Lorenz and Harlow

Lorenz (1935) — imprinting. He split a clutch of goose eggs: half hatched with the mother, half hatched in an incubator where the first moving thing they saw was Lorenz. The incubator group followed Lorenz everywhere; the control group followed the mother. When the two groups were mixed, they separated and went to their own 'mother'. Imprinting occurs only within a critical period (for these birds, a few hours after hatching); if it does not happen then, the birds did not attach at all. Lorenz also described sexual imprinting — a peacock reared among giant tortoises later directed courtship at tortoises.

Harlow (1958) — contact comfort. 16 baby rhesus monkeys were reared with two wire 'mothers'. In one condition milk came from a plain wire mother, in the other from a cloth-covered mother. The monkeys spent most of their time clinging to the cloth mother regardless of which one dispensed milk, and ran to her when frightened. Attachment is not about food but about comfort. Monkeys reared with only the wire mother were the most dysfunctional; those maternally deprived grew up socially abnormal — aggressive, unable to mate, and, as mothers, neglectful or violent towards their own young. Harlow concluded there is a critical period of about 90 days for attachment, after which damage was irreversible.

Evaluation: both studies had huge theoretical impact — Harlow's work directly informed Bowlby and changed care practices in hospitals and children's homes. But generalising from birds to humans is questionable (mammalian attachment is far more complex and two-way), and Guiton et al. found imprinted chickens could later learn normal mating behaviour, challenging the idea that imprinting is permanent. Harlow's study raises severe ethical issues: the monkeys suffered lasting harm, and it is debatable whether the knowledge gained justifies it.

What did Harlow show?

Learning theory — 'cupboard love'

Dollard and Miller (1950) explained attachment through conditioning:

• Classical conditioning — food (UCS) produces pleasure (UCR). The caregiver starts as a neutral stimulus but, through repeated association with food, becomes a conditioned stimulus producing a conditioned response of pleasure. That association is the attachment.
• Operant conditioning — crying leads to a response (feeding), so crying is positively reinforced. For the caregiver, the crying stops when the baby is fed, which is negative reinforcement. This is a two-way system of mutual reinforcement.
• Drive reduction — hunger is a primary drive; food is a primary reinforcer. Attachment is a secondary drive, learned by association with the reduction of the primary drive.

Evaluation: the theory is contradicted by animal research — Lorenz's geese imprinted regardless of feeding, and Harlow's monkeys chose comfort over the mother that fed them. It is also contradicted by human research: Schaffer and Emerson found many babies formed their main attachment to the mother even when other people did most of the feeding. Attachment involves far more than food — reciprocity and interactional synchrony predict attachment quality, and none of that is explained by conditioning. That said, learning almost certainly plays some role: babies may learn to associate a caregiver with comfort and warmth rather than food.

Bowlby's monotropic theory

Bowlby rejected 'cupboard love' and proposed an evolutionary explanation — attachment is innate because it improves survival. Remember it as ASCMI:

• A — Adaptive. Attachment gives an survival advantage: babies who stay close to a caregiver are protected, fed and kept warm.
• S — Social releasers. Innate 'cute' behaviours (smiling, cooing, the baby face) activate the adult attachment system. Attachment is therefore reciprocal.
• C — Critical period. Bowlby saw around two years as the critical window (later softened to a sensitive period, with risk continuing up to about five). If no attachment forms, the child will be damaged for life.
• M — Monotropy. A child's attachment to one particular caregiver is unique and more important than any other. The law of continuity (constant, predictable care) and the law of accumulated separation ("the safest dose of separation is a zero dose") follow from it.
• I — Internal working model. The first relationship becomes a mental template for all later relationships, and for the child's own parenting.

Evaluation: support comes from Bailey et al. (2007), who assessed 99 mothers and found those with poor attachment to their own parents were more likely to have poorly-attached one-year-olds — evidence for the internal working model passing down the generations. But monotropy is challenged: Schaffer and Emerson found a significant minority formed multiple attachments at the same time, and it is unclear that one attachment is qualitatively different rather than simply stronger. The idea is also socially sensitive — feminists such as Erica Burman argue that the law of accumulated separation places an unfair burden of blame on working mothers.

Which part of Bowlby's theory?

The Strange Situation (Ainsworth, 1970)

A controlled observation in a laboratory playroom with a two-way mirror, designed to measure attachment quality. It has eight episodes of three minutes each and assesses five behaviours: proximity-seeking, exploration and secure-base behaviour, stranger anxiety, separation anxiety and response to reunion.

• Type B — Secure (about 66%). Explores happily but regularly returns to the caregiver (secure base). Moderate separation and stranger anxiety. Accepts comfort at reunion. Associated with sensitive responsiveness from the caregiver.
• Type A — Insecure-avoidant (about 22%). Explores freely but does not seek proximity or use the caregiver as a secure base. Little separation or stranger anxiety, and avoids contact at reunion. Associated with a caregiver who rejects the infant's needs.
• Type C — Insecure-resistant (about 12%). Clingy, explores little, huge separation and stranger anxiety, and resists comfort at reunion (seeks contact and pushes away). Associated with inconsistent care.

Evaluation: the Strange Situation shows excellent inter-observer reliability (Bick et al. found agreement on attachment type in 94% of cases) and good predictive validity — secure infants tend to have better later peer relationships and mental health. But it may be culture-bound (an imposed etic): it was designed in the USA and Britain and may not measure the same thing elsewhere, and Main and Solomon argued Ainsworth missed a fourth type, disorganised attachment (Type D) — inconsistent, contradictory behaviour, usually following severe abuse or neglect.

Cultural variations in attachment

Van IJzendoorn and Kroonenberg (1988) meta-analysed 32 Strange Situation studies from 8 countries, comprising 1,990 children.

