The God, the blood, and the fuzzy: reflections on Cornerstones and two target articles

ABSTRACT

In response to Cornerstones of attachment research and the target articles, I reflect on three questions. First, what is “attachment”? Although a natural kind, I argue against an essentialist understanding (i.e. in terms of necessary/sufficient conditions for class membership). Instead, the attachment concept must be allowed to have fuzzy boundaries, partly because of how attachments transform in both phylogeny and ontogeny. Second, how to think about the normative (species-typical) features of the theory vis-à-vis dyadic/individual differences in attachment? Whereas the former are foundational, I argue that the latter largely reflect surface variation. Despite this, the lion’s share of attachment research has horned in on variation and its measurement, to some detriment to the theory’s potential and applications. Finally, what is encouraging and discouraging about recent developments? While applauding large-scale cooperative endeavors (e.g. individual participant meta-analyses, consensus statements) I caution the field not to lose sight of the value of smaller-scale, creative explorations of uncharted territories.

KEYWORDS:

• Attachment
• evolution
• essentialism
• philosophy
• God

The proverbial fly on the wall – I now know who it is. With his Cornerstones, Robbie Duschinsky (2020) has made an unprecedented contribution to the history and sociology (and more) of attachment theory and research. The target articles (Duscinsky et al., this issue; Schuengel et al., this issue) further extend these contributions in important ways. One of many realizations I gained from reading Cornerstones and the target articles is how little we really know, especially about foundational matters. Another one was, however, satisfaction at how much we do know, and how many useful things (like evidence-based caregiving interventions) the field has produced. For heuristic purposes, though, I will now plow into what we still don’t know.

What is attachment?

What is attachment, really? Perhaps a more fruitful way to pose the question is, what are the concept’s boundary conditions? Or, as phrased in Cornerstones, “what relationships qualify”? (p. 571). The honest answer appears to be that we don’t know – not really.

Indeed, Bowlby didn’t quite know either. As repeatedly illustrated in Cornerstones, he switched between “narrow” (e.g. following-response directed to discriminated figures) and “broad” (e.g. including secure base components) usages of the term in a fairly frivolous, carefree fashion. To illustrate how varied Bowlby’s use was, he applied it to the offspring’s selective, affectional relationships with its early caregivers, to spousal relationships (Bowlby, 1980), patients’ relationships to psychotherapists (Bowlby, 1988), and people’s relationships to special places, to inanimate objects, even to the government/state (Bowlby, 1969/1982). It seems clear that with such varied usage, to which others have contributed additional suggestions (e.g. to siblings, groups, pets, God, drugs, imaginary peers, smart phones; see discussion in Granqvist, 2020), the concept runs considerable risk to become confusing and perhaps to lose pragmatic usefulness. This calls for conceptual housecleaning, more than for additional “creative” candidate additions to the list. But conceptual housecleaning will necessarily have to include empirical research, which is designed to test whether particular relationships meet the “criteria” (a topic I shall return to) set up to define attachments. It is, after all, an empirical question how people actually relate to various objects in their lives so this cannot be dealt with by theoretical, a priori considerations alone. And humans are, a posteriori, a peculiar species that acts and thinks in ways that continue to be surprising, even for psychologists.

A key example of human oddity is that with development humans, especially when they come together in large groups, tend to create and treat fictions (e.g. money, ideologies, social contracts, human rights, gods, insurances) as though real, the real deal even (Harari, 2014). A related peculiarity is the pervasiveness by which humans anthropomorphize nature and artefacts, hyperactively attributing human mind properties where they do not belong (e.g. Epley et al., 2007). Granted, humans’ attachment figures are usually other real humans. However, human attachment figures are also, in the words of songwriter Lloyd Cole (2013), “fallible, weak, and so tiresome”. To overcome their shortcomings, humans put their fictionalizing, anthropomorphizing proclivities to use, creatively envisioning perfect (omnipresent, omniscient, all-loving) attachment figures whose very existence even escapes the problem of empirical refutability. Yes, much empirical research indicates that humans treat God, psychologically and functionally, as a noncorporeal attachment figure (Granqvist, 2020; Granqvist & Kirkpatrick, 2016).

