top of page
Phil Harris

Special Feature: Attachment Theory- A Review of the Evidence

Updated: Aug 4, 2020


Infographic courtesy of Aaron Weintraub

In training and lectures on child and adolescent development I explore and discuss the lack evidence to support a central tenant of attachment theory: that relationships with a primary carer shape all the subsequent relationships in the infants future life. Within this I am not saying babies do not need to be cared for. Rather, I question the assumption that all the relational and mental health difficulties that occur in an individual's life stem from their relationship with a faulty primary carer giver. This always causes some considerable and fascinating debates. So I thought I would present the evidence and the limitations of this element of attachment theory and offer people a chance to explore the wider research for themselves.

The first paper you can read is a review of the attachment theories presented by Bowlby and Ainsworth and is good introduction to their work and ideasb(click here). It examines their interpretation of attachment theory and the initial research studies that supported these ideas, largely that infants form an internal model of the world based on the attachment style to their primary care giver, usually their mother. This "internal model" remains with them and shapes all their subsequent relationships. These models are classified as a positive secure attachment, or the more negative avoidant, ambivalent or disorganized attachments. Poor attachments will lead to substantial relational difficulties for the rest of the individual's life.

However, this core assumption that an "internal model" exists within children was never substantiated in either Bowlby or Ainsworth's work. They simply assumed it was there without providing any proof of its existence then or now.

Attachment advocates have tried to address this very significant gap in attachment theory by calling upon neuroscience. Many of its most vociferous advocates will state that the latest neuro-scientific research on brain plasticity has revealed the formation of this internal model. They draw upon brain studies which show that the first three years of a child's life are vital for its successful development. It is during this "sensitive period" that the brain undergoes a surge in synaptic connectivity as brain cells reach out each other and form neural webs and pathways that will "wire" for life. Calling on research that demonstrates the sophistication of rat brains raised in high stimulation compared to those in low stimulated environments, they demonstrate how vital this period is to human brain development. This requires sensitive responsive care in stimulating environments to action mass synaptic connection. Maternal deprivation during this period is therefore a missed opportunity as the brain's developmental door shuts after this time. This then crystallizes patterns of relating in the child that are difficult to amend after this moment.

The first problem with this is that neuroscience research they rely upon is not in fact new but over 30 years old (click here). The second problem is that the neuroscience simply does not support this position (for a review click here). For example synaptic connectivity is simply not an indicator of learning or development. Too much synaptic connectivity actually impedes learning in infants. The result is that many neuro-scientists have disavowed the use of their research in this way, claiming it is an abuse of scientific research for largely political purposes (click here).

The use of neuroscience in attachment theory is often used to lend authority to a position even though the research that is called upon does not support it. For example, Dr William Seers has stated that crying, even for a short time in babies, results in brain damage, lowering IQ and creating behavior difficulties. This is because the brain releases a flood of hormones such as Adrenalin and Cortisol that stop synaptic connectivity. Of the four studies he cites, two are on baby rats, one is on baby primates and only one is on human babies entering into day care. In this study, mouth swabs of babies revealed a 75-100% increase in Cortisol during crying. This sounded shocking and helped to sell Dr Seers books. However, this research paper also showed that the Cortisol levels in babies dropped within 30 minutes and that the increases in Cortisol were not sufficient to cause any harm to the babies at all (for a review click here). So what this research demonstrated was separation from mothers into daycare actually caused these infants no harm at all.

A central challenge for the concept of an internal working model based solely on the relationship with a primary giver is also challenged by cognitive developmental research. This research shows that babies simply do not learn in the way that Bowlby and Ainsworth suggested. Babies simply cannot generalize their learning from one object to another. Teach a baby to jiggle a suspended mobile by shaking its foot that joins it with a red ribbon will bring much joy to the infant. However, change the colour of the ribbon from red to blue and it will be baffled, having to learn the process again from scratch (click here). They cannot transfer red ribbon learning to the blue ribbon. Likewise, their relationship with mother has no bearing on their relationships with others as the cannot generalise this information. This is also reflected in the fact that infants are quite capable of holding different attachment styles with different people simultaneously. An infant securely attached to mum can be avoidant with dad. Furthermore, meta-analysis (a large scale study that combines the results from lots of studies) found that infant attachment styles did not even predict attachment style in the adolescent years (click here).

