I have recently had a series of debates with psychologists, social workers and family workers around the country regarding Attachment Theory. They are surprised when I say that the evidence for an Internal Working Model developing in infancy is almost non-existent. When I share the evidence that does exist, people are somewhat shocked. So, whilst I know that the defense of Attachment Theory is the hill that some practitioners would chose to die on, I thought I would share the study that does purport to have identified the Internal Working Model.
As a brief recap, for Bowlby the Internal Working Model was the mental representation that develops in infancy based on the child's experience of their primary care giver. Whilst he accepted infants can have multiple attachments, he postulated there is a deep biological mechanism in infants that bonds them to the mother in particular. Exposure to mothering behaviors creates deep expectations regarding the nature of relationships, including the capacity to have needs met, and this then becomes a blue print for all subsequent relationships in the individual's life. This Internal Working Model was the driver of attachment. Inconsistent or emotionally unavailable parenting leads to children to experience distortions in their Internal Working Model, leading them to have a dysfunctional relating style for the rest of lives. This can have dire long term consequences. As Bowlby himself observed:
"The key point of my thesis is that there is a strong causal relationship between an individual's experiences with his parents and his later capacity to make affectional bonds, and that certain common variations in that capacity, manifesting themselves in marital problems and trouble with children as well as in neurotic symptoms and personality disorders, can be attributed to certain common variations in the ways that parents perform their roles." (Click here)
Here Bowlby is not claiming that parenting style is a contributory factor as some people assume and wish to correct me on. He is claiming that poor Attachment is the cause of all major mental illness and relational dysfunction. This is not a pick and mix model. It is an extraordinarily big claim and big claims require big evidence. But Bowlby himself had no evidence that such an Internal Model existed. Instead he looked to the research by Konrad Lorenz with Greylag geese. Hatchling Greylag geese will imprint the first that they see as a mother figure, yellow wellingtons for example, and dutifully follow them around. But this is a species specific behavior. We share little in common biologically with the egg laying, migratory goose. Mother Nature provides a wide range of parenting strategies and so clear linkage would have to be established before drawing conclusions between geese and humans. Even in the 1950s, Bowlby was accused of simply cherry picking one animal model that suited his own beliefs without the provision of any evidence it related to humans. For example, 15-20 per cent of Greylag geese are homosexual, have no offspring but serve as a protectors of other Greylag geese with offspring. Are we to also extract parenting lessons from this too?
Advocates of inprinting also fail to mention that Konrad Lorenz himself had been a very passionate member of the Nazi party, took a lead role on Racial Policy research and wrote extensively of the dangers of mixing races (click here). The Nazi's were not keen on Darwinian Evolution and so were looking for other reasons why they were innately the Master Race. Cultural practices figured highly in this.
The central problem for Attachment theory then, and now, is that there has been no evidence to support the idea that Internal Working Model exists. Whilst hundreds of Attachment studies have been published, these are observational studies based on similar methods used by the likes of Lorenz in animal studies. They confuse what a child does (a description) with an explanation of why they do it (a cause). Alternatively they rely on parents retrospective recollection of their parenting experiences. Again, you cannot prove causality in these studies. I can describe a lot cars in exquisite detail but it does not make me much of a mechanic. A million observation and retrospective reflections will not identify a single cause of why something is observed.
Very few, if any studies, are concerned with identifying the specific mechanism of the child's behavior. The Internal Working Model is just assumed to be there and inevitable conclusions are drawn from this. The vast majority of Attachment studies are based on this pre-supposition creating a circular argument. Why cannot baby be soothed in the Stranger Test? Because of the mothering style. How do we know it is the mothering style? Because the baby cannot be soothed.
What about the orphans, both monkey and humans I hear you cry! They have no Internal Working Models and this is ruinous for them! This research looks very compelling-when it is viewed only though an Attachment Theory lens. It must be born in mind that Harlow's research on abandoned infant Rhesus monkeys clinging to a towel mother is not representative of monkey parenting in the wild. Adoption or peer rearing is not part of the monkey's normative behavior and so abandonment of a baby monkey is a death sentence for the infant. Harlow's experiments did not even provide a space for the babies to hide. They had only a forced choice between two fake monkeys. It is a very strong person-situation event. All fear based neurological systems will be sent into overdrive. It is somewhat akin to making judgments about your relationship with your mother based on how you react to being chased down the street by an escaped tiger. However, Attachment theorist present the monkey separation studies as proof that Attachment is a biological necessity, without which the orphaned monkey develops no Internal Working Model and is doomed to the lowest rung of the troop hierarchy.