Unlike God, however, some other suggested attachment targets mentioned by Bowlby (e.g. governments/states) are not typically anthropomorphized and conceivably therefore also less readily identified as targets by the attachment system. Tellingly, empirical research so far indicates that not even the welfare state, and not even by Swedes, is cognitively used as an attachment target (Gruneau Brulin et al., 2018). Most of the other candidate attachment targets listed above have not been adequately addressed empirically, and some (e.g. material objects like smart phones) appear to be used for attachment functions by exception rather than default (Keefer et al., 2014). It seems plausible to hypothesize that anthropomorphic and related organismic features, like individuality and agency, are often key boundary conditions for attachments among humans. Indeed, attachments tend to develop towards attachment “figures”. Nonetheless, despite 50 years of theory and research, we don’t know for sure.

What we know and don’t know, and why

Attachment is not a fictional, merely hypothetical construct. Infants around the world in most mammalian (and many bird) species can be observed to selectively follow, cling to, or use other means to achieve physical proximity to the stronger-wiser others available to them during early ontogeny. So in its narrowest ethological sense, attachment is by all means a real observable phenomenon, probably even a “natural kind” (e.g. Quine, 1969). Mere fictions are rarely both universal and observable. So why do we still not know what attachment truly is?

The reasons are, I believe, simple. First, nature typically does not comprise clear essences, so answering the deceptively simple question of what something really “is” is usually notoriously difficult, often impossible. This also means that semantic boundaries (definitional criteria) are to some extent arbitrary. Of course most of us academics may pretend that they aren’t; we usually think it’s important with definitions that clarify or give exactness to the formal conditions for class membership. In calling for definitional clarity and providing plenty of exemplifying deviations from “best practice”, Cornerstones and the Duschinsky et al. target article would seem like cases in point. Yet, a noble Nobel laurate has expressed disagreement: “Greed and lust I can understand but I can’t understand the values of definition and confinement. Definition destroys. Besides, there’s nothing definite in this world” (Dylan, 1976). Definitions rarely cut nature at its joints but carve out hard-edged freaks out of a nature so smooth. They can become shackles for the mind. Even in visual perception, we are lured into sensing that reality is neater and with clearer boundaries than it is and has (e.g. Gregory, 1997).

This is in keeping with what Wittgenstein (1953) – a former Cambridge colleague of Duschinsky’s not discussed in Cornerstones – sought to convey with his critique against essentialism. As is familiar, he argued for a radically different approach to determining class membership than the essentialist definition ideal that had dominated since antiquity. Because there are no essences in nature, Wittgenstein asserted, there are no frontiers by which we can definitively settle if something belongs to a particular category or not; categories (like “attachment”) are fuzzy sets, with fuzzy boundaries. Instead, all we can hope for are family resemblances. Rather than binary “fit vs not” to definitional criteria, degree of similarity with a category prototype can be used to settle class membership, if such settlement is important. Such a “prototype” approach to determining what relationships qualify as attachments has not, to my best knowledge, been explored but could potentially make a worthwhile contribution (cf. Rosch & Mervis, 1975; Shaver et al., 1987).

A second reason why we don’t know what attachment “is” is that most things that develop, like attachments, usually continue to develop, occasionally even making puzzling later transformations. This is true of attachments both in phylogeny and ontogeny. For example, a baby bird and a human baby primarily rely on signaling and physical proximity maintenance to discriminated figures in their expressions of attachment. Whereas this behavior develops and disappears early in most birds’ (and some mammals’) lives, however, it lingers among humans/primates, with additional relational layers ultimately added in ontogeny. This includes treating their caregivers as safe havens; preferentially seeking proximity (and later symbolic forms of contact) to them when alarmed or distressed, rather than escape to a burrow or den, as other mammals might (on this note, see Bowlby’s enthusiastic letter to his wife, as cited in Hesse & Main, 2000). In cognitively advanced species like humans – born into variable ecological niches and cultural settings, and with long developmental periods of immaturity and dependence – proximity to the caregiver also comes to facilitate learning (including social learning), where the attachment figure is used as a secure base for exploration (Granqvist, 2021). It seems likely then, as both Bowlby and Ainsworth ultimately concluded, that the hominin version of attachment includes strong developmental prospects for safe haven and secure base acquisition in addition to mere proximity maintenance, whereas this is not the case for some other mammals (or birds). Selection pressures have conceivably worked on physical proximity maintenance among birds but additionally on safe haven and secure base components among humans/primates. This also appears to be the general trend of how attachments develop in ontogeny among humans, both to caregivers and peers (Zeifman & Hazan, 2016).

