Children, young people and families

"All psychologists want to solve the secret of their history!"

Dr Robbie Duschinsky, University of Cambridge, meets Dr Judith Solomon to hear about her work in attachment.

16 February 2017

Professor Judith Solomon, thank you for sparing the time to speak about the Strange Situation Procedure. It is exciting that you will be running training in coding the Strange Situation in the UK from 3-14 July 2017 at the University of Cambridge. Over the past decades there have been few opportunities for clinicians and researchers based in Europe to learn this measure, so it is a rare opportunity.

Maybe somewhere to start this conversation would be to explain a little bit about the Strange Situation procedure?

The Strange Situation procedure is a standardised laboratory procedure, developed by Mary Ainsworth. Essentially what happens is that 12- to 18-month-old infants face two occasions of separation from and reunion with their familiar caregiver. With each separation the infant experiences slightly more stress, although nothing more than typical in a baby’s life. We then watch for individual differences is whether and in what way infants are able to make use of their caregiver to provide reassurance following these separations. We take this as an indication of security with the caregiver, by which Ainsworth meant the extent to which the infant expects the caregiver to be available and soothing under conditions of threat or stress. 

What does attachment theory suggest generally happens when a baby is stressed or perceives some kind of threat?

For most babies in all cultures studied, when a baby experiences some stress or threat, he or she feels more or less compelled to get into physical contact with the caregiver. When that contact is gained, the baby’s distress generally subsides, and then the attachment system goes into a lower state of activation which permits the baby to do other things, most especially explore and play. Ainsworth termed this the “secure” response: the infant shows that he or she wants proximity and contact when distressed, stops being distressed soon after comfort is available and then turns attention to other things.

You said “most babies” show this pattern of behaviour. If a majority of infants go to the caregiver when distressed in the Strange Situation and can be comforted – what do the other infants do?

Before developing the Strange Situation, Ainsworth first did a major observational study of mothers and infants in Uganda. She saw a few babies who didn’t act as expected. Rather than going to their caregiver when distressed, they avoided their caregiver and did not communicate their distress to him or her.

When she set up the standardised Strange Situation in the US, Ainsworth found the same thing. About 15-20% in the US samples, even when distressed by the repeated separations, did not go to the parent or communicate their distress. They looked or moved away, paying attention to the toys. Ainsworth termed this an “avoidant” response. John Bowlby’s attachment theory, the theory upon which Aisnworth’s observations were based, predicts that all babies should be upset by separation from their caregivers. And physiological studies, such as of heart-rate, show that these infants who seem unruffled are indeed actually quite distressed! Initially, some people wanted to say that these babies were simply unperturbed by separation, but in fact their heart rate was going zooming upwards. Apparently, they were “hiding” their distress from their caregiver. As it turned out, this made sense: Ainsworth specifically found that the caregivers of infants who show an “avoidant” response in the Strange Situation tended to be less sensitive and more rejecting in response to infant displays of distress.

She had been doing very careful home observations of the small group of 23 babies involved in her first Strange Situation study. Her sample was expanded, but she started with just 23 babies. She had sent her students to do observations once a month in the home looking at the mothers and babies together over the first year of life. So in fact she had access to a great deal of knowledge about these mothers and babies at the time she ran the first Strange Situations.

Now a third, smaller group of babies in Ainsworth’s original small sample were on the opposite end of the spectrum in terms of how they behaved in the Strange Situation. They wanted to be right next to their familiar caregiver right from the start of the Strange Situation. And after separations, they couldn’t settle back to play – they were angry and distressed. Because they had this quality of anger in their interactions with their caregivers, Ainsworth called this a “resistant” response from infants.

Ainsworth believed that the infants who did not communicate their upset to their caregiver (“avoidant”) and those that couldn’t be comforted and showed anger (“resistant”) in the Strange Situation had something in common: she believed that both responses were evidence of the infant’s lack of confidence that their caregiver would be available to them and responsive to their signals. She therefore termed these responses as “insecure” and contrasted them to the “secure” response of seeking comfort directly and being able to be comforted, which is organised by trust in the availability of the caregiver.

