Becoming More Secure: Can Attachment Style Change?
PsyCove Editorial · 10 min read · Updated 2026-10-09
- Attachment is moderately stable, not fixed. Longitudinal work finds continuity across years alongside real movement, with a substantial minority of adults changing category within a few years.
- Insight does not move a working model; disconfirming experience does. You can describe your pattern accurately and still run it, because the pattern executes faster than deliberate reasoning.
- Change usually attaches to an identifiable event: a stable relationship, the end of a chaotic one, parenthood, therapy, or a sustained drop in life stress.
- Progress shows up as shortened episodes rather than a new personality: the same interpretation appears, lasts twenty minutes instead of two days, and is noticed as an interpretation.
- Re-measure with the same instrument at three and twelve months. One score tells you almost nothing; the difference between two scores taken months apart tells you nearly everything.
What the evidence actually shows
Attachment is moderately stable, which is not the same as fixed. Longitudinal work finds that a substantial minority of adults change category across a few years, and that change is often tied to identifiable events: entering a stable relationship, ending a chaotic one, becoming a parent, starting therapy, or a sustained drop in life stress.
The mechanism is straightforward even when the timetable is not. Attachment is a working model — a set of expectations about whether reaching out works. Expectations update when they are disconfirmed repeatedly. One good relationship does not overwrite twenty years of contingency; it begins to, and then continues to.
Why insight alone does not move it
Most people who recognize their pattern can describe it accurately and still run it. That is not hypocrisy; it is the difference between knowing the rule and having the skill. The pattern executes at a speed and in a physiological register that deliberate reasoning arrives after. Practice has to happen in the triggered moment, not in the calm review of it.
A workable sequence
- Map your triggers concretely: the specific cue, the story you build, the action you take, and what it gets you in the next hour.
- Pick one recurring situation and rehearse a different move in advance, so it is available at speed.
- Reduce the background load. Sleep loss, financial stress and work overload lower the threshold for every attachment reaction.
- Choose relationships that reward directness. Environments that punish asking will keep any insecure pattern in place no matter how much you work on it.
- Re-measure at intervals. Comparing the same instrument across months is the only way to know whether the pattern is actually moving.
What change looks like from the inside
Progress rarely feels like confidence. It usually feels like a shorter episode: you still interpret the silence as distance, but the interpretation now takes twenty minutes instead of two days, and you notice it as an interpretation rather than as a fact. That shortening is the real signal, and it is easy to miss if you are waiting to feel permanently different.
You do not graduate out of a pattern. You get faster at noticing it, and the episodes get shorter and cheaper.
When to get help
If the pattern is tied to trauma, to sustained fear, or to behavior you cannot interrupt on your own, working with a clinician is not a failure of effort — it is using the mechanism that changes working models most reliably. Approaches that work directly with the interpretation layer and with tolerance of uncertainty tend to be the relevant ones.
How much does attachment actually change? What longitudinal studies find
The honest answer is that attachment is among the more stable individual differences in social psychology — Fraley's modeling of retest data puts the correlations in the moderate range — and also genuinely changeable, and those two claims are not in conflict. Stability estimates describe a population average over a period; they do not describe a ceiling on any individual. Longitudinal studies that follow the same people for years find continuity, and also find that a substantial minority shift category — usually toward security, and usually in connection with something identifiable that happened to them.
Stability is real, and it is not immobility
A high stability coefficient means people tend to keep their relative rank within a group over time; it does not mean nobody's score moves. Mean-level change also occurs, with samples tending to shift slightly toward security across early adulthood, which fits the idea that accumulated adult relationships do updating work. Practically, expect your own score to be recognizable a year from now and still expect it to be different — those are compatible predictions rather than contradictory ones.
Change tends to attach to events
When researchers examine who actually changed, the change is rarely unexplained. Entering a stable and responsive relationship, leaving a chaotic one, becoming a parent, starting therapy, and sustained reductions in life stress appear repeatedly in the data. This is the strongest available argument that the mechanism is experiential: the expectation moves when the evidence moves, and the evidence moves when your circumstances do. It also explains why change is often uneven across relationships rather than global.
What does not seem to move it
Reading, insight, and intention are the three things people most often expect to work, and the evidence for them as standalone agents is weak. They are not worthless — language shortens the time to recognition, and recognition is the entry point for everything else — but none of them supplies the repeated disconfirmation that a working model requires. A weekend workshop and a two-year relationship are not remotely comparable interventions, even when the workshop is excellent.
How to measure change without fooling yourself
Tracking change is harder than it looks, because the instrument is self-report and self-report is exactly what an insecure pattern distorts. Two safeguards do most of the work: use the same instrument every time, and treat any single reading as noise. The comparison that carries information is the same items, months apart, answered with the same relationship in mind.
- Use one instrument and stay with it. Different measures place their boundaries differently, so an apparent shift may be a change of instrument rather than a change in you.
- Answer with a specific relationship in mind and note which one. Global answers average across contexts and hide the movement that actually matters.
- Space the readings. Three months is the shortest useful gap; weekly re-testing mostly tracks mood rather than the pattern underneath it.
- Record episodes, not only scores: how long the last incident lasted, and whether you noticed it while it was running. Shortening appears here before it appears in the score.
- Add one external observer. A partner or close friend who sees you regularly will notice changes in the distance you keep before you notice them yourself.
- Do not measure during a crisis. Acute stress inflates both anxiety and avoidance scores in anyone, secure people included, so the reading will not tell you about the underlying pattern.
