Next Steps in Attachment Theory

Abstract

Thanks to the phenomenal success of attachment theory, great progress has been made in understanding child and adult relationships. The success of attachment theory opens the way to new research directions that can extend its successes even further. In particular, more work on the fundamental nature of attachment that respects recent biological research is important, as is concentrated effort on the related caregiving system.

Shaver and Mikulincer (2012) provide an excellent survey of the current state of attachment research and its application to the family. It is incredible how productive Bowlby’s (1969/1982) initial intuition and conceptualization have been over the past half century. As attachment researchers have followed up on Bowlby’s original theory, some of its limitations are becoming apparent. We can see how some limiting elements in Bowlby’s theory and its extensions by others resulted from constraints attributable to the social and intellectual context of his time.

The first constraint on Bowlby’s thinking was that he began to formulate his ideas in the 1940s and 1950s, a time when Freud dominated the conceptualization of childhood (Bowlby, 1958; Karen, 1994). One of Bowlby’s decisions was to distinguish his work from Freud’s by emphasizing proximity as the goal of attachment and by rejecting feeding (Freud’s emphasis) from his conceptualization of attachment (he later expanded the proximity goal to include security; Bowlby, 1969/1982). This choice did not turn out to disadvantage attachment theory, but it had a severe negative impact on the development of caregiving theory because it created a reactive conceptualization of caregiving—caregivers reacted to the child’s distance or the child’s security distress, and proactive behaviors like preparing and delivering food were excluded from the theoretical purview of caregiving—as if protecting the child from starvation or malnutrition had a different motivation from protecting the child from danger.

Bowlby’s reaction to Freud, by limiting the scope of his theorizing, aided him in constructing a compact and consistent theory of how an individual can maintain physical contact with an other who is already disposed to maintain proximity. In the years of development of attachment theory, attachment researchers have relaxed this constraint. Researchers who have applied attachment ideas to adult relationships have mostly come to treat Bowlby’s literal conceptualization of proximity as a metaphor. Thus, most researchers of adult attachment conceptualize emotional closeness and affection rather than simple physical proximity, as can be seen in prominent measures of attachment (Bartholomew, 1990; Bartholomew & Horowitz, 1991; Brennan, Clark, & Shaver, 1998; Feeney, 1999; Hazan & Shaver, 1987).

The second constraint on Bowlby was that he was writing at a time when cognitive views of the sources of behavior were in the ascendency. As a result, he developed his cognitive conception of attachment in two directions. At the level of the day-to-day and moment-to-moment dynamics of attachment he developed a cybernetic model of an attachment behavioral control system that emphasized thinking and deciding (Bell & Richard, 2000a, 2000b). This view reduced much of attachment behavior to a mechanical theory of the regulation of distance and later regulation of security. In general, attachment researchers have found this cybernetic model to be of little research interest. Instead, attachment research has focused on Bowlby’s second cognitive conception, the internal working model, as the new heart of attachment theory. I’ll have more to say about this focus on the working model and resultant attachment styles as strategic choices.

The third constraint on Bowlby’s theorizing was the primitive state of research on the brain and evolution. This limitation is paradoxical because a major reason for the persuasiveness of attachment theory was Bowlby’s sensitivity to biology and evolution. Bowlby had an awareness of evolutionary forces and the biological basis of relationship motivations. Thus, he viewed attachment behavior as emerging from an attachment system in the brain, and he hypothesized caregiving as a complementary brain system. However, more recent neurobiological research has revealed a more complex picture than Bowlby’s original conceptualization. In particular, the fear, caregiving, and attachment systems, though cross-connected, are considered separate systems (Panksepp, 1998).

Although he had no way to know it at the time, Bowlby’s emphasis on proximity in attachment and caregiving resonates well with that has been learned about the attachment and caregiving systems. Both of these systems appear to be strongly associated with oxytocin, a hormone and neurotransmitter that is produced with physical contact as well as during dyadic interaction and communication (Bell, 2010; Uvnäs Moberg, 2003; Uvnäs-Moberg, Arn, & Magnusson, 2005).