Secure, avoidant and resistant are the classifications for the Strange Situation introduced by Ainsworth, and which became well known.

What kind of impact did Ainsworth’s observations make in the psychological research of the period?

Psychoanalysts of the time didn’t think it was possible for babies to show defences against distress: they thought of defence as a product of ego development. So Ainsworth’s observations of infants avoiding a caregiver they might reasonably expect to reject them caused quite a bit of controversy and discussion. This line of thinking was developed further by Ainsworth’s student, Mary Main, who talked about the “avoidant” and “resistant” responses as conditional strategies, in the sense used by evolutionary biologists. In calling them strategies, Mary meant that these patterns of response are (not necessarily intentional) ways that babies have  of organising their behaviour with respect to the caregiver that permit them achieve what availability is possible from a caregiver expected to show various kinds of insensitivity. The strategies allow the baby to get something of what they need from the attachment figure even though they may not be getting the whole experience associated with secure attachment – of having warm and reassuring contact and being permitted to explore at their own pace.

Later researchers have found that these different responses or “strategies” in the Strange Situation have predictive value. What among this research has particularly interested you?

Different classifications in the Strange Situation have been found to be predictive of a variety of outcomes. These differences have effects on children’s functioning, for example in terms of social competence, behaviour in pre-school, levels of aggression. They can also predict differences in symptoms of later mental illness.

To me personally, the most powerful line of research that has come out of it is looking at the stress physiology and the mental health consequences of security as opposed to insecurity. We know now, mainly from work begun by Megan Gunnar but also many other people, that you can see that when a baby is stressed, for example in the doctors office, having their attachment figure with them, if you measure stress in term of cortisol increase, will lower that stress, and this is particularly true for secure babies. When secure babies (secure as defined in the Strange Situation) go to the doctors' office with their mother, you can see that cortisol levels certainly don’t go up as high as other babies in that very stressful circumstance. 

Ainsworth’s work with the Strange Situation has some counterintuitive conclusions!

If a child is distressed by separations – one might think that this is bad and a sign of a lack of resilience. But for Ainsworth the inability to show distress when appropriate indicated insecurity. Likewise, if a child was is clinging the whole time and not letting go – one might think that this demonstrates affection. But for Ainsworth it is regarded as a sign of insecurity, of lack of confidence in the caregiver.

There was a fair amount of criticism of Ainsworth’s work. Some alleged that behaviours that she attributed to an infant’s expectations about their caregiver’s response to attachment signals were really temperament differences – that some babies simply were predisposed to cry less, and others more. However Ainsworth found that the infants who were “avoidant” in the Strange Situation in fact were especially likely to cry and show anger at home. These were not personality traits, but the result of expectations about the relationship based on past experience. These expectations are certainly capable of change. But researchers found that they could also remain stable over long periods, and even across generations.

Clearing up that controversy took a long time and has had a major impact. It has helped people realise that early caregiver/baby interaction can result in patterns of relationship, and that these can have important consequences for later development.

Is it obvious that an assessment that takes between 22-25 minutes of a separation/reunion paradigm would be able to predict such a wide variety of different outcomes?

Certainly it’s not obvious at all. It’s amazing. That is why having training in the measure is important rather than using the classifications loosely. It’s important to be able to observe this behaviour very carefully and very reliably because essentially what you’re trying to do is predict major aspects of baby’s social development from a little 20-minute segment. The reason it’s possible is because training allows you to find the signposts of these patterns discovered by Ainsworth. I suspect that clinicians who don’t have this training can no doubt see important things, but will also miss quite a bit because they don’t know what to look for. Being able to identify security and differentiate it from insecurity can make a huge difference for making clinical judgements, providing an anchor for these judgements.

