Attachment is not fate: what changes in adulthood and what resists

“I understand now where this comes from in me. So why does nothing change?”

I hear that question more often than any other. Someone sits across from me, able to describe how it was at home, aware that their mother was often absent and their father unpredictable, able even to name their attachment pattern — and then on Friday evening, when a partner does not reply to a message, the same hole opens in their stomach as when they were seven. The insight is there. The change is not. And everything therapy does sits between those two points.

In this piece I am not going to promise that an attachment pattern can be outgrown. I will write something more useful: what shifts relatively quickly in adulthood, what shifts slowly, and what is stubborn enough that living with it beats trying to erase it. I will not repeat the basics here — I wrote about how these same patterns repeat from relationship to relationship in my piece on attachment patterns.

An attachment pattern is not character, it is a default setting

Let us first clear up a mistake that makes change look impossible. An attachment pattern is not a trait you carry in equal measure across every situation, the way you carry your height or the colour of your eyes. It is closer to a default setting: what your system does when you do not adjust it — and when it is under pressure.

That has two consequences. The first is encouraging: defaults can be overwritten. The second is less so: when the system has no resources — when you are tired, ill, frightened or in conflict — the default returns, because it is the cheapest option available. This is why people so often say they have “gone backwards” when in fact they simply hit a day with no reserve left.

The pattern is also relationship-specific. With one person you can be fairly secure, with another anxious, and with a manager avoidant. That is not inconsistency. It is evidence that the pattern is made in relationship, and changes in relationship too.

Three levels that do not move at the same speed

When I watch what actually shifts in a person, I see three levels, each with its own tempo.

Behaviour is what you do: whether you call, whether you go silent for three days, whether you leave the room, whether you ask. Narrative is the story you tell yourself about it: “he is not interested”, “I will be faster on my own”, “if I show how I feel, she will leave”. The felt sense is what happens in the body: the tight chest, the heat in the face, the emptiness under the ribs that arrives before any thought does.

These three levels do not move together in therapy, and they do not move in a random order. It nearly always goes the same way: behaviour first, then narrative, and the felt sense last. That lag is exactly why the middle stretch is so unpleasant — you are doing things differently while feeling precisely the same.

Behaviour moves first — and that counts as change

The first thing that changes in people is almost always what they do. Someone with an anxious pattern learns not to send a fourth message after three go unanswered. Someone avoidant learns to say “I need an hour to myself and then I will come back” instead of disappearing. This happens relatively quickly — for many within weeks or a few months, because it is something you can practise.

People often dismiss this shift: “I was only performing, inside I felt exactly the same.” I do not want you to dismiss it. Behaviour is the only level you have direct access to, and every time you act differently you walk through an experience that contradicts your narrative. You did not send the fourth message — and the relationship did not fall apart. That is data your system did not previously hold.

The narrative moves second

Once enough of that data accumulates, the story starts to loosen. Not because you argued it down, but because it stops being self-evident. Instead of “he is not interested” you get “maybe he is not interested, or maybe he is just tired” — and that doubt is the progress. Security is not being right every time; it is having more than one explanation available.

For most people this shift is a matter of months, not weeks. Anything that loosens your fusion with a thought helps here: cognitive defusion, work with the inner critic, and writing. Not because the thoughts are untrue, but because a thought you can look at from the outside loses its status as fact.

The felt sense comes last and is the most obstinate

And then there is the level that convinces people they have made no progress. The body. The clamp in the chest when a partner says “we need to talk”. The heat when someone praises you and you do not know where to put it. The cold that comes over you when someone moves in too close, too fast.

This level changes most slowly because it was never learned in words. It formed before you had words — in thousands of repetitions of what happened when you cried. The body learned to predict, and a prediction is not deleted; it is overwritten by a new one, which takes a great many repetitions. This is the same mechanism I described in my piece on neuroplasticity and therapy: nothing disappears, but the new becomes stronger than the old.

The honest forecast is this: at the level of the body we are talking about a year or two of steady work, and for people with early, repeated trauma, longer. That is not pessimism. It is the reason not to give up in month six, when the behaviour already holds and the feeling does not.

Why the old pattern returns under stress

Now for the part that confuses people most: relapse. Someone works for six months, hits a conflict with a partner at two in the morning — and within fifteen minutes is back exactly where they started. Shutting down, accusing, walking out, sending eleven messages.