Thus, attachment is not the same entity across species or developmental periods. No wonder then that the concept has been used in different ways – indeed, it must be. That said, I agree with Duschinsky and colleagues that increased clarity on its use in specific contexts would help prevent confusion and miscommunication. Their examples of the different connotations this and other key terms (e.g. internal working models, trauma, security) can take across different contexts are illuminating and provide a service for the field, especially for its dialogue with practitioners and the public.

As discussed in Cornerstones, because of the variable ways in which attachment had been used, Ainsworth (1985) made a more concerted effort to bring conceptual clarity to the term than Bowlby had. She did so in an implicitly essentialist manner, however, defining attachment as an affectional bond between two individuals (none of which is interchangeable); in which the attached person selectively maintains proximity to the attachment figure(s); protests involuntary separation; mourns and grieves the loss of the figure; and perceives the figure as stronger and wiser, as witnessed by using him/her as a safe haven and secure base. Notably, if all italicized features are treated as necessary conditions, then very (too) few relationships would warrant class inclusion among humans, and most other animals would not develop attachments. To me it appears more plausible to treat these features as central to the category prototype (in humans, the infant toddler’s relationship with its principal caregivers) to which other candidates can be empirically compared to establish degree of resemblance.¹

The normative vs dyadic/individual difference components of attachment

An issue touched upon but not discussed at length in Cornerstones or in the target articles is how to think about the species-typical (or “normative”) vis-à-vis the dyadic/individual difference components of attachment theory. The general way most attachment scholars, me included, probably think about these is that the normative features are truly foundational; strongly evolutionarily canalized, load-bearing, and applicable – in variable ways – to many, many species, especially primates, including humans. In contrast, dyadic/individual differences (security and organization) refer to variation that is “surfacy” enough, however locally adaptive, to be observable in a sufficient number of dyads/individuals in most populations. In comparison with the normative features, this variation is generally less important, both biologically and psychologically.

To illustrate, children who grow up in strongly species-atypical environments where no familiar caregiver can be found may fail to develop attachments, with massive developmental damage, including serious psychopathology, as a consequence. That children are allowed to develop/maintain attachment(s) is thus imperative. Children who grow up with attachments that deviate somewhat from the most common, usually “ideal” (secure) pattern with one caregiver have generally experienced slightly lower sensitivity from that particular caregiver (though they may of course develop secure attachments to other caregivers). Perhaps as a consequence of this, or of associated factors, they generally display a slightly less ideal developmental trajectory than securely attached children. Thus, it is desirable to help children and their caregivers achieve secure attachment, but by no means is secure attachment a prerequisite for favorable development. This message can also be illustrated by considering that secure attachment is no immunity but “merely” one protective factor among other ones against other developmental adversities.

The relative weight I have assigned to the normative (much weight) and dyadic/individual difference components (less weight) may understandably be surprising for readers who are not so familiar with the foundations of attachment but it is likely uncontroversial among attachment scholars. Uncontroversial, when most of what we’ve been doing the last four decades has concerned dyadic/individual differences, and their measurements and correlates? Then why have we assigned so much time, resources, and prestige to something that actually carries less weight in development? Why have we horned in on the “not as important” parts of the theory, when other, more important parts have been readily available? These might be the most puzzling questions about the development of the attachment “paradigm”, and yet so rarely posed. When something has this combination of qualities – puzzling and yet not addressed – it clearly deserves our attention.

This is one of the key matters I had hoped Cornerstones would discuss at some length, but the discussion is relatively brief, mostly referring to what was expected in North American developmental psychology from Ainsworth’s days onwards. Granted, the field of attachment, like every other field, has been squeezed, pushed and pulled in certain directions by external forces partially outside of the field’s own control. And yet, from the outset our field has had some of the strongest, most hard-working, and stubborn leaders of psychology plow their way through the thick terrains surrounding them, fiercely battling every obstacle. It is hard to imagine these leaders just resigning to external pressures.