Bowlby’s fourth constraint resulted from his early focus on separation and loss (Bowlby, 1973, 1980), and the fear generated thereby. This contextualizing of attachment with separation and loss led to the conceptualization of attachment only in the context of fear (most often expressed by attachment researchers with the milder term distress). Thus, attachment anxiety consists of attachment in the context of the fear of loss of nurturance, resulting from intermittent and inconsistent caregiving from attachment figures. And attachment avoidance consists of attachment in the context of a fear of entering relationships, resulting from prior loss or unavailability of attachment figures. Thus, attachment as the term is currently used is a compound concept, a combination of fear and something else (Bell, 2009, 2010; Main, 1999).

I was happy to read Shaver and Mikulincer’s (2012) discussion of the strength of attachment. Strength of attachment is a concept that many attachment researchers have avoided—or even denied. Some of the most eminent attachment researchers have insisted that attachment is a concept that describes types of attachment, but not amount or strength of attachment (Ainsworth, 1972; Cassidy, 1999; Main, 1999; Sroufe & Waters, 1977; Weiss, 1982). Some authors have argued that attachment can be activated at different levels, but these authors insist that this does not reflect the strength of attachment (Ainsworth, 1982; Ainsworth, Blehar, Waters, & Wall, 1978; Cassidy, 1999). However, the idea that attachment can be viewed as having strength is not new (Berman & Sperling, 1994; Kahn & Antonucci, 1980; Levitt, Coffman, Guacci-Franco, & Loveless, 1994). I am happy with Shaver and Mikulincer’s recognition of strength of attachment because I believe we will find that the strength of attachment is the “something else” that operates in conventional attachment alongside fear. I hope that other attachment researchers will take up this view.

When attachment researchers begin to deal with strength of attachment, this development will, however, require some serious reconceptualization of measurement strategies. As of now, our measures of attachment are categorical, such as the Strange Situation (Ainsworth, 1964), the Adult Attachment Interview (Hesse, 1999) and style description paragraphs (Bartholomew & Horowitz, 1991; Hazan & Shaver, 1987), and dimensional measures of styles (Brennan et al., 1998; Collins & Read, 1990; Fraley, Waller, & Brennan, 2000). These measures all focus on measuring, in one way or another, the amount of fear or the consequences of fear in an attachment relationship; none is designed to measure the amount of attachment. In the current conceptualization of attachment, secure attachment is attachment without fear. Yet there is no room in contemporary definitions for strength or levels of secure attachment (secure attachment is attachment without either fear of losing the relationship or fear of entering into relationships).

My own work, published in this journal (Bell, 2009), has suggested how we might conceptualize strength of attachment after separating it from fear. I have proposed that at base the strength of attachment should be conceptualized as the level of trust a special other. Trust, of course, captures the focus of attachment on security. This trust is a trust in the other’s caring and support, and not the self-interest-based predictability of others’ actions (Fukuyama, 1995; Macy & Skvoretz, 1998; Molm, Schaefer, & Collett, 2009). And I have proposed that the level of trust has a causal impact on openness; the willingness to reveal one’s needs to the special other; and on depending, expressed in Japanese as amae (Bell, 2010; Doi, 1981), or the willingness to depend on the special other for help in meeting one’s needs.

The new model of attachment strength—or trust—will allow researchers to focus on the incremental changes from zero through full-fledged attachment. This focus on moment-to-moment changes in activation strength of attachment should serve to bring attention back to the dynamics of attachment. In fact, attachment researchers seem to have almost completely abandoned Bowlby’s interest in the situational and relational dynamics of attachment (see “Attachment as Contextual and Dynamic” herein).

Caregiving

Bowlby’s fifth constraint was cultural as well as theoretical. Bowlby chose to focus on attachment as the foreground in the parent–child relationship, relegating caregiving to the background. In fact, he devoted only a few pages to caregiving in Attachment (Bowlby, 1969/1982), and most of that section is about retrieval.