Both as a clinician and as a researcher, attachment has been a key concern of yours. What was it that drew you initially attachment theory?

I didn’t start out there. I started out in my graduate work as a comparative psychologist, specialising in biological psychology. I continued to do that as a graduate student in my first few years when I got to Berkley. I knew nothing about attachment until my advisor, a comparative psychologist, said “Look, you are about to take your qualifying exams, and you need someone to be on your committee who is not a comparative psychologist so why don’t you go and meet Mary Main and work with her and have attachment as one of your areas of sub-concentration?” So that’s what I did and of course Mary Main had me read John Bowlby’s original work on attachment theory and Ainsworth’s work on the Strange Situation. And I worked in Mary’s lab. It was a revelation to me, that I could use my interest in evolution and comparative psychology and make it a foundation of my interest in babies and parents and the development of personality. All psychologists want to solve the secret of their history! Okay, maybe not social psychologists, but everyone else. I was bowled over, and after my qualifying exams I came to work more with Mary Main.

One of the pieces of work that you’re most well known for is the introduction, together with Mary Main, of a fourth classification for the Strange Situation in addition to Ainsworth’s three. 

We introduced a classification called insecure-disorganised/disorientated. It captures infants who show behaviours that don’t fit the Ainsworth classifications and instead suggest conflict, fear or confusion in relation to the caregiver. The classification emerged over many years, mainly during my time in graduate school and a little bit after that. My attention to these kinds of behaviours came in part because I was learning how to do Ainsworth’s classification system at the same time as I was looking at a sample of abused babies. Working both with Mary Main’s normative sample and certainly looking at the abused babies, I kept having trouble making classifications. I kept seeing things that made me think: “this is not secure, this is not avoidant, this is not resistant - this is something else”. And it was particularly common among the abused babies.

About the time that this was happening, many other people, especially those working with high risk samples, were similarly struggling with what to do with babies who didn’t fit the Ainsworth classifications. Pat Crittenden and and Karlen Lyons-Ruth, for instance, were struggling with how to classify babies who were definitely maltreated and whose behaviour didn’t look simply like the familiar Ainsworth patterns. In psychology usually you just throw anomalous cases out, assume that there was some measurement error. But the fact that these cases were more common in maltreatment and at risk samples made us think that there may be something important here. I was also heavily influenced by my background in ethology and comparative psychology to believe that behaviour – even rather odd and surprising bits of behaviour – should be assumed to be meaningful, even if the meaning is not obvious on the surface.

The way Ainsworth classified babies, and the way people continue to do it now, is not by having a set of very firm a priori guidelines, but by having a template that encompasses many different variables. By comparing a particular case to your template you can decide if the baby’s behaviour fits in to any of groups. So if the baby’s behaviour does not fit any of your templates then you begin to consider that you have something else going on. Since I already had begun taking extremely compulsive notes, Mary Main asked me to continue doing that and we developed a set of indices or guidelines for deciding whether a baby was in fact classifiable into the secure, ambivalent or avoidant group or had to be set aside, and, ultimately, we made a new disorganised/disoriented classification for these infants.

When children were showing these kinds of behaviours what did you and Mary Main think was going on at the level of mental processes?

The behaviour of these babies, in the first instance, seemed not make sense. For instance, if a baby is otherwise showing a secure pattern, representing the direct expression of the desire for comfort when distressed, why might this be suddenly interrupted without a prompt from the caregiver by standing extremely still? Another baby might show bits of avoidance together with bits of approach behaviour: I remember one baby beginning to approach the mum, but then getting down on her tummy and crawling backwards to her mother. She seemed to be wanting to avoid her mother, but to be unable to do so as she was also compelled to approach. 

Those are examples of behaviour that seem to be in violation of the idea of a coherent strategy, in which the behaviour can be seen to be organized in terms of achieving what availability can be expected from the caregiver. That was, at any rate, our early thinking. 