Though brief, Duschinsky’s reference to “North American” and “psychology” is apt, because they provide converging paths to the same destinations: the individual (the topic of psychology), individual differences (distributed with monstruous, peacocky tails in American society), and individualism (nowhere as proud and blatant as in America). Not coincidentally, since the birth of psychology in the 19^th century, the focal interest has steadily tilted from normative processes to individual differences (e.g. Leahey, 1987). This can probably best be understood as a function of the triumphs of liberalism and associated individualist values, both legitimizing and causing increases in individual differences over time. Indeed, individual differences have become so important that we often put a blind eye to the foundational commonalities that unite us as individuals of a certain species. This is how attachment itself – a relational, dyadic construct of utmost importance – ultimately became mistreated as individual differences in attachment quality (as see the Duschinsky et al. target article).

With the foregoing discussion I do not mean to downplay the importance of Ainsworth’s discovery of natural variation in attachment behaviors (also anticipated by Bowlby). Variations in attachment quality often do qualify (or moderate) the output of the attachment system, and that bears saying. In addition, as researchers we thrive on natural variation, especially when we cannot (for ethical reasons) experimentally cause that variation ourselves. If the population has no variation, there’s nothing for us to study with conventional methodology in psychology. So, as discussed in Cornerstones, Ainsworth’s discovery filled a portal pragmatic function for the emergence of attachment as an empirical “research program” (Lakatos, 1980). The field could now engage not just with serious deviations from species-typical processes but with normal samples from the normal population. What I do intend to do with the foregoing discussion, however, is to help us regain a sense of proportion, with proper weight assigned where due.

It’s helpful to think about the emergence of the dyadic/individual difference aspects of the theory (along with their associated measurement procedures) as a double-edged sword, serving as both the boon and bane of attachment theory (Granqvist, 2020). Our journals, textbooks, lectures, and presentations tend to deal with the boon part, so there’s no need to repeat that here. Regarding the bane part, the undue weight assigned to the dyadic/individual difference components in research, theorizing, and applications, has contributed to many reifications, oversimplifications, exaggerations, and misapplications of the theory (profitably discussed both in Cornerstones and the target articles). What’s worse, these have begun to have tragic consequences for children, caregivers, and families. To give one example from an increasing list, some children have been removed from their caregivers only because social authorities have estimated – via poorly validated attachment measures – that these children have had insecure and disorganized attachments to their caregivers (Granqvist, 2016). This is despite the fact that permanent or repeated removals of children from their caregivers are likely to have more seriously negative effects than “merely” insecure or disorganized attachment does. Indeed, as discussed in Cornerstones, Bowlby’s theory was founded to explain the untoward effects that may stem from prolonged/repeated separations. If we had been more engaged in disseminating the normative features of the theory, then practitioners might not have been misled into thinking that individual differences in security and organization are the “real deal”; by no means do they supersede or replace attachment-normative processes.

A second regrettable consequence of the field’s preoccupations is the weight assigned to how dyadic/individual differences should be measured (observations, interviews, questionnaires?) and subsequently treated (categories or dimensions?). It seems clear that attachment categories, unlike attachments, are not natural kinds, as illustrated by results of taxometric analyses (e.g. Fraley & Spieker, 2003). As discussed in Cornerstones, even Alan Sroufe – a champion of the categorical approach – is willing to engage with the idea that attachment categories are fictions, however useful (cf. instrumentalism as a philosophy of science). It is disheartening to consider how much blood that has been shed over something like this: Fictions, revolving around surface variation, when many foundational issues have been at stake yet overlooked. Like bad religion run amok. Hang down your head, John Dewey (e.g. Dewey, 1998).

The disagreements have been particularly heated when it comes to adult attachment (the developmental/AAI vs social-personality/self-report traditions). Duschinsky is right in Cornerstones to note that I never picked sides in response to the disagreements. Why? First, both threads of adult attachment theory and research are perfectly legitimate (unbeknownst to me, Bowlby apparently agreed, as see Cornerstones). Second, I have deep personal antipathy against most forms of “cancel culture” and rank-closing, especially when the ranks close around variation and fictions rather than foundational matters. Third, there are many parts of the elephant for us blind researchers to explore. Though some ways of exploring might yield a better portrait of the whole animal, we learn more from trying different approaches than from sticking to one. Relatedly, different attachment measures are called for in different situations, depending on research question, design, age-group, access to grant, attachment target, etc. Along with psychometric considerations, pragmatic issues should settle measurement choice.

