what about attachment parenting?

During my excursion into the world of child development theory, I’ve been acutely aware of one important group I’ve so far overlooked; the advocates of attachment parenting (AP). Attachment parenting is an approach derived from John Bowlby’s theory, but includes practices like extended breastfeeding, ‘babywearing’ (carrying the baby close to the parent’s body), co-sleeping, and a high level of responsiveness to the baby’s needs. I must admit to being sympathetic to AP as an approach; I must also admit (like many other parents) to being something of an attachment parenting backslider. Both my children were in a buggy by the age of six weeks because I couldn’t find a sling that didn’t hurt my back, due to a series of complications I stopped breastfeeding my son at four months, and one child slept happily in bed with me whereas the other simply wouldn’t.

A significant influence in the AP movement is William (‘Dr Bill’) Sears, a paediatrician and author of a number of best-selling parenting books. Dr Sears’ wife Martha is a director of Attachment Parenting International (API), an organization founded in 1994 by Barbara Nicholson a learning disabilities specialist, and Lysa Parker, who also has a background in special education. Parker also worked with La Leche League International, as did four other of the API’s eight directors. (La Leche League was founded in Illinois in 1956 by a group of mothers who wanted to promote breastfeeding, a practice then rapidly dwindling in the USA.)

Dr Bill’s shoes have to an extent been filled by Gordon Neufeld, founder of the Neufeld Institute in Vancouver, Canada. Unsurprisingly, Neufeld’s approach is firmly grounded in attachment theory, with separation, for example, being seen as playing a key role in the ‘epidemic of anxiety’ that apparently afflicts our children. Neufeld’s ideas are very popular, but it’s instructive to read what a group of parents who have used the AP approach with their special needs children have to say here.

Why attachment parenting?

The predominant reasons given for the attachment parenting approach are that it involves ‘instinctive’ and ‘natural’ child-rearing practices that parents have used for thousands of years. In essence this is the same argument used by Bowlby and Bruce Perry in support of their models of child development. I think it’s flawed in two respects. One is that ‘instinctive’ behaviours have also led to the adoption of practices that attachment parenting advocates are unlikely to approve of, such as infanticide, cranial deformation and genital mutilation. You could argue, of course, that such practices might be instinctive but they’re not ‘natural’. Unfortunately, ‘nature’ isn’t always benign either.Throughout human history, infants in hunter-gatherer societies have been at high risk of death from predation, starvation, injury and infection. And if hunter-gatherer societies appear to be healthier than those in the developed world, it’s often because of the price paid by their ancestors and the weaker members of their community. Although AP proponents often advocate as natural a lifestyle as possible, few in the developed world would expect mothers and children to be self-sufficient, live in huts, cook over open fires or fail to take advantage of modern medical interventions if required.

Evidence or belief?

In other words, neither the ‘instinctive’ nor ‘natural’ justifications for attachment parenting provide sufficient evidence to support it. I suspect that rather than AP consisting of practices based on a careful evaluation of the evidence, AP is actually based on a set of beliefs. They’re not unreasonable beliefs; breastfeeding, carrying babies in a sling and being responsive to a child’s needs are highly likely to be of benefit because they optimize nutritional intake and reduce the risk of gastro-intestinal infection, keep the infant warm, comfortable and within the parent’s sight, and increase the likelihood of the child’s needs being responded to promptly. In other words, they are beneficial to the child for good, demonstrable reasons, not because they are ‘instinctive’ or ‘natural’. Other practices are less obviously beneficial; Attachment Parenting International has issued safety guidelines in respect of children sleeping in the parent’s bed, for example, and numerous parents have described attachment parenting resulting in outcomes such as a bad back and sleep deprivation.

Personally, I have no problem with parents basing their parenting approach on a set of beliefs; as long as children aren’t harmed, parents are free to bring up their children as they see fit. They are also entitled to share their beliefs with others and to try to persuade them that those beliefs are right. I don’t even have a problem with practitioners such as Sears or Neufeld advocating a particular approach to child-rearing, because parents and practitioners committed to an approach don’t generally claim to have carried out an exhaustive evaluation of the relevant evidence. It’s usually fairly obvious that they are propagating beliefs and are using evidence selectively to back up their views. What I do find worrying is researchers adopting the same approach. Even though their credentials suggest they are able to evaluate all the relevant evidence in a reasonably objective manner, some follow the same strategy as attachment parenting advocates and base their theoretical models on beliefs backed up with supporting evidence only. What’s also worrying is that an uncritical evaluation of the evidence appears to be acceptable to some peer-reviewed journals.