Caregiving, however, can be viewed as having preceded attachment, both evolutionarily and dyadically. As I have shown elsewhere (Bell, 2001), caregiving evolved before attachment (imagine the fate of the first mammalian child with a newly developed attachment system who attempted to elicit caregiving from a mother who did not have an already-functioning parental caregiving system). Similarly, caregiving from a parent or other caregiver precedes the development of the attachment system in the child. Oxytocin seems to have a central role in the growth and consolidation of brain neurons in both the caregiving and attachment systems. In mothers, the prebirth production of oxytocin primes her caregiving system to proactively care about the child emotionally and to care for the child behaviorally (Uvnäs-Moberg et al., 2005). Both tender physical contact and focused social interaction also cause the production of oxytocin (Bell, 2010; Panksepp, 1998; Uvnäs-Moberg et al., 2005). Thus, for fathers and others, physical contact and interaction contribute to the production of oxytocin so as to strengthen their caregiving system toward the child. Furthermore, the attachment and caregiving systems can be activated in humans toward a wide variety of special others, including other adults, siblings, and pets.

When it first evolved, caregiving was a very simple system, but after a hundred million years or so, it began to show characteristics associated with the feelings of affection that we recognize in humans. Rabbits with a minimal cortex feed their offspring and huddle with them but will not retrieve them when they wander from the burrow (MacLean, 1990). In rats, caregiving includes retrieving offspring to the nest, Bowlby’s prototypical proximity caregiving system. Active protective caregiving behavior has been observed in the ungulates (hoofed mammals), from deer and elk to giraffes (Altmann, 1963; Masson & McCarthy, 1995). Protective caregiving behavior directed toward young is performed not only by mothers in these species but also by other adult herd members. Such a system corresponds to the safety caregiving system proposed by Solomon and George (George & Solomon, 1999; see also George & Solomon, 1996) and by Cassidy (1999), with a caregiving set goal of maintaining the safety of the partner. Ungulates also retrieve their young and keep them close.

In felines and canines the caregiving system has expanded beyond protection to teaching and emotional support (Lopez, 1978; Rheingold, 1963; Schneirla, Rosenblatt, & Tobach, 1963). This expansion in scope seems to correspond to the set goal of felt security in the Bowlbian caregiving system (Bartholomew, 1990; Heard & Lake, 1997; Sroufe & Fleeson, 1986; West & Sheldon-Keller, 1994).

It is certainly arguable—and has been argued by many authors (Blood, 1969; Erikson, 1963; LaRossa & LaRossa, 1981; Rapoport, Rapoport, Strelitz, & Kew, 1977; Ruddick, 1989)—that human parental caregiving is attuned to all of a child’s needs, from proximity to safety to emotional security and beyond. For a caring parent, it is the full range of the child’s needs that determines the parent’s response. Thus, human caregiving can be viewed as a motivation to meet the needs of a special other. The caregiver responds to the other’s traditional attachment (fear) motivation by being a safe haven. The caregiver responds to the other’s autonomy/exploration motivation by being a secure base. The caregiver also responds to needs for food, clothing, shelter, identity, and validation. This view of the evolution of the caregiving system suggests corresponding expansions in the attachment system. Thus, we can also imagine that the attachment system has evolved across species to respond to this expansion of the caregiving system.

Attachment as Contextual and Dynamic

Beginning to see caregiving as autonomous and proactive thus provides an expanded perspective on attachment. Shaver and Mikulincer’s (2012) description of attachment styles emphasizes avoidant attachment behavior as negative and passive (“deactivated,” in their terms), whereas anxious attachment is “hyperactivated.” This focus on the individual-level internal working model, though clearly accurate as a description of a person’s attachment strategy (to use a military term), glosses over the everyday variation in attachment behavior (what may be considered attachment tactics). As Doi (1981) notes, children get their needs met both actively and passively, both positively and negatively.