Infants capacity for multiple attachment is completely missed in wider research. Attachment theory calls upon a wide range of animal behavioural studies on parenting that are simply not applicable to human beings. Most famous is Lorenz work with Greylag geese. On hatching these goslings attach to the first stimuli they see as if it is their mother and will dutiful follow a yellow pair of Wellington boots through what Lorenz referred to as inprinting. This process is largely restricted to birds whose birth and brain structures are completely different to humans.

Other research, such as Harlow's studies, have focused on the impact of maternal separation on monkey species. This research has identified considerable resultant dysfunction in baby monkeys who are removed from their mothers as a parallel to human infant experiences. Again there is a considerable risk in generalizing this research to humans. In all other primates, the mother is the sole caretaker of infants in harem mating structures. Abandonment by mother will result in almost certain death of the infant. Primates can only have one infant at a time and childhood is curtailed swiftly by the arrival of a new infant. Humans have a unique parenting style in the primate world of alloparenting. We have evolved a multiple parenting approach to offspring, meaning greater social supports exists within extended human family and tribal units. We can thus have more than one infant at a time and have the longest childhood in the animal kingdom (for a general review click here, for an anthropological view click here and for an interesting study click here). For human infants, maternal separation in an environment of multiple care givers is very different to other primate experiences. It also suggests that areas of the brain relating to parenting, like the septal nuclei region, is an open system in humans. The release of bonding hormones like oxytocin, which acts upon this region, is not limited to bonding with our own offspring. Other children, individuals, lovers and even pets can trigger activity in this region in humans.

Another central problem in Bowlby's and Ainsworth's work is the complete neglect of genetics in their studies and in follow up research. Babies are not born as blank slates waiting to be written on by their parenting experience. They are born with inherited temperaments and these make a significant contribution to the personality of the child which influences their attachments. Likewise, parents also have temperaments which may influence their parenting style. (For a deep a review click here and here.) Research has shown that identical twins share 50 per cent similarity in temperament whilst non-identical twins share on only about 25 per cent agreement on temperament. A less well know area of study is that of "virtual twins." These are non-related infants who are adopted into the same family at the same time. As they do not share genes, similarities in these children would be solely determined by shared family environment (attachment, parenting etc). Research has demonstrated similarities in these children is zero (click here for a general review or here for a deep review). Virtual twins temperaments are no more alike than two random strangers picked out of a crowd.

Genetic aspects of temperament in infants stay remarkably stable across time (click here, here and here). So low anxiety parents may simply sire low anxiety children whilst anxious parents may sire anxious children. As such, relationships are not simply modeled by internal models but are also mediated by the consistency of temperament. In reality, attachment and genetics are intertwined and emerging evidence in epigentics is revealing how gene actions can be switched on and off by environments (for a review click here). The point here is that care giving behaviours are acting upon an infant whose temperament shapes their responsiveness to it.

Temperament raises important issues of causality. A critical lesson for all researchers is that correlation is not causation. Just because two things happen in close proximity does not mean one is causing the other. This is the central problem identified in the use of attachment assessment tools for children, which can be fatally flawed (click here). Establishing causality is very difficult in research terms and simply not appreciated in a great deal of attachment research. For example, are infants insecurely attached because of aloof parenting or are parents aloof because of the babies difficult temperament? It is important to note that causality cannot be established retrospectively in research. Therefore, many attachment assessment tools that explore parents own past attachment style are not proof of causal relationship between their own attachment and that of their child. Rather they establish a relationship between parenting and the "perception" of ones of own childhood.

Maternal deprivation is often confused with material deprivation in attachment research. However, to sustain a high level of psychical, relational and material support for children demands resources. Resources that a rapidly drained through poverty. In this way, poverty has a direct impact on parenting capacity (click here, here, here and here). Finally it is also worth noting the attachment focuses on the family environment. However, children also operate in non-family environments such as school and peers. Research has also examined the impact of peers and the wider social connections on development. This research demonstrates how children are socialized not by parents but by peer contexts (click here and here).

What is the take home message of this? I am not saying parents are not important, especially when the child's agency is low and the parent control of the environment is high. But we probably see a shifting scale of influence, as agency increases and parental control of environments decrease. So parents do matter, it is just that they are not the only thing that matters. When assessing children and young people it is essential we take a critical view over the full expanse of the child developmental experiences, not simply one element. Otherwise we are not assessing them as an individual in their own right but confirming what we already know.


See also "Where is the internal working model?" click here.

Personality Part I: The research base. click here,

I must give a big thanks to Aaron Weintraub for the use of his excellent infographic.


9,097 views0 comments

Recent Posts

See All
bottom of page