Later, though, the theory became refined. Mary Main’s group proposed that having a parent who was alarming to the baby put the baby in a paradoxical situation. Infants are wired to want to go to their caregiver when alarmed. But what if the caregiver is themselves the source of alarm? The baby can’t go to someone who alarms them, but equally the baby is compelled to want to go to them – with each escalating the intensity of the other. Influenced by Robert Hinde who had documented odd behaviours in animals experiencing conflict between motivations to approach and avoid, Hesse and Main suggested that the conflict to approach and flee an alarming caregiver is expressed in the odd behaviours used to classify disorganised attachment. It is important to keep in mind that caregivers can alarm their baby for many different reasons, not necessarily reflecting physical abuse. For instance, a traumatised parent may dissociate and be suddenly unavailable to their baby for substantial periods. This explanation, though, helped make sense of why disorganisation was especially common in maltreated samples, as maltreatment could be expected to directly make a baby alarmed by their caregiver.

At the same time that I was working on the disorganised classification, Mary Main was working with other of her students and trying to capture aspects of representation of relationships. And she found that the coherence of the autobiography told by caregivers was associated with the behaviour of their children in the Strange Situation. It was an incredible finding. Unlike Ainsworth, we didn’t have home observations: we didn’t know what these mothers and babies were doing in the home. But Mary Main found that many of the parents who seemed to have unresolved loss had babies that we had placed in to the disorganised group.  So that was an important hint that there was a connection between the baby’s behaviour and something of this experience with the mother. But it took a while for that all to be fleshed out into a full theory, incorporating in full the role of frightening and dissociative caregiver behaviours in causing behaviours coded as disorganised.

Thirty years after the introduction of the classification, what relevance does the classification have for researchers and clinicians now?

In terms of research it has emerged that children who are classified as disorganised in infancy or pre-school are more likely to be aggressive or hostile. There also has been a link identified by some people between a disorganised classification in the Strange Situation and later dissociative behaviour. Both findings have relevance for understanding the development of some types of mental illness, whereas the links between maladaptation and the avoidant and ambivalent classifications have not emerged very strongly. The latter finding has been a surprise to many in the field, whereas it is fair to speculate that the predictive power of disorganized attachment is its relation to concurrent trauma or the parent’s own earlier traumatic experiences. This is consistent with what has been a growing understand in psychology and psychoanalysis on the powerful role of trauma in shaping pathology.

Subsequent to your work with Mary Main, you went to train as a clinician. It would be interesting to hear about how knowledge of the infant attachment classifications was relevant or helped you in your work as a clinician.

It was immediately of use. I trained at the Infant-Parent Program, a psychoanalytically informed kind of early intervention which was the brainchild of Selma Fraiberg. Thanks to understanding the avoidant and resistant strategies, I was able to see a lot about how an infant’s expectations of the caregiver grew into entrenched aspects of their relationship, which they both sustained in a way. And having been sensitised by Mary Main to thinking about fear, I was able to see in my parent-child parent-infant cases the ways in which the child was afraid of the parent and often the parent was afraid of the child.  These ideas allowed me to explicate a lot of what was going on in terms of parent-child dynamics and then think about how to address those difficulties. And this is still the case for me as a clinician.

Why might classification be something that clinicians or researchers are interested in today? We’ve seen over the last 15-20 years quite a rise in the idea that you can’t put humans in boxes; humans are too heterogeneous to be meaningfully put in to categories.

Nobody can disagree with that notion. Of course people can’t be put into categories in any neat way. But as humans we like to see the world in terms of categories. It’s a part of our evolutionary inheritance, how our minds are set up. A key reason for this is that when we use categories it is then easier to see multidimensional patterns of behaviour or of relationships, even if we are working from idealised templates. In a way you can say it is heuristically useful to have these classifications.