When we view attachment as more than a response to distress, we can see how the tactics of attachment mirror the attachment style strategies for all children of whatever attachment style. Most children get their needs met without any effort on their part because of the proactive caregiving of a parent or other person. Thus, any child who has a need met by a caregiver adopts a secure orientation—even a child who is anxiously insecure about being validated can be secure about being fed and clothed. When the child has a need that has not been met by the caregiver’s proactive action, the child can, overtly (“May I have more?” at the dinner table) or subtly (scraping a fork across a plate emptied of food), indicate the need. If the need has been made known to the caregiver but not met, the child will complain gently as a rebuke or loudly as a temper tantrum, an implicit or explicit complaint against the caregiver’s failure to nurture (Bell, 2010; Doi, 1981). This is the tactical equivalent of anxious attachment’s hyperactivation, carried out by children as well as adults of any attachment style. If the need is still not met, the child or adult may move on to a sulk, the tactical equivalent of attachment avoidance.

Next Steps and New Directions

My own view of a profitable direction for attachment theory is to recognize the recent developments in the biology of attachment and caregiving (Panksepp, 1998; Uvnäs- Moberg et al., 2005) as well as cross-cultural forms of attachment (Doi, 1981) and to conceptualize attachment (separated from fear) as the direct counterpart of caregiving (also separated from fear) as a biological system that develops to different levels depending on an individual’s level of experienced nurturance. When we do this, we will see attachment as a general motivation to have our (many) needs met by people we trust. The trust that I refer to here is the trust in the other’s caregiving motivation. At the individual level, we can see some people as having a low level of attachment in general. We will probably have to distinguish those persons whose brains never developed an attachment system (e.g., unnurtured orphans; Chisholm, 2000) from those who developed an attachment system but have suppressed it because of loss or disappointed attachment efforts in the past (Bowlby, 1980). This low level of attachment is characterized as attachment avoidance. As Mikulincer and Shaver (2007) have noted, many avoidant people may have a strong attachment motivation but are afraid or reluctant to express it by depending on others; but as suggested here, some may never have had an opportunity to develop such a motivation.

At the other end of the attachment strength dimension, there are those with a highly developed attachment system who freely ask those they trust to help them meet their needs. When we consider other motivations, we will distinguish persons with high attachment and also a self-reliant self interest motivation, who can easily switch from meeting own needs when feasible to asking for help when needed, from a person with high attachment motivation but a low or suppressed self-interest style, who becomes overly dependent on others. And there are those intermediate on the dimension of strength of attachment.

The heuristic I use to understand caregiving and attachment as biological systems is to imagine the density of attachment and caregiving neurons in the brain (Bell, 2010). More density means more capacity. Individuals who have received high levels of appropriate nurturance will have developed dense, consolidated neurons in the caregiving and attachment systems, and they will have high capacity to activate and express those motivations. Other individuals without experiences of nurturance may never have developed the neurons. Still others may have developed the neurons, but their systems may be suppressed by fear or other motivations. These neuronal developments determine the capacity for caregiving and attachment. But at any moment, one of these systems can be inactivated, can be partially activated, or can be fully activated. I distinguish the level of attachment (or capacity) from the level of activation of the attachment system (Bell, 2010).

Although most attachment research has focused on individual-level trait attachment styles, it needs to be remembered that attachment and caregiving are fundamentally dyadic processes. Toward a few special others, an individual may develop a strong attachment system (often called an attachment bond). Toward most others, the strength of the individual’s dyadic attachment is zero—the person has no reason to trust in the other’s caregiving motivation. Over time toward a given other, a person’s attachment motivation may increase. This is partially a cognitive process as the person begins to perceive the other’s caregiving motivation develop toward the person; but this is also partially a noncognitive emotional process produced by physical contact and social interaction (Bell, 2010; Montagu, 1986).

None of the issues raised here is new to attachment theory. Each of them has been implicit from Bowlby’s earliest work. However, academic-political climates and focus on establishing and verifying early intuitions about attachment moved attachment theory and research in its historical direction. Bowlby’s avoidance of caregiving considerably simplified his theoretical project. But this focus has thereby limited the ability to deepen the conception of attachment by looking at its relationship with caregiving. Similarly, the focus of attachment research since Bowlby on attachment styles as general experience-based strategies has diverted attention away from anxious and avoidant elements within each style as individuals respond to contextual experiences of nurturance and support. There is much more yet to be learned about the interface between attachment and caregiving.