In a previous post I compared Leo Kanner’s and Bruno Bettelheim’s approaches to the early evidence relating to autism. Kanner’s initial model of autism was based on his evaluation of evidence in the light of contemporary child development theory. As time went by, he repeatedly revised his model as new evidence became available. Bettelheim, by contrast, not only based his model on a belief about the cause of autism (parental behaviour), but also used evidence selectively to support it, and his background in philosophy appears to have persuaded him that this was valid way of handling evidence. I don’t know if Bettelheim’s approach is embedded somehow in the world of child development research, but it keeps cropping up amongst child development researchers. Bowlby’s essential dismissal of genetics and unquestioning acceptance of the idea that the origins of psychiatric disorders are to be found in childhood experiences, contrasts starkly with his painstaking analysis of concepts such as instinctive behaviour and emotion. Allan Schore goes into considerable technical detail about the development of the orbitofrontal cortex, but doesn’t question Bowlby’s model of attachment, or the findings of researchers he uses to support his orbitofrontal model, despite Bowlby’s theory and the function of frontal areas of the brain being the subject of considerable debate. Perry’s discussion of the evolution of human socio-emotional behaviour is pretty speculative, and although on the face of it his brain scan findings are persuasive, digging a little deeper suggests that he hasn’t paid sufficient attention to defining concepts such as neglect, nor to the range of other possible causes of abnormal brain development.

Carrot or stick?

Another issue that’s concerned me is what the attachment model is used for. AP proponents tend to take a resoundingly positive approach to parenting – AP practices are recommended because they are seen as being good for babies and their families and ultimately, society at large. Parents are encouraged to use strategies flexibly and adapt them to their own lifestyles, even though some AP supporters might get a bit over-zealous and risk making parents feel needlessly guilty. When attachment theory filters through into public policy, however, a rather different picture emerges; one in which there’s a real risk of attachment theory being used as a stick with which to beat parents. Poor attachment is blamed for poor health, antisocial behaviour and psychiatric disorders, locating the source of those problems firmly within the family, particularly with parents and notably mothers.

Despite references being made in policy documents and textbooks to Bronfenbrenner’s ecological systems theory, his framework is rarely applied except in the sense of recommending interventions at the community, state or international level to prevent or compensate for the damage caused by poor parenting. So far, I’ve found little consideration given to the possibility that problems manifested by individuals could have causes at all levels, from genes through physiology, to economic, social and cultural pressures at the community, national or international level.

In my next post, I look at a book by a critic of current child development theories, John Bruer’s The Myth of the First Three Years.

Acknowledgement: with grateful thanks to Jennifer Skillen for information about the history of La Leche League.

another post about nature and nurture: Allan Schore’s model

Allan Schore’s work is widely cited in the child development and child protection literatures – notably in documents relevant to public policy. In this post I look at Schore’s paper ‘Attachment and the regulation of the right brain’ because it deals directly with Bowlby’s theory of attachment. Schore seeks to map Bowlby’s model of attachment onto recently acquired knowledge about how the brain develops, and a shortened version of this paper forms the foreword to a reprint of Bowlby’s classic book Attachment.

Schore is described on his website as “the American Bowlby”, so I expected to see similarities between his model of attachment and Bowlby’s. There is indeed a resemblance, but Schore differs from Bowlby in the use of three terms; ‘emotion’, ‘imprinting’ and ‘environment of adaptedness’. These differences sound trivial, but in fact are quite important.

Schore summarises Bowlby’s model of attachment as follows (page references are for Attachment unless otherwise indicated);

“…that attachment is instinctive behavior with a biological function, that emotional processes lie at the foundation of a model of instinctive behavior, and that a biological control system in the brain regulates affectively driven instinctive behavior.” (Schore p.24)

Emotion

A key departure from Bowlby is encapsulated in the phrase ‘emotional processes lie at the foundation of a model of instinctive behavior’. Schore’s reasoning goes as follows;

“Bowlby emphasizes the salience of ‘facial expression, posture, tone of voice, physiological changes, tempo of movement, and incipient action’ (p.120). The appraisal of this input is experienced ‘in terms of value, as pleasant or unpleasant’ (pp.111–112) and the movements ‘may be actively at work even when we are not aware of them’ (p.110); in this manner feeling provides a monitoring of both the behavioral and physiological state (p.121). Emotional processes thus, he says, lie at the foundation of a model of instinctive behavior.” (Schore pp.28-29)

Firstly, I think Schore differs from Bowlby over emotional processes lying at the foundation of anything. Bowlby’s reasons for avoiding using the term ‘emotion’ are pretty clear; he sees emotion as part of a range of ‘feelings’ and those feelings as being part of an organism’s appraisal of its environment; feelings would include hunger, thirst and discomfort – not usually classified as emotions. To clarify his point Bowlby quotes philosopher Susanne Langer’s analogy of the heating and cooling of iron:

“When iron is heated to a critical degree it becomes red; yet its redness is not a new entity which must have gone somewhere else when it is no longer in the iron. It was a phase of the iron itself, at high temperature.

