworth in contrast to other influential figures in the history of attachment research, or in the history of psychology. Our recruitment of research leaders using purposive sampling naturally also limits the present study. Not least, it excludes the perspectives of attachment researchers who are not heads of laboratories, which is a major limitation. The richness of the heritage of developmental attachment research may be a special selling point for attracting younger generations to the research program, despite other challenges the program faces in adapting to the demands of contemporary developmental science. If so, then understanding more about the training, perspectives and priorities of this potential “fourth” generation, and their interpretation of the tradition they seek to inherit, would be a valuable next step.