However, Rhesus monkeys reared without mothers but with a peer showed little difference in social adjustment compared to their mother-lavished kin (click here). So maybe mum was not so important, but social contact is. Social animals must rehearse and practice social skills. An abandoned monkey not only has no mother but they have no rehearsal at anything. Peers can provide this as equal as mothers, they are just not as experienced themselves to be the learned teacher that an adult is.
Again, in these studies little attention is paid to any other possible causes for the infants behavior. Some infant monkeys are highly anxious even with their mothers and this appears to be a facet of their underlying genetic temperament (click here). More recent studies have also identified that the milk formula given to Rhesus monkeys in separation experiments lacks key fats. This deficiency leads to brain impairments in Rhesus monkeys (click here).
A tenant of faith in the Attachment research is that these once separated monkeys can be rescued from their plight if adopted by highly nutrient mothers. Here the implied assumption is that Internal Working Model can be re-activated in these adoptees, and these nurtured orphans go on to become fully functioning and well-adjusted members of their troops. Stephen Suomi's monkey research is cited heavily in this area and to read a free chapter by him on this subject click here.
Now I am not an expert but I am sure that facilitating monkey adoption is no easy business. Pairing fretful infants with nurturing mothers must be very difficult. The trouble with the oft cited Suomi studies is they have never been published in a peer-reviewed academic journals. As Judith Riche Harris (click here for a review) points out, these research findings are only referred to in book chapters. (Look at the end of Suomi's chapter for his references and you will see what I mean-they are books not primary research studies). In other words there is no systematic presentation of what these adoption studies were, how many monkeys were involved, what procedures were utilized or any statistical analysis of the outcomes. In other words, there is no scientific scrutiny necessary to deem them as validated. This does not qualify as evidence in any scientific sense but is a trust exercise and sometimes we are too trusting. For, example, one adoptee study often cited is not even an adoptee study at all but a peer raised monkey study.
This does not prevent people citing Suomi's research as hard fact though. For example, just check out this triumphant piece entitled The Verdict is In: The Case for Attachment Theory (click here). Here, among many spurious claims, they state:
"One example is the work of biological psychologist Stephen Suomi, who's been working with monkeys in highly controlled experiments. These studies have shown that two genetic variations that have been associated with alcohol abuse or impulsiveness in humans are linked to totally different outcomes when the infant monkeys are reared by a group of highly nurturing foster mothers rather than by their birth mothers."
The assertion here is that Nurture trumps Nature. Attachment is far more important that the genetic predisposition and this proves it. Now if you return to the article and scroll down to the references in this article, you will see that the Suomi evidence of Attachment is... a book chapter. So what exactly was the highly controlled experiment? Where are the studies? What were the totally different outcomes? And why have they not been published as primary research if the evidence is so compelling?
I am one of those annoying people that checks out references in research. And there is a pattern in the Attachment research of playing very fast and loose with the studies they cite. For example, as I reviewed earlier (click here), the sensitive period of development in the brain and the damaging levels of cortisol in the brain if babies cry are not supported by the research base that these theorist cite. However, over time these claims have mutated into given facts and then become established truths. This is always a danger when University lecturers and students rely on secondary research sources, not the primary research.
Similar issues dominate the use of human orphans studies, who are raised in highly impoverished environments. This terrible experience is not restricted just to the Romanian orphanages sadly. However, very few orphan studies include the type and depth of deprivation that is experienced by these children. For example, whilst some institutions are characterized by global neglect, others are characterized by meeting the children's physical but not emotional and intellectual needs. Ward sizes, whether there is any consistency in the care giver, premature births, consanguineous births, birth weight, disease, whether children are bracketed by age in these institutions or not, all make a significant difference in the environments that these children experience. Unraveling these issues is very difficult and little attempt is made to do so. Even very basic factors such as nutrition and undernourishment are not considered, when this has a dramatic effect on the physiological development of the infant brain. The sole factor that is isolated for study is just the lack of Attachment.