As soon as feeling is regarded as a phase of a physiological process instead of a product of it – namely a new entity metaphysically different from it – the paradox of the pyschical and the physical disappears.” (p.108)

Certainly emotional processes (or more accurately ‘feelings’) are an important part of Bowlby’s model of attachment because the expression of feeling is the means of communication between infant and carer. But the phrase ‘emotional processes lie at the foundation of a model of instinctive behavior’ to me implies that in order to be foundational, emotional processes must have an existence independent of the process of appraisal – exactly the opposite of what Bowlby and Langer are saying.

The second point is that Schore doesn’t say that emotional processes lie at the foundation of a model of attachment, but at the foundation of a model of instinctive behaviour. I suspect many of the zoologists cited by Bowlby in his analysis of instinctive behaviour would disagree. Even according to Bowlby’s definition (which excludes reflexes) many behaviours he would consider instinctive don’t actively involve emotion as far as we can tell. He cites for example, imprinting (e.g. ducks and geese following the first object they see after they hatch) and nest-building as instinctive behaviours, but although these behaviours might be associated with emotion (difficult to assess in birds) there is as far as I’m aware no evidence that emotional processes are foundational to them. Of course it could be argued that Schore is referring exclusively to humans, in which case he would be departing from Bowlby’s model again, because Bowlby clearly sees human instinctive behaviour as on a continuum with instinctive behaviour across the animal kingdom.

Despite Bowlby explicitly choosing not to use the term ‘emotion’ Schore’s paper contains 69 mentions of it in 26 pages. Here are a couple of them;

“… Bowlby’s scientifically-informed curiosity…envisioned the center stage of human infancy, on which is played the first chapter of the human drama, to be a context in which a mother and her infant experience connections and disconnections of their vital emotional communications.” (Schore pp.23-24)

I’m not sure this Bowlby actually did this, since he saw emotional communications as only part of the complex interactions involved in attachment, he wasn’t happy with the use of the term ‘emotion’, and goes into some detail about why he doesn’t use it.

“… in his second volume Bowlby (1973) attempted to define more precisely the
set-goal of the attachment system as seeking not just proximity but access to
an attachment figure who is emotionally available and responsive.” (Schore p.26)

I haven’t read Bowlby’s two later volumes (Separation and Loss), but searching their text on-line shows Bowlby using the word ‘emotionally’ descriptively in case studies, but only one instance of him referring to the mother being “physically present but ‘emotionally’ absent” (Loss, p.43; the quotes around ‘emotionally’ are Bowlby’s). Again, he’s guarded about the use of this word and doesn’t seem to see emotion per se as central to the mother-child interaction.

Imprinting

‘Imprinting’ is a term that has several different meanings. In studies of development, it refers to an animal learning to recognise the features of a particular stimulus – regardless of what happens as a result. Konrad Lorenz’s famous experiments showed that goslings imprint on the first object they see, whether that object is their mother or Lorenz’s boots. But Schore doesn’t use ‘imprinting’ in quite the same way.

“Attachment theory, as first propounded in Bowlby’s (1969) definitional
volume, is fundamentally a regulatory theory. Attachment can thus be conceptualized
as the interactive regulation of synchrony between psychobiologically attuned organisms. This attachment dynamic, which operates at levels beneath awareness, underlies the dyadic regulation of emotion. Emotions are the highest order direct expression of bioregulation in complex organisms (Damasio, 1998). Imprinting, the learning process it accesses, is described by Petrovich and Gewirtz (1985) as synchrony between sequential infant maternal stimuli and behavior…” (Schore, p.34)

I think this paragraph simply lumps together attachment, emotion and imprinting and assumes they form a coherent whole. The first part of the paragraph – about attachment – makes sense in the light of Bowlby’s explanations. But no supporting evidence is offered for the claim that attachment underlies the dyadic regulation of emotion, and Damasio’s claim about emotions doesn’t necessarily have anything to do with dyadic interactions. And the final sentence is unclear about what ‘it’ refers to or why imprinting necessarily involves maternal stimuli or synchrony – young birds have been reported imprinting onto inanimate objects.

