Letters from Ainsworth: Contesting the ‘Organization’ of Attachment

Abstract

As Main (1999) noted in her obituary for Mary Ainsworth, ‘she was interacting weekly by letter and manuscript with her mentor and friend, John Bowlby, and their academic correspondence formed an important part of her, and implicitly our, life.’ These letters of Ainsworth to John Bowlby during the 1980s, available in the Wellcome Trust Library in London, contain valuable reflections on the work of her pupils to extend her system of classifying infant behaviour in the Strange Situation Procedure. Mary Main proposed a D classification and interpreted such behaviour as a breakdown in an ABC strategy caused by a conflict between a disposition to approach and flee from the caregiver. Patricia Crittenden extended the subtypes of A and C, using a developmental model of information processing, and observed that they could be used together in A/C combinations. These letters offer insight into Ainsworth’s intellectual rationale for encouraging both Main and Crittenden in their conflicting endeavours.

Introduction

John Bowlby (1960) observed that separation from an attachment figure would serve as a source of anxiety and trigger the attachment system. Based on this insight, Mary Ainsworth and her assistant Barbara Wittig designed a structured laboratory situation to examine individual differences in the responses of 56 middle-class non-clinical infants aged 11 months to the departure of a caregiver. They termed this the Strange Situation Procedure. Two categories were initially proposed by Ainsworth and Wittig (1969) for coding patterns in infant responses to their study. Infants classified as Anxious-Avoidant (Group A) did not exhibit distress on separation, and ignored the caregiver on their return. Infants classified as Secure (Group B) used the caregiver as a safe base from which to explore, protested at their departure but sought the caregiver upon his or her return. Ainsworth and Wittig noted, however, a third category of behaviour in a minority of cases. These infants, classified as Anxious-Ambivalent/Resistant (Group C), showed distress on separation, and were clingy and difficult to comfort on the caregiver’s return. These patterns were found to be applicable cross-culturally, and predictive of a range of later outcomes for the child (Bretherton, 1985).

From the late 1970s, however, with the support of new video-recording technologies, researchers began to note behaviours that did not fit Ainsworth and Wittig’s classifications for coding Strange Situation Procedures. These behaviours included combinations of A and C behaviour, signs of disorientation, and infants’ fear of their own caregiver. However, as Egeland and Sroufe (1981) noted, most of the infants displaying such behaviours were placed in B, Secure, as the best available fit for classifying their reunion behaviour. What could be the cause and significance of these behaviours? Riding on this issue was the very meaning of attachment, and the validity and reliability of the Strange Situation Procedure. Different answers to this question emerged in the early 1980s from two former pupils of Ainsworth: Mary Main and Patricia Crittenden.

Extending Ainsworth: Main

Main relates that from graduate school she ‘had already been intrigued by odd-appearing behaviors of animals in conflict situations (see Hinde 1966) and – after observing one “unclassifiable” infant in her doctoral study fling her arms about her head while in an anomalous position on parent entrance – Main continued to pursue the problem of “unclassifiable” infants in this light’ (Main, Hesse, & Hesse, 2011). She set out by analysing these behaviours in her sample of infants aged 12 months as extreme or failed responses to the predicament of avoidant infants, given that she expected avoidance to be ‘highly predictive of interactive and affective disturbance’ (1974). This focus must be seen in the context of Bowlby’s (1973) emphasis, in the second volume of his trilogy, on the significance of an infant’s avoidance of the caregiver on reunion as the key indicator of attachment pathology.

On February 1, 1974, Ainsworth wrote to Bowlby that ‘we have found plenty of evidence that the mothers of A babies dislike physical contact, and that it is through behaviour relevant to physical contact that they (at least in large part) express rejection. Mary’s theory is that this puts babies in a double bind, for they are programmed to want contact and yet are rebuffed (or at least have unpleasant experiences) when they seek it. Mary’s hypothesis is that the avoidance (detachment-like) defensive behaviour characteristic of A babies stems from the double-bind.’ Ainsworth wrote to Bowlby that ‘I believe that we now really have a better ‘handle on’ the difference between A and C babies. I must say that this is largely due to Mary Main (letter of March 12, 1976). Beyond this particular analysis of conflict behaviour, Ainsworth’s letters to Bowlby show admiration for Main’s efforts: ‘She is clearly dedicated to her work,’ Ainsworth wrote on September 16, 1976, ‘as well she might be, with such talent.’

