Secure Attachment in Infants: What the Research Describes, and What It Does Not

Attachment in infancy is not a measure of how much a baby loves a parent, and it is not a score for how well the parenting is going. It describes something narrower and more specific: what a baby does with a caregiver when they are frightened, tired or hurt, and whether that caregiver works as a base to go back out from.

The purpose of the system, in John Bowlby's original argument, is protection. An infant who stayed near an adult when alarmed survived, so human babies arrive with the behavior built in: crying, clinging, following, searching. None of it is manipulation or a habit the adult accidentally taught, which is what the childcare advice of the mid twentieth century generally assumed.

The part that gets missed is the exploration half. Mary Ainsworth and Silvia Bell's 1970 paper reported that a mother's presence encouraged exploratory behavior in one-year-olds while her absence depressed exploration and heightened attachment behavior. Security is not a baby who stays close. It is a baby who can leave, because leaving is safe when coming back works.

Ainsworth spent years observing families at home, first in Uganda in the mid 1950s and then in a longitudinal study in Baltimore, before building the procedure she is known for. The Strange Situation is a short sequence of episodes in an unfamiliar room with toys, in which a caregiver leaves and returns and an unfamiliar adult is present for part of it. The point of the design is to raise the stakes gently enough that the attachment system switches on and the pair's usual way of handling it becomes visible.

The classification comes from the reunion. How distressed a baby becomes during the separation reflects temperament as much as anything, which is why it is not the scored part. What the raters attend to is the return: whether the child seeks the caregiver, whether they take the comfort offered, and whether the comfort works well enough that they go back to the toys.

Ainsworth and colleagues described three patterns on that basis, usually labeled secure, avoidant and resistant, the last also called ambivalent. A minority of infants did not fit any of them, and in 1990 Mary Main and Judith Solomon published indices for identifying a fourth, disorganized pattern, so that researchers could code it consistently.

Two things about all of this matter for a parent reading it. These are descriptions of a pair in a research procedure, not properties of a child, and they are assigned by trained raters watching video of a standardized sequence. There is no home version.

The table at the end of this page sets the research description beside the everyday version, which is the more useful one for anyone who is not coding video. Read it as background rather than as something to apply.

The everyday column is deliberately loose, because the honest translation is loose. A child who is clingy on a Tuesday and independent on a Thursday is a normal child, and the patterns describe tendencies across many such days rather than the behavior of an afternoon.

In the first year it mostly looks ordinary, which is part of why it is reassuring. A baby settles reasonably well when picked up by a familiar person, not instantly and not always, but the comfort generally does something. They show a clear preference for a few familiar people by the second half of the year. They protest being left, which is a sign the system works rather than a sign of fragility, and the protest ends when the person comes back.

The piece that researchers watch for, and the one parents can recognize without grading anything, is the cycle: distress, contact, settling, back to whatever was interesting. The settling does not need to be fast. What characterizes the pattern is that the contact is used rather than refused or unable to be absorbed.

Separation anxiety around eight to ten months and stranger wariness in the same period are developmental milestones rather than warning signs. A baby who has started objecting to being handed to a stranger has built the discrimination the system needs.

By two or three the base-and-explore pattern is visible across a room. A securely leaning toddler at a playground goes further out than they did a year ago, checks back with a look rather than a return trip, and comes all the way in when something goes wrong. The checking in is the behavior, not the staying close.

The other visible feature is bringing things. Toddlers who are using a caregiver as a base share: a leaf, a discovery, a complaint about another child. Sharing is the same system running in a calm state, and a child who brings you things is telling you the base is working.

Big feelings do not count against any of this. Toddlers in every classification have enormous tantrums, refuse to leave the park, and fall apart at bedtime. What differentiates the patterns is not the size of the storm but whether the child comes to you during it and whether being with you helps. And that too varies by day, by illness, by how much sleep everyone had.

The most consistent finding is about sensitive responsiveness: noticing a signal, reading it roughly right, and responding in a way that fits, often enough. Marianne De Wolff and Marinus van IJzendoorn pooled 66 studies in 1997 and found observed sensitivity associated with security at about .24 in the core observational subset, a moderate relationship, which led them to call sensitivity an important but not exclusive condition of security.

Two features of that finding are worth noticing. Moderate means other things also matter, including the child's own temperament, the family's circumstances, and the stress load on the caregiver. And often enough means often enough. Nobody reads a baby correctly every time, and the research describes nothing that requires it.