But there’s an additional reason. It’s not just that it’s heuristically useful. Since categorical distinctions are also used in how people think – such as yes/no, good/bad – they come to structure reality. “Yes, I have a good mummy” or “no, I don’t have a good mummy”. Children and babies obviously think in those kinds of terms, and there is a kind of psychological reality to creating categories on this basis.

That being said, using categories has serious problems. It can reify flows of behaviour that are unique to circumstances and relationships. It can lead us to lose track of all the things that change in the lives of babies and children and adults. We can certainly miss a great deal if we think only in terms of classification. In terms of statistical work within research, categories are also much more unwieldy to work with and reduce statistical power.

If categories have these problems, though, then why would someone want to come and learn how to make classifications of the Strange Situation?

Because if you’re interested in attachment theory, or if you’re interested as a clinician in making sense of the relationship that you see playing out before you, it helps to have a very clear notion of these basic types or templates. You can’t see variations in any coherent way if you can’t see central tendencies. I have certainly seen clinicians make what I think are unfortunate mistakes because they don’t have this information, and can’t sort out and make sense of the complexities they are dealing with. For instance, they may lack a sharp template for what security looks like, and so don’t have a benchmark for what aspects of a relationship to support or cultivate. Correlatively, lacking a clear image of ordinary variations, they are at risk of “pathologizing” expectable difficulties in parent-child interactions.

Besides learning classifications, what else might someone learn in the course of training in the Strange Situation?

In the course of the training, I talk a lot about attachment theory and behaviour, about feelings and representations of others in relationships. I talk about clinical examples of attachment patterns and behaviors that the clinicians will likely encounter in their work, especially clinicians who work with parents and babies but also with people who work with adults. 

There has been a whole variety of applications of the Strange Situation in care proceedings, custody proceedings, clinical work and assessments and a whole variety of different kinds of research. Is there anything in particular you would wish to draw out as particularly exciting or interesting or valuable usage and also anything you’d like to draw out as usage of the classifications system that you think you might have cautions about?

Because attachment theory has become so well accepted and so pervasive, people have tended to want to think with and use attachment classifications whether it’s for research or for the courts or for intervention with high risk parents. 

But it is a major drawback if clinicians they don’t know the Ainsworth’s system that it is all based on! A little knowledge can be especially dangerous in this area. And reading about Ainsworth’s research, though I would absolutely encourage this, is only part of the picture. You can read about a child avoiding, you can read about a child being secure and calming quickly with the parent, but if you don’t know what it looks like and what it doesn’t look like and what it looks like when it is distorted or disrupted you are really hampered in your thinking or assessment. Your templates are likely to be based on your presuppositions, for instance about what  the words “avoidance” or “sensitivity” or “disorganisation” mean – not what researchers have actually been measuring.

Because so many maltreated children and so many mother/baby relationships in other kinds of high risk situations tend to be classified disorganised, that classification has become very important. Clinicians and social workers may identify behaviour they regard as “disorganised”. But without training, it is not at all clear that what they are identifying is related to the evidence base from research. This can lead people to go really wrong, even with good intentions. I would encourage clinicians and researchers who want to really understand attachment and to link their work to the research that has taken place over decades to consider learning to code the Strange Situation. It is the foundation on which our field has been built. Similarly, so many researchers and clinicians are fascinated by research or clinical adaptations of studies of adult representation of attachment. But adult patterns of representation and defense have their origins in parent-infant interaction, directly or indirectly. It helps enormously to understand adult patterns when you see them in their basic, elemental forms in the behaviour of infants in the Strange Situation.

- Dr Judith Solomon holds a Visiting Research Fellowship in the Department of Public Health and Primary Care at the University of Cambridge. She is known for her research and theory-building on attachment and caregiving, including the introduction of the disorganised attachment classification and the first longitudinal study of infants in separated and divorced families. With Carol George, she served as co-editor of Attachment Disorganization (Guilford Publications,1999) and Disorganized Attachment and Caregiving (Guilford Publications, 2011).