Emotion and imprinting are lumped together again here;

“Emotionally focused limbic learning underlies the unique and fast-acting processes of imprinting, the learning mechanism associated with attachment, as this dynamic evolves over the first and second years. Hinde (1990, p.162) points out that ‘the development of social behavior can be understood only in terms of a continuing dialectic between an active and changing organism and an active and changing environment.” (Schore pp.30-31)

Again, an unsupported claim is made about emotion and imprinting and Hinde’s comment about social behaviour being an interaction between an organism and its environment involves a dialectic much broader than that between an infant and primary caregiver.

Another anomaly in the use of the term ‘imprinting’ occurs here;

“… I offer evidence to show that attachment experiences, face-to-face transactions of affect synchrony between caregiver and infant, directly influence the imprinting, the circuit wiring of the orbital prefrontal cortex, a corticolimbic area that is known to begin a major maturational change at 10 to 12 months and to complete a critical period of growth from the middle to the end of the second year.” (Schore p.30)

‘Imprinting’ has a number of different meanings, including psychological imprinting (the way it’s usually used in studies of development), but it can also mean to impress or make a mark, so can be used in reference to neural circuits. It’s quite likely that imprinting in the behavioural sense is associated with imprinting in the neural sense, but to use the same term with different meanings without clarification is confusing.

Environment of adaptiveness

Bowlby frequently refers to what he calls the ‘environment of adaptedness’, the environment that shaped human instinctive behaviour. In other words, it’s an environment to which the behaviour is adapted (past tense). But Schore uses the term ‘environment of adaptiveness’ instead. I assumed he must explain elsewhere why he does this, until I noticed that he misquotes Bowlby;

“More specifically, it [the control system for instinctive behaviour] evolves in the infant’s interaction with an ‘environment of adaptiveness, and especially of his interaction with the principal figure in that environment, namely his mother’ (p.180).” (Schore p.28; Bowlby has ‘adaptedness’ in the original from which Schore quotes)

Schore also appears to use the term ‘environment of adaptiveness’ to mean something different to the ‘environment of adaptedness’;

“… Bowlby concludes that the mother–infant attachment relation is ‘accompanied by the strongest of feelings and emotions, happy or the reverse’ (p.242), that the infant’s ‘capacity to cope with stress’ is correlated with certain maternal behaviors (p.344), and that the instinctive behavior that emerges from the co-constructed environment of evolutionary adaptiveness has consequences that are ‘vital to the survival of the species’(p.137).” (Schore pp.28-29)

For Bowlby, the ‘environment of adaptedness’ was the evolutionary environment that interacted with genetic endowment to produce the instinctive behaviours typical of human beings. For healthy development to take place, a child needs to be raised within the limits of that environment. For Schore the ‘environment of adaptiveness’ is an environment co-constructed with the mother. In relation to attachment, the two would be very similar, but Bowlby’s environment is much broader than Schore’s. It includes, but isn’t limited to, the interaction with the mother. Schore not only narrows Bowlby’s concept considerably, he actually changes Bowlby’s terminology without – in this paper at least – explaining why.

Placing mother-child interactions centre-stage

I have three main problems with models like Freud’s, Bowlby’s and Schore’s, that make parental behaviour central to the development of child behaviour. The first is that making any factor central by definition marginalizes other factors. There’s a difference between saying “My theoretical model recognizes that there are n factors involved in the development of child behaviour and I’m going to focus on factor a” and saying “There are n factors involved in the development of child behaviour but only factor a is important, so it’s the only one I’ll include in my model”. Some factors are undoubtedly more important than others, and parental behaviour might be one of them, but that doesn’t mean we can assume it’s central to child development. Bowlby bases his assumption that it is, on the correlation found by psychoanalysts between pathologies of personality and childhood trauma. The problem with that correlation is that although childhood trauma might be the cause of a personality pathology it doesn’t mean it’s necessarily the cause or that it’s the only cause. There might be people with pathologies who didn’t have a traumatic childhood, or some who have had a traumatic childhood but haven’t developed pathologies. Psychoanalysts would be unlikely to see many patients in the latter group.