That is not lost progress. It is a system built to reach, under threat, for the fastest and most well-worn response it has. The new pattern is still expensive: it costs attention, language and time. The old one is free. When arousal is high — the same mechanism I describe in my piece on hyperarousal and insomnia — the expensive things switch off first.

So a more useful measure than “did it happen” is this: how long did it take me to come back. The first time, a week. A year later, perhaps two days. Two years on, an afternoon. The pattern does not stop appearing; it becomes shorter, shallower and more recognisable from the inside.

What the research says about stability

The data are more interesting than you might expect. Attachment patterns do persist from childhood into adulthood, but weakly: the associations are moderate rather than strong, and they weaken as the gap in years grows. Fraley’s analyses show a stability large enough not to be chance, and small enough not to be fate.

I like the second finding even more. Change is not random: people move towards security mainly when they spend a long stretch inside a relationship that actually gives them that experience. Not through technique, but through exposure to something different. A partner, a friendship, a therapeutic relationship — all work by the same principle, because all offer a repeatable experience that a response arrives.

And a third: attachment predicts how therapy will go, not whether it will work. People with an anxious pattern attach faster and end with more difficulty; people with an avoidant pattern need longer before they speak at all — yet outcomes are comparable for both if they stay.

What changes faster than you expect

Some things move surprisingly quickly, because they are not the pattern itself but its consequence. How fast you notice that you are inside a reaction. Whether you can say what you need instead of signalling it through silence. Whether after an argument you can come back and say “I was harsh earlier”. Repair is a skill, not a personality, and repair is what contributes most to a relationship becoming secure.

Who you are drawn to shifts quickly too. People often notice that the type who used to pull them in suddenly holds no interest. That is not a healed attachment system. It is, however, a very good sign that the measure has moved.

What is slow or stubborn

For honesty’s sake, the other side as well. Slow is the bodily expectation I described above. Slow is the sense that you are worth someone staying for — in many people that stays fragile long after they have learned to ask for closeness. Stubborn are the responses formed under prolonged threat in childhood, because there the lesson was not about relationships but about survival.

One more thing is stubborn: change is not even across relationships. Someone becomes secure with a partner and remains entirely avoidant with their mother. That is not half a success. The relationship with parents is often the last to move, and sometimes it never moves — and then the task is a different one: setting boundaries rather than getting closer.

Where this piece does not hold

Now let me argue against myself. Everything I have written assumes that the core of the difficulty is an attachment pattern. Often it is not.

If closeness exhausts you because you have broken nights and two small children, that is not avoidant attachment but depletion — and the answer is sleep, not insight. If a partner’s silence sends you into panic because a previous partner left without explanation, that is the trace of a specific event, not of childhood. And if all this is happening in a relationship where someone genuinely is unpredictable, your “anxiety” is not a fault in your system but an accurate perception.

Attachment is also a weak frame for a good deal of what we hang on it: for depression, for anxiety disorders, for physical symptoms with no findings. If you have weeks when you cannot get up, you do not need a pattern analysis, you need treatment for a condition. And finally: working with attachment patterns carries a real risk of becoming self-justification. “That is just me, I am avoidant” is not an explanation but an excuse, whenever it closes a conversation instead of opening one.

How to tell whether it is moving

Because the feeling changes last, you need measures that do not rest on it. These four are useful: how long it takes you to come back after a trigger. Whether you can notice, mid-reaction, that you are in one. Whether you are at least sometimes the one who returns first after a conflict. And whether you can say what you need in one sentence without a reproach in it.

All four are observable even when you feel exactly as you did a year ago. I have gathered more on what actually counts as progress in my piece on signs that therapy is working.

What to remember

An attachment pattern is not character but a default setting — and defaults get overwritten, even though they are not deleted. Behaviour shifts in weeks, narrative in months, the felt sense in years; that middle stretch, when you already act differently and still feel the same, is a normal part of the road, not proof of failure.

Under stress the old pattern returns, because it is the cheapest response you have. Do not measure progress by whether it happened, but by how quickly you came back. And do not forget the dullest finding in the research: what moves you towards security is not understanding, but a long enough experience of a relationship in which the response genuinely arrives.