Main argued that Type A behaviour is best ‘seen as a necessary shift in attention away from the attachment figure – a shift away from a theoretically irresolvable conflict’ between ‘approach, withdrawal and anger’ (Main, 1981). This shift is framed as a ‘conditional strategy for maintaining proximity under conditions of maternal rejection.’ Avoidance serves proximity in two ways: it keeps the caregiver relatively near without alienating him or her through approach behaviour; it helps the infant ‘gain control over, i.e. flexibility in his own behaviour, a thing he will not have should he... break into disorganised distress’ (Main, 1977). As such ‘this shift of attention is in fact only an attempt to reorganise or to maintain organisation’ (Main, 1981). Type A behaviour is ‘a search for control when disorganisation threatens,’ and is continuous with disorganisation to the extent that it is ineffective at successfully diverting attention from the conflict between approach, withdrawal and anger (Main, 1981). Main therefore theorised Type A behaviour not primarily as characterised by avoidance of the caregiver but as ‘avoidance of behavioural disorganisation’ in the service of proximity (Main, 1981).

In Main’s doctoral research, she had asked her undergraduate coders to compile a list of ‘odd behaviours,’ and these had included ‘hand-flapping; echolalia; inappropriate affect; and other behaviours appearing out of context’ (Main, 1977). This list was expanded and coalesced into indices of a new attachment category through discussions between Main and her graduate student Judith Solomon, who ‘emphasized the importance of the existence of unclassifiable patterns of infant behavior’ (Main & Weston, 1981). Commenting on these investigations of anomalous cases by Main’s lab, Ainsworth wrote to Bowlby on June 6, 1979: ‘I really do think that she is doing exciting things. Very creative.’ From 1980–1981, Main began to note that avoidant infants can at least direct their attention away from the conflicting demands of the attachment system to both approach and flee from the caregiver. By contrast, other infants appeared so overcome by this conflict that they could not develop any coherent strategy for achieving proximity with their caregiver in the Strange Situation. For instance, Main came to reconceptualise ‘approach-avoidance’ behaviour, in which the infant combines proximity-seeking with suddenly veering away, not as an extreme form of avoidance (e.g. Main 1980) but as a sign of what Main and Solomon began to call type ‘D.’

Extending Ainsworth: Crittenden

In May 1983, Crittenden completed her doctorate under Ainsworth at the University of Virginia. Crittenden conducted Strange Situation Procedures with 73 infants and toddlers, most of whom had been severely maltreated. Crittenden noted that ‘not all infants can be placed easily into the three categories described above. As Ainsworth has pointed out (personal communication) it is unlikely that a sample of 56 normal, white, middle class infants would exhaust the possible patterns’ (Crittenden, 1983). Drawing inspiration from the analysis of different forms of cognition in Chapter 4 of Bowlby’s text on Loss (Bowlby, 1980), Crittenden proposed that the A and C responses can be regarded as excluding ‘some classes of information’ relevant to ‘the activation of the attachment system’ (Crittenden, 1983). Crittenden proposed that both A and C behaviours should be seen as strategies for maintaining the availability of the caregiver by ‘interfering with one’s ability to process’ different kinds of information’ (Crittenden, 1983). She disagreed with Main that disorganised behaviour should be expected in maltreated samples.

Though it was not identified at the time, part of the disagreement between Main and Crittenden hinged on different interpretations of the set-goal of the attachment system and the term ‘organised.’ Main used the term ‘organised’ to mean behaviour which aims at physical proximity with the caregiver when the attachment system is activated by stress or anxiety. Notably, this was the technical meaning given to the concept of ‘organisation’ by Ainsworth in her published texts in the period 1969 to 1973 (see e.g. Ainsworth, 1972), as she battled social-learning accounts of the attachment types as merely reflections of different, fixed personality traits. These were precisely the years in which Main was completing her doctorate, and shortly afterwards Main collaborated with Ainsworth on a manuscript, never published, on ‘the importance of physical contact to the attachment behavioural system’ (Main, 1977). In a letter to Brain and Behavioural Science, Main (1979) criticised Bowlby for underemphasising the centrality of proximity-seeking, which should be regarded as ‘the sine qua non for infant survival.’