The second thing the evidence supports is repair. Ordinary interaction is full of small misses: a signal missed while answering the door, a response that was wrong, an adult who got short-tempered at the end of a long day. What the research associates with security is not an absence of those moments but the pattern of coming back and putting it right.

There is also encouraging evidence that this responds to teaching rather than being fixed. Across 70 studies, Marian Bakermans-Kranenburg and colleagues found that randomized programs moved insensitive parenting by about 0.33 and infant insecurity by about 0.20, and that the most effective ones used a moderate number of sessions and a clear-cut behavioral focus rather than an open-ended format.

A great deal of what parents worry about does not carry the weight the internet assigns it. Child care is the clearest case. The NICHD Study of Early Child Care followed 1,153 infants and their mothers across ten sites and reported no significant main effects of child care experience, including quality, amount, age of entry, stability and type, on attachment security. Elevated insecurity appeared only where low maternal sensitivity combined with poor-quality care, more than minimal amounts of care, or multiple arrangements. Day care by itself was not the variable.

Working parents, bottle feeding, a baby who sleeps in a cot rather than a bed, a toddler who had a tough two weeks while you were ill: none of these is the determining factor the anxious version of attachment writing implies. Neither is a single bad day, a lost temper, or a week where you were too flattened to be delightful company.

What does carry weight is the accumulated pattern over many months, and the single best use of a page like this is not to audit yourself against it. It is to notice that the thing being described is repeated ordinary responsiveness, which is both less glamorous and much more achievable than the version where every interaction counts.

The Strange Situation was built in the United States with a particular set of assumptions baked into it, including that being briefly left with an unfamiliar adult in an unfamiliar room is a mild, ordinary stressor. Where that assumption does not hold, because infants are rarely separated from a caregiver at all or are routinely cared for by several adults, the same procedure is not putting the same question to the child.

The proportions landing in each classification accordingly differ between samples and between countries, and the review literature treats this carefully. Judi Mesman, Marinus van IJzendoorn and Abraham Sagi-Schwartz, surveying the cross-cultural studies that followed Ainsworth's work in Uganda, describe the overall picture as a balance between universal trends and contextual determinants rather than one attachment pattern per culture. Variation between families within a country is substantial in every sample anyone has collected.

For a parent, the practical reading is modest in two directions. The core idea, that a protective base makes exploration possible, is observed broadly rather than being a local Western finding. And the percentages in any single classification table deserve less weight than they usually get, particularly when they are being used to say something about a culture's childrearing.

This is deliberate. Attachment classifications are assigned by trained raters from video of a standardized procedure, with careful attention to behavior that is easy to misread, and no published checklist converts a parent's observation of their own child into a classification. An attachment pattern is also not a diagnosis, it is not in any diagnostic manual as a condition, and it describes a relationship rather than a trait a child carries.

There is a practical reason to stay out of it as well. A parent watching their own child for signs is no longer doing the thing the research actually links to security, which is responding to the child in front of them. Observation of this kind tends to make interaction stiffer, and the stiffness is the opposite of what helps.

If you have read this far looking for reassurance, the honest version is that there is no test here that would give it to you, and that the behavior associated with security is the ordinary kind: being approachable when they are upset, coming back after the moments that went badly, and letting them go further out than feels comfortable.

Take it to a person rather than to a page. A pediatrician is the right first stop, and they can refer to a clinician who works with young children if there is something to look at. Bring specifics: what you see, when it started, what happens when they are hurt or frightened, what helps and what does not.

A few situations warrant professional input sooner rather than later, and none of them is about an attachment label: a child who seems to have no preference between a familiar adult and a stranger, a child who does not seek comfort when clearly hurt, a marked change after a particular event, or a caregiver who is depressed, frightened or struggling enough that responding feels out of reach. The last one is not a failure and it is one of the more treatable items on the list.

What is not useful is a label applied from a web page. It cannot be accurate, and it tends to settle into a story about the child that outlasts the worry that produced it.

The free Psychology.com Attachment Style Quiz is for adults, about their own close relationships, and it asks how you generally think and feel in them. It is not a child assessment and cannot be used as one. It is a self-report questionnaire for education and reflection, not a clinical or diagnostic instrument, and an optional in-depth report is available afterward for readers who want a longer write-up.

It is still relevant here, for one reason. Research on the transmission of attachment across generations finds modest links between a caregiver's own way of handling closeness and their child's security, and it finds that caregiving sensitivity does not fully account for those links. Your own pattern is the part of this you can actually look at, and attachment theory, the adult attachment styles index and the insecure attachment overview are the places to do it.