Attachment Theory in Adult Relationships: What the Science Says (Plus a Downloadable PDF)

Search “attachment theory in adult relationships” and you tend to get two kinds of results: academic papers locked behind paywalls, and quizzes that hand you a four-letter label and treat it like a horoscope. Neither is the thing most people actually want, which is a clear, honest primer they can read once and keep.

Attachment theory in adult relationships is the idea that the bond you form with a romantic partner runs on the same emotional system that once tied you to your earliest caregivers: a system built to seek closeness, protest separation, and use another person as a safe base to venture out from. It began with John Bowlby’s work on infants and was carried into adult love by Hazan and Shaver (1987). The modern version does not sort people into four sealed boxes. It places each of us somewhere on two continuous dimensions, and where you land can shift from one relationship to the next. This guide traces that arc from its origins to what the science does, and does not, claim.

Where attachment theory in adult relationships began: Bowlby to Hazan and Shaver

Long before anyone talked about attachment styles in dating, the theory was about babies. The psychiatrist John Bowlby proposed that human infants arrive with an evolved behavioural system whose job is to keep a caregiver close, because for a small child, proximity to a protective adult is a matter of survival. A distressed infant cries, reaches, and clings not out of habit but because those behaviours summon the person who keeps them safe. Hazan and Shaver (1987) open their landmark paper by summarising exactly this lineage: the observation that an infant uses a caregiver as a secure base to explore from and as a safe haven to return to when frightened.

Developmental researchers had already sorted infants into a few recognisable patterns based on how they behaved when a caregiver left the room and came back. Some children were upset at the separation but soothed quickly and went back to play; some stayed anxious and hard to comfort; some seemed indifferent and kept their distance. Hazan and Shaver (1987) borrowed these three patterns, secure, anxious, and avoidant, and asked a genuinely new question: what if adult romantic love runs on the very same system?

To test it, they published a “love quiz” in a newspaper and analysed the responses of hundreds of adults. The results were striking. Adults described themselves across the same three patterns, and in roughly the proportions seen in infant studies: about 56 percent secure, around 25 percent avoidant, and close to 19 percent anxious (Hazan & Shaver, 1987). More than that, the patterns predicted how people experienced love. The secure group described their most important relationship as trusting and durable; the avoidant group reported discomfort with closeness and difficulty depending on a partner; the anxious group described a preoccupied, sometimes turbulent love marked by longing and a fear of not being loved back.

The deeper claim was structural, not just descriptive. Hazan and Shaver (1987) argued that the hallmarks of the infant bond, seeking proximity, protesting separation, using the other person as a safe haven and a secure base, reappear almost feature for feature in adult romance. Your partner becomes the one you want near when you are scared, the one whose absence you protest, the base you feel steadier venturing out from. You can watch it in ordinary life: the person you text first when a medical result comes back worrying is your safe haven, and the relationship that makes it feel possible to take the risky job or the far-off trip is your secure base doing its quiet work. That reframing is what turned a theory about nurseries into a theory about grown-up love, and it is the foundation the rest of the field, and the broader map laid out in our attachment styles pillar guide, was built on.

The modern dimensional model: two continua, four regions

Hazan and Shaver’s three categories were a beginning, not the last word. The picture researchers eventually converged on treats attachment not as a set of types you belong to but as a position on two sliding scales.

The pivot came from Bartholomew and Horowitz (1991), who proposed a four-category model organised around two underlying dimensions: your internal model of yourself and your internal model of other people, each running from positive to negative. A positive model of self means you feel basically worthy of love and not braced for abandonment; a positive model of others means you expect people to be available and trustworthy. Cross the two axes and you get four regions. Secure sits at positive self and positive other. Preoccupied is positive other but negative self: wanting closeness while fearing you are not enough to keep it. Dismissing is positive self but negative other: comfortable alone and wary of leaning on anyone. Fearful is negative on both: wanting closeness and distrusting it at the same time.

Those two axes are the ancestors of the terms the field uses today. The self dimension maps closely onto what is now called attachment anxiety, how much you worry about rejection and crave reassurance, and the other dimension onto avoidance, how much you keep emotional distance and suppress your need to depend on a partner. Low on both is the secure region. High anxiety with low avoidance is the preoccupied corner. Low anxiety with high avoidance is the dismissing corner. High on both is the fearful corner.

Here is why this matters, and why it is worth resisting the pull of the four tidy labels. In the dimensional model, the four “types” are not boxes with walls between them. They are regions on a continuous map, and most people do not live at the extremes. You might sit moderately high on anxiety and only slightly above the middle on avoidance, a combination no single label captures well. Two people both called “anxious” can be a world apart if one is near the centre and the other is far out toward the edge. One might simply prefer a goodnight text and feel briefly unsettled without it; the other might spiral when a message goes unanswered for an hour. Same word, very different lived experience, and collapsing them into one “type” flattens exactly the information you need. The label is a neighbourhood, not an address, and definitely not a personality.

Seeing your own coordinates on that plane turns out to be more useful than collecting a label, which is why a visual adult attachment style chart that plots the two dimensions tends to click for people faster than a list of four descriptions. Once you can read the map as a space rather than four bins, the rest of attachment theory in adult relationships gets much easier to hold honestly.

How adult attachment is actually measured

If attachment lives on two dimensions, then measuring it should give you two numbers, not one identity, and that is exactly how the standard adult measures work. In their review of the field, Fraley and Shaver (2000) describe how self-report questionnaires ask you to rate a series of statements about how you think, feel, and behave in close relationships, then score you on the two dimensions of anxiety and avoidance. Your “style” is simply which quadrant those two scores land in. Nothing in the process discovers a hidden type. It locates a point.