Crittenden, by contrast, used the term ‘organised’ to refer to patterned behaviour which aims at maintaining the availability of the caregiver as a source of protection. Considering ‘organisation’ as any patterned behaviour oriented towards protection, rather than only proximity-seeking behaviour, was broadly in line with Dante Cicchetti’s use of the term in 1981 (Cicchetti & Serafica, 1981). It was also in line with Ainsworth’s ruminations in her dialogue with Bowlby in the early and mid-1980s, that from around 18–24 months the set-goal of the attachment system might well be the availability of the caregiver rather than physical proximity. Citing the correspondence, Ainsworth would write that ‘Bowlby holds that “availability of the attachment figure is the set-goal of the attachment system in older children and adults.”’(Ainsworth, 1990, citing a personal communication from Bowlby in 1987). This crucial divergence regarding the very meaning of the term ‘organised’ between Main (and 1970s Ainsworth) on the one side and Cicchetti and Crittenden (and 1980s Ainsworth) was later to be acknowledged by Crittenden (2001), who stated that ‘I disagree, however, that an inability to access the attachment figure when under stress implies disorganisation’ and ‘the terms organisation and disorganisation should more closely reflect their dictionary meanings’: the presence of a determinate pattern.

A more expansive interpretation of organisation, together with a much more maltreated and a slightly older sample (the eldest was 24 months, and the average age 13.7 months), helped lead Crittenden to different conclusions than Main. In her dissertation, Crittenden identified behaviours that could be seen as patterned and strategic if understood as aiming at maintaining availability, but which would be seen as not organised if the goal of attachment is taken to be physical proximity. Crittenden argued that ‘the most severely abused infants are not difficult; they are co-operative. The most severely neglected infants are either cooperative or passive/withdrawn...Only the extremely depressed neglecting mother and her infant remain withdrawn, unable to contact one another’ (Crittenden, 1983). Crittenden thus agreed with Main that some infants would not be able to organise a strategy, and that these would be among the most concerning cases (though Crittenden’s elaborated account of such non-strategic behaviour would only appear twenty five years later in Ringer & Crittenden, 2007). Yet, in contrast to Main, Crittenden proposed that behaviours unclassifiable with Ainsworth’s ABC may indeed be patterned strategies for maintaining the caregiver’s availability i.e. ‘organised’ in Crittenden’s sense, though not necessarily Main’s.

For instance, Crittenden noted that ‘those children who were most severely maltreated were found to show an unusual combination of responses in the strange situation which was proposed as a possible new pattern of attachment’ (Crittenden, 1983). The combination of terror and lack of consistency in the caregiving of abused-and-neglected infants produced a distinct pattern which included both avoidant and resistant behaviours: ‘It appears that all abused-and-neglected infants show an A/C pattern as do a few who are only abused and also a few who are only neglected.’ (Crittenden, 1983). Crittenden (1985) cites Ainsworth’s support for the inclusion of the A/C construct: ‘According to Ainsworth (personal communication), both avoidant and ambivalent manifestations might be expected to co-occur in some extremely anxious children.’ Another pattern that Crittenden noted was a cluster of maltreated children, generally slightly older than the others, who appeared to display false positive affect; they were ‘unusually accommodating and can only be classified as co-operative. These babies pose some very interesting questions. Why are these children co-operative when their experience with their mother should provoke a passive or difficult response? And why do so many of them seem concurrently ill at ease?’ (Crittenden, 1983).

Ainsworth’s Reflections

The extraordinary fact which confronted Ainsworth’s students was that the ABC classifications appeared to account for the large majority of middle-class infants cross-culturally. Both Main (1990) and Crittenden (1995: 368) therefore were brought to ask: ‘Why are there only three patterns of attachment when mothers are highly varied?’ On the one hand, the activation of the attachment system when an infant is anxious appeared to be an innate psycho-physiological mechanism. On the other hand, the quality of this behaviour differed as a function of the infant’s caregiving environment in systematic ways.

Main’s approach pivoted on the concept of conflict behaviour, in turn underpinned by a version of Bowlby’s (1969) control systems model of human learning and motivation from the first volume of his trilogy. Main hypothesised that ABC were universal because ‘all infants tend to become attached except under the most depriving of conditions’ (Main, 1974). Once the attachment system is activated by stress or separation, Main suggests that most infants will either express their attachment needs through direct proximity-seeking behaviours (Type B) or direct their attention away from them in order to avoid rebuff from their caregiver (Type A). A small minority, whose needs have been only erratically met by their caregiver, will alternate clingy attachment behaviours with expressions of anger (Type C). Main argued that those infants who are not able to muster one of these responses will show anomalous behaviours, not oriented towards achieving proximity, which may be the displaced expressions of a conflict between fear, anger and a desire for comfort (Type D). By contrast, Crittenden’s approach pivoted on the concept of information processing, and was underpinned by a version of Bowlby’s (1980) account of different kinds of memory as the frame for different forms of human affective and cognitive experience. She theorised that the A and C responses can be regarded as excluding ‘some classes of information’ relevant to ‘the activation of the attachment system’ (Crittenden, 1983).

