Some people re-read a text three times before sending it, then check if it was delivered, then check if it was read, then start composing a follow-up in their head just in case. Other people feel a small flicker of claustrophobia when a partner says "we should talk about us" — not because anything is wrong, but because the conversation itself feels like a demand. Neither of these is a personality flaw. Both are, in a very literal sense, decades-old science: patterns first mapped in infants, then carried, mostly unmodified, into how adults do closeness.

Where the idea comes from: Bowlby, Ainsworth, and a locked door

The theory starts with the British psychiatrist John Bowlby, who in 1950 was asked by the World Health Organization to look into the mental health of children who had lost their parents in World War II. Watching how those children responded to separation and loss, Bowlby became convinced that the bond between an infant and a caregiver wasn't a byproduct of feeding or comfort — it was its own biological system, built by evolution to keep a helpless infant near someone who would protect it. He laid the idea out across three volumes published between 1969 and 1980, drawing partly on ethology — Konrad Lorenz's work on ducklings imprinting on the first moving thing they see — to argue that proximity-seeking behavior like crying, clinging, and following is instinctive, not learned.

Bowlby's collaborator, the developmental psychologist Mary Ainsworth, is the one who turned the theory into something you could actually measure. In the 1970s she designed a lab procedure called the Strange Situation: a caregiver and a roughly one-year-old are in a room together, a stranger enters, the caregiver briefly leaves, and then returns. What predicted an infant's attachment pattern wasn't whether they got upset during the separation — almost all of them did — it was what they did when the caregiver came back. Securely attached infants sought comfort and settled. Anxiously attached infants stayed distressed and hard to soothe even once reunited. Avoidantly attached infants shrugged the reunion off and ignored the caregiver, though later physiological studies found their stress response hadn't actually switched off — they were still activated, just outwardly composed. A fourth pattern, disorganized attachment, was added afterward by the researcher Mary Main to describe infants whose behavior didn't fit either strategy at all.

From the nursery to the relationship: Hazan and Shaver's leap

Attachment theory stayed a child-development topic for almost two decades. That changed in 1987, when social psychologists Cindy Hazan and Phillip Shaver published a study asking a genuinely strange question: what if adult romantic love runs on the same wiring as infant attachment? They put a version of Ainsworth's three infant categories in front of adults — describing each one in a sentence or two and asking people which sounded most like them — and ran it, somewhat famously, as a self-report survey in a local newspaper. The proportions that came back were close to what Ainsworth had found in infants decades earlier, and the adults who picked different categories described strikingly different romantic histories: how much they trusted partners, how they handled a partner needing space, how they reacted to being needed. Hazan and Shaver's argument was that the same "internal working model" a person builds toward a caregiver in infancy — is this person reliably there for me or not — gets reused, mostly without our noticing, as the template for how we approach partners as adults.

Two dials, not four boxes

For a while, a four-box version of adult attachment became the standard picture. Psychologists Kim Bartholomew and Leonard Horowitz refined Hazan and Shaver's three infant-derived styles into four in 1991, splitting "avoidant" into two distinct types — one built on a negative view of others but an okay view of self (dismissing), and one built on a negative view of both self and others (fearful) — based on a model of how a person's internal picture of themselves and of other people combine.

But boxes are a simplification of something messier, and, it turns out, more continuous. In 2000, R. Chris Fraley, Niels Waller, and Kelly Brennan published an item-response-theory analysis of existing adult-attachment questionnaires that became the basis for the Experiences in Close Relationships-Revised (ECR-R) scale. Their statistical analysis argued that attachment isn't four separate types a person either belongs to or doesn't — it's two continuous dimensions that everyone sits somewhere along, and the four "styles" are just useful shorthand for the four corners of that space.

• Low anxiety, low avoidance — Secure: comfortable depending on a partner and being depended on.
• High anxiety, low avoidance — Anxious (preoccupied): craves closeness, stays alert to any sign it's at risk.
• Low anxiety, high avoidance — Avoidant (dismissing): values independence, keeps emotional distance even when close.
• High anxiety, high avoidance — Fearful-avoidant: wants closeness and distrusts it at the same time.

This is also why a two-axis attachment assessment scores you on anxiety and avoidance separately instead of just handing you a single label — a quadrant is a more honest picture of where you sit than a box, because most people aren't parked in a corner. They're somewhere in the field, closer to one edge than another, and that position can shift.

What each pattern tends to look like

Secure attachment, which shows up as roughly half or more of adults in most samples, looks less like the absence of relationship problems and more like a working recovery system. Secure people get hurt, get jealous, get anxious sometimes — they just tend to say so directly, trust that the relationship can hold the conversation, and don't assume a bad night means the whole thing is ending.

Anxious attachment tends to show up as a heightened radar for a partner's warmth or distance — noticing a shorter reply, a change in tone, a canceled plan, and reading it as evidence something is wrong before checking. It often comes packaged with real generosity and depth of feeling; the cost is that reassurance tends to wear off fast, so the checking repeats.

Avoidant attachment usually looks like genuine self-sufficiency that turns into distance under pressure — handling stress alone by default, feeling crowded by a partner's need to process something out loud, and pulling back right when the relationship is asking for more closeness. It isn't that the connection doesn't matter; needing it openly just feels risky.

Fearful-avoidant attachment carries both pulls at once: wanting closeness enough to seek it out, and finding closeness frightening enough to sabotage or retreat from it once it arrives. It's often, though not always, associated with a chaotic or unpredictable early caregiving environment, where the person who was supposed to be a source of safety was also, sometimes, a source of fear.

None of these four is a character flaw or a diagnosis, and none of them means a relationship is doomed. They're strategies that, at some point, made sense — usually as a response to how reliable or unreliable the people a person depended on early in life actually turned out to be.

A pattern, not a life sentence

The most important finding in the modern attachment literature might be the one people hear about least: these patterns are stable on average, but not fixed for any given person. Longitudinal work from Fraley's lab estimates that roughly three in ten people show a meaningful shift in attachment style over the years, usually tied to a specific relationship — a securely attached partner, a good stretch of therapy, or simply enough evidence, accumulated over enough time, that someone can be trusted to stay.

The developmental psychologist Mary Main described this directly in what she called "earned security" — adults with genuinely difficult childhoods who, through therapy, a stable adult relationship, or plain reflection, end up able to talk about their past coherently and without being overwhelmed by it, and who test as securely attached despite an insecure start. Attachment isn't destiny; it's closer to a habit, formed early and reinforced by repetition, that responds — slowly, and with real effort — to different evidence.

Knowing you lean "anxious" or "avoidant" in the abstract is a start. Knowing where you actually fall on both dimensions — how much you fear losing closeness, and how much closeness itself unsettles you — is a more useful place to begin.