Orphans studies also tend to focus on the differences in orphans who experience developmental deprivation such as low IQ, indiscriminate attachment styles and pseudo-autism in higher numbers than the general population. Whilst these differences are real, they are by no means the universal outcomes for these children. Rather they show elevated levels compared to the general population and some only slightly so. Nor do these studies pay much attention to the rapid adjustments that most of these children make post release from institutional care-an adjustment that might be considered impossible according to Attachment Theory. For a thorough and deep review of the research on these wider issues of orphan studies and institutional neglect click here. This wider view does negate the claim that having no Internal Working Model leads to complete disorganization of an individual's capacity to achieve social adjustment.
Certainly, if the Internal Working Model is a blue print for all subsequent relationships, we could at least surmise its existence by identifying that people relate to others in a consistent manner, as predicted by their early Attachment style. So whilst we may not demonstrate the Internal Working Model itself- we might identify its expression. But alas, this is not so. Even advocates of Attachment Theory have found it hard to find a repetitive pattern of relating as suggested by Bowlby that exists in real life. As Pietromonaco & Barrett noted in their extensive review of the research of Internal Working Models:
"Little direct evidence exists that is relevant to the organization of working models. Several studies, however, bear indirectly on this issue and are consistent with the ideas that people hold both general and relationship-specific working models and that different working models might exist at the specific level. Generalized and relationship-specific measures of working models of the self and others are positively associated (e.g., more positive general models of the self are associated with more positive relationship-specific models of the self), but the correlations are small to moderate (less than .40), indicating that they are not identical." (Click here).
So, we hold multiple attachments with multiple people which are sometimes repeated patterns of early our Attachment, in the case of healthy relationships, to some degree. However, disorganised, ambivalent or insecure attachments, the ones we really want to worry about, do not appear at all. This is a far cry from Bowlby's very specific assertion that:
"To explain why individuals of different sorts should continue to exhibit the characteristics described long after they have grown up, it seems necessary to postulate that, whatever representational models of attachment figures and of self an individual builds during his childhood and adolescence, these tend to persist relatively unchanged into and throughout adult life. As a result he tends to assimilate any new person with whom he may form a bond, such as spouse or child, or employer or therapist, to an existing model (either of one or other parent or of self), and often to continue to do so despite repeated evidence that the model is inappropriate." (Click here)
So, it is with great fanfare I will share a research study that did purport to have identified Internal Working Model. Johnson, Dweck and Chen believed that they have identified clear evidence for its existence. As they rightly note, when it comes to the existence of an Internal Working Model:
"Yet no study has ever directly assessed internal working models of attachment in infancy. In the present study, we sought to do so." (To read the study click here)
This team identified 21 infants attachment style in a Stranger Situation test. 10 of whom were classified as securely attached and 11 who were deemed insecurely attached. These infants were then shown cartoons. In short, a big blob (representing mother) and small blob (representing baby) stand together at the bottom of an incline. Big blob ascends the slope and stands on a small plateau. Baby blob bounces up and down to the sound of a baby crying. Mother blob either descends or ascends the incline, toward the baby blob or away. The finding was that securely attached infants watched this screen of the responsive mother blob for 10.2 seconds, compared to the insecure infants who watched unresponsive blob for 8.0 seconds. This was because it defied their internal expectations of mothering.
And that is it. The evidence of Internal Working Model.
This study shares the hallmarks of many problems in the attachment research studies. Firstly, although the diagnosis of the infants attachment style was done blind coded by a third party, this party was assessing infants in a Stranger Test which already presumes that the reactions of infants are because of parenting style. No other measures are examined, like the temperament of the child for example. The evidence for the heritability of temperament is extraordinary high (click here) but not high enough to be included in these studies .
Furthermore, outcomes derived from small sample sizes tend to evaporate very quickly when scaled up include larger sample populations. Statistically, small numbers operate very differently from big numbers. Very often promising small sample results begin to recede into general averages as sample size scales up. If I beat you in a game of heads & tales 3 out of 4 times, this is no great shakes. If I beat you 3 million out of 4 million times, you will want to inspect my coin a lot more closely.