Main and Solomon’s ‘discovery of a new, insecure disorganized/disoriented attachment pattern’ would be announced in print in 1986; Crittenden’s information-processing, developmental theory would first be published in 1992. Yet, as we have seen, already in 1982 they had come to propose different accounts regarding the effects of fear on child development. Ainsworth, aware of this theoretical divergence, called her two students into her office in Virginia for a meeting in the winter of 1982. In a letter marked December 7, 1982, Ainsworth reported to Bowlby on Main’s visit to Virginia: ‘she and Pat Crittenden (who has been involved with a very deviant group of infants and young children of maltreating families) looked at each others’ tapes of strange-situation behaviour too deviant to fit comfortably into our A/B/C classificatory groups, and this is the kind of thing that can pave the way toward eventual refinement and extension of the classificatory system.’

Ainsworth’s diplomatic approach appears to have had an intellectual rationale in the value Ainsworth saw in the work of both students. Writing later in the same letter (December 7, 1982) of Main and Weston’s work of 1981 investigating the meaning of the ‘unclassifiable’ category, Ainsworth stated that ‘almost each paragraph contains something very important.’ A few months later, she would write to Bowlby: ‘I myself have learned a lot from Pat [Crittenden]’s work – and think she is the best researcher presently working in the field of maltreatment. She does not come up with precisely the same findings that Main and George do – but I think that she knows more about the variety of effects that maltreatment has on a developing child. She has observed maltreating families more directly, over a long period of time, and through more ‘instruments’ (letter dated March 22, 1983). In print Ainsworth described Crittenden’s work as ‘excellent research on maltreated children’ (Ainsworth, 1985).

These meditations on child maltreatment eventually led Ainsworth to formulate a letter to Bowlby in early 1984 expressing her concerns about Main’s ideas. This letter was then, by accident, posted not to Bowlby but to Mary Main! In her obituary for Ainsworth, Main (1999) describes receiving this letter: ‘I discovered a mismanagement had occurred in the form of address, since the enclosure inadvertently began, “Dear John”...The most interesting paragraph included some slightly querulous remarks regarding the new, disorganized/disoriented attachment category.’ The text of Ainsworth’s next attempt to send this letter to Bowlby, dated February 27, 1984, ran as follows:

My reservations about Mary Main’s findings? Well! I will tell you them...

1. She is convinced that I have discovered the three patterns of attachment – that she believes to hold not only for one-year-olds but throughout the life span. This is very flattering. Also I must confess I think that they are indeed the three major patterns. But on the other hand, I can identify subgroups, and believe that both the infants and their mothers as dyads differ from other subgroups e.g. A1 and A2 within group A. Furthermore, I cannot quite believe that apart from the groups and subgroups I have identified there are [not – inserted by Bowlby] other less frequent occurring patterns that may be impossible to comprehend within these three major groups (A/B/C). To say nothing of cross-cultural variations.

2. To be sure, Mary has found cases that do not fit within the A/B/C classifications – and these she has identified as U – i.e. unclassified (or better unclassifiable). She feels strongly that these cases ought not to be forced into whichever A/B/C classification they fit best, but that they form a separate group. However, she also believes that somehow the U patterns merely represent dis-organisations of the A/B/C patterns. So she would classify a baby as U/B2 if the baby seemed to fit the B2 better than any other subgroup, but not yet closely enough to be classified as B2. Just what constitutes the disorganisation is not clear to me, although I might very well be able to see how the baby did not fit the specifications in B2.’

   (To clarify: Bowlby made a handwritten emendation to Ainsworth’s printed letter, clarifying that Ainsworth intended meaning was that she was not convinced that ABC are the only patterns).