My second misgiving is that a link between childhood trauma and pathology is intuitively appealing. Most of us can recall childhood events involving our parents that we still find painful many years later and examples of other parents dealing with their children inappropriately readily spring to mind. But we don’t often hear people say “I reckon it’s due to a 7R variant of DRD4” or “I suspect serotonin levels, myself” or even, surprisingly since it affects up to 20% of some populations, “Have you considered lactose intolerance?” partly because all these possible causes of abnormal behaviour require more specialized rather than general knowledge and also because they are not obvious – they need to be tested for.

The third reason is related to the second – and it’s that the link between childhood trauma and pathology is superficially simple. Most people will know of examples that appear to confirm it. Fewer people are likely to be familiar with the complex physiological pathways triggered by genes that can affect brain chemistry, or the even more complex interactions between a constantly changing brain chemistry and a constantly changing environment, so fewer people are likely to opt for those explanations. However, when you look at Freud’s, Bowlby’s or Schore’s attempts to map out a possible biological mechanism for child-parent interactions, you realize that the apparent simplicity of the idea is deceptive.

Brains and behaviour

I haven’t yet discussed the second part of Schore’s paper, about the development of the orbitofrontal cortex and its links to the limbic system of the brain. That’s partly because I don’t have sufficient knowledge about brain development to question the accuracy of Schore’s account. It’s also because, although an overview of the development of the orbitofrontal cortex might be informative, I can’t see how it adds support to Bowlby’s theory. Schore asks;

“So the next question is, 30 years after the appearance of this volume [Attachment], at the end of the ‘decade of the brain’, how do Bowlby’s original chartings of the attachment domain hold up?” (Schore, p.29).

Schore’s answer is “In a word, they were indeed prescient.” My response would be a bit wordier; that since Bowlby’s theory (despite the fact that I think it’s flawed in some respects) was based on careful observations and thoroughly grounded in behavioural theories that Bowlby had worked through from first principles, it doesn’t actually need any evidence from brain research for it to ‘hold up’. The reason for this is that if a behaviour is occurring, we know that the part of the brain involved in that behaviour must be working, otherwise the behaviour wouldn’t happen. So if a child recognizes faces, it’s a pretty safe bet that the parts of the brain that deal with face-recognition are up and running. If a child can speak and understand language, we can be pretty sure that the language areas are functioning. If a baby cries when hungry and laughs when playing peek-a-boo, the control mechanism involved in the regulation of feelings is probably on track. It’s when there are deficits in face-recognition or language or sad/happy responses that we’d start to wonder about the pathways from the eyes to the fusiform facial area, from the ears to the fronto-temporal areas of the brain or in the limbic-orbitofrontal pathway.

The problem for Freud’s, Bowlby’s and Schore’s theories is that there are many factors in addition to maternal behaviour that can interfere with the typical development of the orbitofrontal cortex. A variation in a tiny section of genetic material can result in, say, the absence of an enzyme that regulates the way neurons connect to each other, or that changes levels of neurotransmitters. A genetic variation might result in a food intolerance in mother and/or baby, affecting the balance of nutrients available to the infant. These factors could affect the baby’s behaviour, impacting on the mother’s response, in turn influencing the baby’s behaviour, in a complex chain of interactions culminating in a unique pattern of attachment behaviours and in a unique brain structure in the infant.

In other words, if a child is behaving unusually, it’s highly likely that a brain scan would show ‘differences’ in brain structure and/or function. But that doesn’t tell you anything about the origins of the differences; they could be genetic, or environmental, or both. In order to draw useful conclusions about those origins, you’d need a DNA profile for mother and infant, a detailed history, some careful observations of the way they interact, and probably a long wait before the relevant gene-behaviour pathways were mapped out by researchers. Unfortunately, DNA profiling, brain scans, detailed histories and careful observation are expensive and require specialist expertise; it’s simpler and cheaper to attribute all behavioural anomalies to poor parenting.

The implications of Schore’s model

So if, as I claim, Bowlby’s model of child development is questionable and Schore’ model lacks Bowlby’s rigour, why take the time and trouble to critique what Schore has to say? The reason is that Schore’s work has had a significant impact. It’s frequently cited in policy documents and his emphasis on emotion is echoed in the child protection literature – Davies and Ward’s Safeguarding children across services: Messages from research mentions ‘emotional’ 261 times in 226 pages. Damasio was right to point out that research into emotion has been a neglected field of research, but as Bowlby explains, one of the reasons for that is because ’emotion’ is a construct that’s challenging to define. It would certainly be useful to find the mechanisms underlying specific emotions such as anger or sorrow, but the usefulness of ‘emotion’ as a general concept – despite its current popularity – is debatable.

I mentioned brain scans and behaviour; in my next post I’ll look at some of Bruce Perry’s work.