That distinction should change how you read any result. When a questionnaire tells you that you are “anxious,” it is rounding two continuous scores into the nearest labelled region. Someone one point away on either scale might get a different word for a barely different reality. Fraley and Shaver (2000) treat this dimensional structure as one of the more settled conclusions of the research, and it is the reason a careful adult attachment questionnaire reports where you fall on each dimension rather than just stamping you with a category and calling it a day.

The other reality the labels hide is that attachment is not entirely one fixed thing about you. Fraley and Shaver (2000) discuss the difference between a person’s general, or global, attachment orientation and their attachment within a specific relationship. You can carry a broad tendency and still feel noticeably more secure with one partner than you did with another, or more anxious with a new person who has not yet proven reliable than with someone you have trusted for years. That is not inconsistency. It is the system doing what it is built to do: responding to the actual person in front of you and the history you share with them. The same person can be the secure partner in one pairing and the anxious one in the next, which is why a global label pinned to your name will always be a little too coarse to be the last word.

Fraley and Shaver (2000) are also candid about the parts that remain unsettled. Self-report measures and interview-based measures of attachment do not always agree, and they may be tapping somewhat different things, so no single questionnaire should be treated as the final verdict on you. It is also worth noticing the pull to over-identify, to read one result and start narrating your whole romantic history through it. The pattern is a lens, not a diagnosis, and the moment it starts explaining everything a little too neatly is the moment to loosen your grip on it. This is a good reason to hold your result lightly. A measure is a snapshot of two dimensions, at one moment, from one instrument, in one relational context. It is genuinely useful, and it is not an X-ray of your soul.

What attachment theory in adult relationships does not claim

Because the labels are sticky and a little dramatic, it is easy to inflate attachment theory into something it was never meant to be. Three overreaches are worth naming plainly.

First, it does not claim your attachment is fixed. The definitive scholarly synthesis of the adult research is titled, pointedly, Attachment in Adulthood: Structure, Dynamics, and Change, and Mikulincer and Shaver (2007) spend much of it on how attachment patterns shift over time and in response to experience. A steady, responsive relationship can gradually move someone toward greater security; loss, betrayal, or a long stretch of stress can move them the other way. Picture someone who spent years bracing for withdrawal with a distant partner, then finds that a reliably present new partner slowly makes the bracing feel unnecessary. That is not a personality transplant. It is the attachment system updating on new evidence, which is what Mikulincer and Shaver (2007) mean by change. Fraley and Shaver (2000) put a finer point on it from the data side: attachment patterns show meaningful but far from perfect stability across time, which is precisely what you would expect from a system that is real and enduring but not carved in stone.

Second, it does not claim to be deterministic. Your position on the two dimensions describes a tendency, a rough probability for how you are likely to react, not a script for any single moment. Mikulincer and Shaver (2007) describe attachment as a dynamic system that switches strategies depending on the situation: under threat, people who lean anxious tend to hyperactivate, turning up the protest and the demand for reassurance, while people who lean avoidant tend to deactivate, turning down the felt need and pulling back. Which strategy shows up, and how strongly, depends on the partner, the context, and the moment, not on a fixed dial set once in childhood.

Third, it does not claim to be a horoscope. A label is not an explanation for everything you do, an excuse for how you treat a partner, or a fixed personality you are stuck performing. Used well, attachment theory in adult relationships describes patterns so you can recognise them and work with them. Used badly, it becomes a way to file yourself and your partner into permanent categories and stop trying. The patterns are worth taking seriously precisely because they can change, which is a very different message from “this is just who I am, so deal with it.”

None of this dissolves the theory. The dimensions are stable enough to shape the way you and a partner communicate and predictable enough to be worth learning. They are simply not a life sentence, and the whole attachment styles pillar guide is built around that more hopeful, and more accurate, reading.

What to try this week

Give yourself five quiet minutes and two questions. First, on a scale of zero to ten, how much do you generally worry about a partner pulling away, or find yourself needing reassurance that things are okay? That is your rough attachment anxiety. Second, on the same scale, how much do you tend to keep a little distance, downplay your needs, or feel uneasy depending on someone? That is your rough avoidance. Plot the two numbers as a single point on a grid. That point, not a label, is you right now.

Then do the more interesting part: run the same two questions for two past relationships, or for a current one on a good week versus a hard one, and watch how the point moves. The shift is the whole lesson. It shows you, in your own history, that attachment is relationship-specific and responsive rather than a fixed trait. Because these patterns also shape how you give and receive care, it can help to pair this with how attachment interacts with love languages; a quick Love Language Quiz is one easy way to see how you most readily take in reassurance. If the same painful position keeps showing up across every relationship, a structured Twogle Check-In can give you and a partner a low-stakes place to name it, and a few sessions with a therapist trained in attachment work is a reasonable next step, not a sign that anything has failed.

Hold your coordinates lightly, watch how they move, and let the map be a starting point rather than a verdict.

Read deeper

• Attachment styles: the complete pillar guide
• Adult attachment style chart
• Adult attachment questionnaire
• How attachment styles affect communication
• Attachment style and love languages
• Take the Twogle Love Language Quiz

Sources

• Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61(2), 226–244. DOI: 10.1037/0022-3514.61.2.226
• Fraley, R. C., & Shaver, P. R. (2000). Adult romantic attachment: Theoretical developments, emerging controversies, and unanswered questions. Review of General Psychology, 4(2), 132–154. DOI: 10.1037/1089-2680.4.2.132
• Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52(3), 511–524. DOI: 10.1037/0022-3514.52.3.511
• Mikulincer, M., & Shaver, P. R. (2007). Attachment in Adulthood: Structure, Dynamics, and Change. Guilford Press.