Babies vary in their curiosity about lots of things but tend to be drawn to anything which is unusual. For example, Johnson, Dweck and Chen inspiration came from a study conducted by Gergely et al that showed 12 month year old babies were curious about the movement of shapes that appeared to defy their expectations (click here). Their conclusions were more modest, suggesting babies responses were an expression of emerging rationality, and specifically in regard to shapes and space. Babies do a similar thing when confronted with magic tricks that defy their inherent sense of physics. (To see a wonderful short film of this click here). However, we cannot conclude from this that they have a special relationship with a magician. A description is not a cause.
I think that there is a tension in the Attachment model. It must be remembered that Bowlby saw himself primarily as a psychoanalyst in the Freudian tradition. Psychotherapy is primarily concerned with the exploration of the subjective sense of self. Here people are free to develop their self-narratives which helps organise their understanding of themselves in ways which hold personal meaning to them. I do not have a problem with this idea. As adults we are free to make sense of our lives in anyway we chose and if this is life-enhancing then this is a invariably good thing, regardless of the evidence. But it does not make it true-it makes it useful.
However it does become more problematic when professionals are making judgments about the care of children based on speculative theories with little empirical evidence to support them. "Child X has Attachment issues" is any easy thing to write on a social welfare report. (To see how easily this approach can be dismantled by a questioning Judge, click here.) However infants, children and adolescents experience problems in their life that are multi-causal and may be nothing to do with their mothers. For example, there was a time when both autism and schizophrenia was squarely blamed on refrigerator mothers. These are cold, hard distant mothers who imprison their children in a concentration camp of themselves (click here). We may bulk at this idea now but Attachment Theory is in the same slip stream of thinking. I still hear Attachment advocates continue make similar claims at Conferences to this very day based on the most trivial anecdotal evidence. I think we owe vulnerable children and young people more consideration than this.
I am also not saying that children who are exposed to wide ranges of abuse cannot be damaged by it. What I would suggest is that not all problems children experience are the result of abuse. Attachment theorist interchange the worlds Attachment and Trauma very easily, especially when challenged on issues in Attachment per see. The question is how do you identify what problems are the result of abusive parenting and what problems are rooted in other factors? Does a proclivity towards Attachment Theory mean that practitioners don't even think beyond Attachment? The worlds leading Pediatricians have identified this concern and so have recommended that Attachment Assessments are no longer used in general practice as they are unable to discriminate between these two issues with any reliability (click here).
So I would question the assumption that all children difficulties are solely located in abusive parenting practices. Equally, I would also question that all abusive parenting practice is located in the parents own childhoods and that all mental health problems are the result of early attachment. Longitudinal studies show that parenting attachment style has very little impact on people's experience of mental health in the long term, even when they utilize harsh and disciplinary approaches (click here). I am certainly not advocating corporal punishment for children here, I am just emphasizing that even this extreme form of parenting demonstrates little long term impact in contrast to the assumptions of Attachment Theory.
For advocates of attachment, I would say it is not bad to advise parents to love their children. It is bad advice to assume that all the problems their children experience are due to a lack of love. Trauma hurts children but not all hurt is anchored in just one source. More love may not be the answer for many children's difficulties when a lack of love is not the driving force. And those not in receipt of love at an early age are not doomed to a loveless life. This is very important as the clinical outcomes for children and young people are actually in decline as the popularity of Attachment theory has increased. And there is not one single psychologically evidence based approach based on Attachment Theory that has demonstrated its effectiveness to a required standard to be considered validated. In fact the opposite is the case. For example, Attachment Therapy was a model based on Attachment Theory that promotes re-parenting techniques that has led to multiple deaths of adopted and orphaned children in particular (click here or here) with demands that its use by made illegal as a result.
This begs a number of important questions. If parenting style is not driven by Attachment, what does shape parenting? What causes mental health problems to evolve? Why is Attachment Theory so popular? And what are the more optimal forms of intervention for children and young people? These are some of the questions I will be addressing in forth coming specials. Stay tuned.