Early in 1985, Ainsworth’s doubts about the need for a ‘D’ classification were quelled. At a four day workshop at the University of Washington, ‘Mary Main was the star of the meeting, demonstrating how she identified infants ‘unclassifiable’ in the A, B, C strange situation categories. Various people brought problem tapes and we spent most of the time discussing these. I believe everyone there was most impressed with the need for adding a new ‘D’ or disorganised category to the classification system’ (letter dated February 14, 1985). As she learned about Main’s theories over the course of 1985 and became increasingly impressed with the promise of Main’s Adult Attachment Interview, Ainsworth remarked to Bowlby: ‘You were right that I am in a sense a student of Mary Main’s’ (letter dated December 23, 1985).

Acceptance of the utility of adding a ‘D’ category and delight in Main’s theorising did not, however, inhibit Ainsworth from continuing to support Crittenden in elaborating a theory that did not utilise a D classification. In a letter of February 10, 1986, Ainsworth wrote that from Main’s ‘Berkeley sample the one major antecedent of the disorganised pattern that has emerged is the parent’s lack of resolution of mourning for the loss of a highly significant attachment figure.’ Integrating ideas from Crittenden and Main, Ainsworth hypothesised to Bowlby that ‘one kind of D or Disorganised pattern’ may be ‘an A/C pattern’ caused by ‘frightening super-inconsistency in parental behaviour.’

Into the late 1980s, Ainsworth continued to support both students in their conflicting extensions of her work. She encouraged Crittenden’s focus on developmental adaptation and the subtypes of the A and C strategies that could be observed in maltreatment samples: ‘Pat Crittenden sent me for comment a truly important paper that she has prepared for some rather obscure British journal which draws together all of her thinking and findings together on her various groups of maltreating and adequate mothers and their various relationships, together in an integrated way...it is the best thing she has done yet’ (letter dated August 27, 1987). As well as giving Crittenden verbal support, she urged that Bowlby talk to Crittenden about internal working models and the role of different forms of information processing in shaping attachment patterns. Following Crittenden’s visit to Bowlby in London, Ainsworth wrote ‘indeed that you had the opportunity to hear about her ideas. I myself think she is top-notch, and is making an important contribution’ (letter of August 23, 1988; cf. Crittenden, 1990 citing her conversation with Bowlby).

At the same time, however, looking back at Main’s career in 1988, Ainsworth wrote to Bowlby that ‘she has certainly turned out to be a top-notch speaker as well as an exciting researcher. Despite the fact that I am well informed about her research, I am always agog to hear what she is going to say this time – her talks are always full of drama and intriguing surprises’ (letter dated April 28, 1988). In 1990, Ainsworth (1990) put in print her blessing for the new ‘D’ classification, though urging that the addition be regarded as ‘open-ended, in the sense that subcategories may be distinguished.’ In the 1991 chapter which would be her final publication, Ainsworth reports that her research ‘confirmed the Main-Hesse conclusion that a parent’s failure to resolve mourning for the loss of an attachment figure through death is associated with disorganisation and/or disorientation in infant attachment behaviour.’ However, they also note that ‘5 out of our 15 Ds were not associated with unresolved mourning on the part of the mother. In 3 of these 5 cases we identified sources of unresolved traumatic experiences severe enough to account for the disorganisation of the baby, although in the remaining 2 cases the interview yielded no evidence’ (Ainsworth & Eichberg, 1991, our italics).

Conclusion

Based on Ainsworth’s letters to Bowlby, we can see that Ainsworth saw strengths in both of her protégés’ extensions of her work, despite the fact that they understood the A and C strategies differently, had alternate theories of the set-goal of attachment, and therefore had competing accounts of the effects of child maltreatment. The letters suggest that Ainsworth saw Main as a talented and innovative researcher. She saw predictive value in the ‘D’ construct, though she wondered whether it might at times function as a rather too encompassing a category. At the same time, Ainsworth valued Crittenden’s expertise regarding child maltreatment, her focus on developmental adaptation and sub-type analysis, and encouraged her emergent thinking regarding the A and C strategies as different forms of information processing. The letters also indicate that Ainsworth hypothesised fear and super-inconsistency as both potentially playing a role in the aetiology of behaviours unclassifiable by her coding system. Main and Crittenden have subsequently advanced their thinking well beyond those ideas they discussed with Ainsworth. For instance, Main has explored the link between disorganised attachment behaviours and dissociation; Crittenden has explored the impact of sexuality on attachment and ecological adaptation. Both elaborated theories of trauma in the 1990s. Yet there are also continuities between the ideas of the 1980s and the present, which make Ainsworth’s reflections suggestive for contemporary discussions.