I call on the field to regain sight of the foundational matters rather than remain absorbed by variation and its measurement. It is particularly important that young researchers are encouraged to do something more creative than devise yet another method to measure attachment variation at some particular developmental period. Enough is enough as far as measures are concerned, and our field can do better than be a “measurement paradigm” (cf. Lyotard, 1984).

Current trends and future prospects: what’s encouraging, what’s not?

It’s encouraging that the field is now coming together in collective efforts, most notably individual participant meta-analysis (IPMA) and consensus statements. As further testimony to Duschinsky’s importance for the field at present, he has been one of the driving forces involved in both initiatives.

First, IPMA is a further refinement of the invaluable meta-analytic tradition initiated by Van IJzendoorn with colleagues and further extended into a grand “meta-meta” analysis in Schuengel et al.’s target article. IPMA is a most welcome contribution because it boosts both power and precision compared to both traditional meta-analysis and individual studies. The power issue is particularly acute at this moment in the history of science because the replication problem (“crisis”, some say) and the concomitant “open” science movement have created a demand for large or small-but-aggregated study samples (see Van IJzendoorn & Bakermans-Kranenburg, this issue). Notably, this demand also partially stems from most psychologists continuing to rely on the frequentist (“p-level”) approach to statistical decision-making, despite its inability to test support for the null-hypothesis and the arbitrariness and historical peculiarities of p < .05 (e.g. Cohen, 1994).

Second, the consensus statements have been needed because of a steady demand for, coupled with increasing misapplications of, attachment-related knowledge in professional practice, such as social and court work. From working on these statements, I have been moved to witness, first-hand, the cooperative stances offered by a large and quite prominent group of academics and clinicians, all of whom – for the greater good – have been able to let darlings slide, not quibble about relatively less important details, and avoid driving a hard bargain, instead generously offering time to provide wise, critical feedback, sometimes even entire written sections. Some say academics are unable to do this; they are wrong (p < .05, as it were). Both of these collective efforts – the IPMAs and the consensus statements – are telltale signs of a promising future. They have rapidly made impact and hold the prospect of putting the science of attachment on a more solid and respected footing while steering its applications in directions that actually promote rather than undermine children’s and families’ best interests.

While generally applauding these collective developments, including the field’s engagement with the “nurturing care” framework (Schuengel et al., this issue), I raise warning flag against the possibility that large-scale collective efforts, building on consensus, may come to dominate the field too heavily in the future. This warning is perhaps particularly pertinent considering the trend for increased impact of intervention-, practice-oriented research and the decreased impact of research on attachment-foundational matters (Schuengel et al., this issue). As in evolution, culture, and society alike, science requires diversity to enable future “progress”. And progress-feeding creativity does not always take place within large application-oriented consortia, which build on what has previously been done, but also within the minds of single individuals and in smaller groups determined to go where others have not gone or won’t go. Therefore, as the field now moves on with large collective efforts heralded by a third generation of public- and practice-oriented research leaders, we must also celebrate – remain open and interested in – the smaller initiatives that will hopefully continue to be undertaken by scholars and students whose gazes are focused elsewhere, such as on diverse cultures, blind spots, and other uncharted territories. As one of the initiators of two recent consensus statements (Forslund et al., 2021; Granqvist et al., 2017), I stress in particular that we must not suppress differences of opinion on pressing matters. Phrased as a heuristic, the field’s progress will be facilitated if the research collective’s cumulative growth of pragmatically useful knowledge co-exists with perceived individual freedom to express bold conjectures and risky predictions – and to make mistakes.

In closing, every worthwhile field of science profits from high-quality historical and sociological scrutiny. Not every discipline is fortunate enough to have scholars such as Robbie Duschinsky do the scrutinizing, however. The field of attachment can clearly benefit tremendously from Cornerstones, along with the many other initiatives undertaken by its author. So I end this with a personal plea: Mr. Duschinsky, please don’t fly away! Or, if you do, return once in a while – will you – to have a peek also at the future developments of this field.

Acknowledgments

I am indebted to Tommie Forslund, the guest editor, and two anonymous reviewers for helpful feedback on previous drafts of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. For pragmatic reasons, like in dialogue with certain audiences, it may still be necessary to pretend that whether or not something is an attachment is a simple question of kind rather than a complex one of degree. Or else children’s best interests could be jeopardized; for example, we could inadvertently sanction that children are re-directed to “attach” to a tablet or smartphone rather than to a real, familiar, individual human being such as a preschool teacher.