• Secure attachment was the most common type in every country — from 75% in Britain to 50% in China.
• Germany had unusually high insecure-avoidant rates (about 35%) — consistent with a culture that values independence and non-clinginess.
• Japan had unusually high insecure-resistant rates (about 27%) and almost no avoidant attachment — Japanese infants are rarely separated from their mothers, so the separation episodes are unusually distressing.
• Variation within countries was about 1.5 times greater than variation between them — sub-culture and social class matter more than nationality.

Simonelli et al. (2014) found lower rates of secure attachment in Italy than earlier studies, which they attributed to increasing numbers of mothers working long hours — showing cultural patterns change over time.

Evaluation: the meta-analysis has a very large combined sample, reducing the impact of anomalous results. But 27 of the 32 studies came from Western countries, and 18 from the USA alone, so the 'cross-cultural' claim is overstated. Most seriously, comparing countries confuses country with culture, and using an American-designed tool abroad risks an imposed etic — 'avoidant' behaviour in Germany may be a sign of independence rather than insecurity.

Read the data

Maternal deprivation and the 44 thieves

Bowlby (1951) distinguished deprivation (an attachment is formed and then lost) from privation (an attachment is never formed). He argued that continuous care from a mother-figure is essential, and that prolonged separation in the critical period (the first 2½ years) causes irreversible harm:

• Intellectual development — abnormally low IQ (he cited Goldfarb, 1947, who found lower IQ in institutionalised children than in fostered children).
• Emotional development — affectionless psychopathy: an inability to feel guilt or strong emotion for others, which prevents normal relationships and is associated with criminality.

The 44 thieves study (1944). Bowlby interviewed 44 teenage thieves and their families and compared them with a control group of 44 non-criminal but emotionally disturbed young people.

• 14 of the 44 thieves could be described as affectionless psychopaths.
• 12 of those 14 had experienced prolonged separation from their mothers in the first two years.
• In the control group, only 2 of 44 had experienced such separations.

Evaluation: the evidence is weak. Bowlby carried out the interviews and the psychopathy assessments himself, knowing his own hypothesis — a serious risk of investigator bias. The design is correlational, so separation may not be the cause. Lewis (1954) replicated the study with a larger sample of 500 and found prolonged separation did not predict criminality or difficulty forming relationships. Bowlby also confused deprivation with privation: as Rutter argued, the severe damage he described follows from a failure to form any attachment, not from losing one.

Romanian orphan studies

Under Ceausescu, Romanian orphanages held huge numbers of children in extreme neglect. Their adoption in the 1990s created a natural experiment.

Rutter et al. (2011) — the English and Romanian Adoptee (ERA) study. 165 Romanian orphans adopted into British families were assessed at 4, 6, 11 and 15, and compared with 52 British adoptees.

• On arrival, half the Romanian children showed signs of delayed intellectual development, and most were severely undernourished.
• Mean IQ at age 11 depended on the age at adoption: adopted before 6 months — 102; adopted between 6 months and 2 years — 86; adopted after 2 years — 77.
• Those adopted after 6 months showed disinhibited attachment: attention-seeking, clinginess and indiscriminate affection towards any adult, including strangers.

Zeanah et al. (2005) — Bucharest Early Intervention project. The Strange Situation was used with 95 Romanian children aged 12–31 months who had spent most of their lives in institutional care, and a control group of 50. 74% of the control group were securely attached, compared with only 19% of the institutionalised group; 44% of the institutionalised group showed disinhibited attachment, compared with under 20% of controls.

Evaluation: the research has enormous practical value — it has reshaped adoption policy (children are now adopted within the first week where possible) and the running of children's homes, which now assign key workers so that children can form attachments. Methodologically, the children were not randomly allocated to age-of-adoption groups: the most sociable babies may have been adopted first, which is a confounding variable. And these orphanages were exceptionally poor — a case of extreme privation, so the findings may not generalise to institutional care in general.

Read the ERA data

The influence of early attachment on later relationships

Bowlby's internal working model predicts that early attachment quality carries into childhood friendships, adult romantic relationships and parenting.

• Childhood. Kerns (1994) found securely attached babies tend to go on to form the best-quality childhood friendships, while insecurely attached children have friendship difficulties. Myron-Wilson and Smith (1998) surveyed 196 London children aged 7–11 and found secure children were unlikely to be involved in bullying, insecure-avoidant children were the most likely to be victims, and insecure-resistant children were the most likely to be bullies.
• Adult romantic relationships. Hazan and Shaver (1987) printed a 'love quiz' in an American local newspaper, analysing 620 replies about current attachment experiences, attachment history and attitudes to love. 56% were classified as secure, 25% as insecure-avoidant and 19% as insecure-resistant. Secure respondents were the most likely to have happy, lasting relationships; avoidant respondents were the most likely to fear intimacy.
• Parenting. Bailey et al. (2007) found the majority of the 99 mothers studied had the same attachment classification to their own mothers as their babies had to them.

Evaluation: most of the evidence is correlational and relies on retrospective self-report (asking adults to recall infancy), which is open to inaccuracy and social desirability bias. A confounding explanation is temperament — Kagan argued an innately calm, sociable temperament produces both secure attachment and good later relationships, so attachment may not be the cause. Finally, the deterministic implication is overstated: many people with poor early attachments go on to form perfectly good relationships, so the internal working model is a probabilistic, not fixed, influence.

Spot the confound

The critical period

Cupboard love

Attachment type from behaviour

Tap a behaviour from the Strange Situation, then the attachment type it indicates. Ainsworth classified on reunion behaviour above all.

Study and finding

Tap an item on the left, then its partner on the right.

The big ideas to know

You have covered the whole of AQA 4.1.3 Attachment. Press Finish to see your score.