The routes that actually move a working model, ranked by evidence
Three routes produce the disconfirming evidence that revises an expectation, and they are not equally efficient. They also work better together than separately: a relationship supplies the raw input, therapy supplies the conditions for taking it in, and deliberate practice supplies the repetition at the specific moments where the pattern runs.
- A relationship that reliably answers. The most efficient single route, because it delivers the exact disconfirmation — reaching out worked — hundreds of times, in conditions you cannot arrange artificially.
- Therapy, particularly approaches that work directly with the relational pattern. The therapist functions as a temporary secure base and, critically, does not disappear when the defensive sequence fires.
- Deliberate practice on one specific situation. Pick the moment that reliably goes wrong and rehearse one alternative action until it is available at speed.
- Reduced background load. Sleep loss, financial pressure, and work overload lower the threshold for every attachment reaction, and lowering the load makes the other routes affordable.
- Time. Expectation revision operates on the order of months to years; any protocol promising a shorter timetable is selling something other than a revised working model.
What change looks like week by week
Movement toward security is usually misread because people wait for a feeling — calm, trust, confidence — while the actual signal is a change in the shape of episodes. The sequence below is the common pattern rather than a promise, and most people find it uneven: better in one relationship while unchanged in another.
Weeks one to four: recognition
The first change is that you notice the pattern while it is running rather than after it has finished. Nothing else improves yet, and this can feel like getting worse, because you are now watching something you previously only suffered. The practical marker is a log: the cue, the story, and the action, written down within an hour.
Months two to six: shortening
The same interpretation still appears, but it does not hold as long. What used to occupy two days now occupies an afternoon, and recovery requires less from other people. This is the phase partners usually notice first, because the observable change is behavioral: fewer tests, fewer disappearances, more sentences.
Year one and beyond: the default shifts
Later change is quieter still and harder to date. The interpretation appears, is recognized, and fails to recruit the action that used to follow it. At this point re-taking the original measure is the only honest check — the change is real but does not feel like an achievement, which is why many people conclude nothing happened while the score says otherwise.
Security is not the absence of the old reaction. It is a shorter distance between the trigger and the moment you recognize it as a reaction.
Becoming secure vs. becoming indifferent vs. becoming compliant
A common failure mode is to mistake suppression for change. Reducing the intensity of your reactions by deciding the relationship does not matter is not movement toward security; it is movement toward avoidance, and it often converts an anxious pattern into a dismissive one. Security is defined by flexibility rather than by calm: the secure response to a gap is to notice it, form a benign explanation, take proportionate action, and return to what you were doing.
Compliance is the second failure mode: staying in contact you do not want because you have read that avoidance should be corrected, or tolerating distance you find painful because you have read that the anxiety is yours to manage. Both are adaptations to a framework rather than to a relationship. The test is whether your range has widened — whether you can now do the thing that used to be unavailable — or whether you have simply substituted one rigid rule for another.
When the pattern will not move, and when to get professional help
There are conditions under which self-directed work reliably fails, and they are worth naming so that failure is not misread as personal inadequacy. If the people around you are genuinely unreliable — if asking is punished, if affection is withdrawn unpredictably, if there is contempt or coercion — the pattern is an accurate reading of the environment and will not move until the environment does. No amount of practice substitutes for leaving a relationship that keeps rewarding your worst prediction.
Separately, some patterns are anchored in experiences that self-guided practice does not reach. Where the pattern is tied to trauma, to sustained fear in childhood, or to behavior you cannot interrupt on your own, working with a clinician is the efficient route rather than the last resort. The mechanism is the same one that operates everywhere else: a real relationship that does not disappear, in which the old expectation can be tested and survives being tested.
Frequently asked questions
How long does it take to become more secure?
Measurable shifts are typically reported over months, with larger category changes over a few years. The variable that matters most is not time but repeated disconfirmation: how often your expectation is contradicted by a real, safe response.
Can I do it without a partner?
Partly. Non-romantic relationships — a therapist, a reliable friend, a stable group — update the same model. Romantic relationships are simply a high-intensity version of the same input, not a different mechanism.
Does reading about it help?
It helps by giving you language and by shortening the time to recognition, which is genuinely valuable. It does not, on its own, provide the repeated disconfirmation that moves the expectation.
How do I know whether I have actually changed?
Re-take the same instrument at three and twelve months and compare the two. A single reading is close to noise; the difference between two readings taken months apart, with the same relationship in mind, is the measurement that carries information. Track episode length as well, because shortening shows up there first.
Can becoming more secure mean caring less?
No, and that confusion causes real damage. Reducing your reaction by deciding the relationship does not matter is movement toward avoidance, not security. Security is flexibility: you still want contact, you can ask for it directly, and a delay no longer requires an emergency response. If you have gone quiet rather than steady, the pattern has changed shape rather than improved.
References
- Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78(2), 350–365.
- Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1–25.
- Fraley, R. C. (2002). Attachment stability from infancy to adulthood: Meta-analysis and dynamic modeling of developmental mechanisms. Personality and Social Psychology Review, 6(2), 123–151.
- Fraley, R. C., Gillath, O., & Deboeck, P. R. (2021). Do life events lead to enduring changes in adult attachment styles? A naturalistic longitudinal investigation. Journal of Personality and Social Psychology, 120(6), 1567–1606.
- Fraley, R. C. (2019). Attachment in adulthood: Recent developments, emerging debates, and future directions. Annual Review of Psychology, 70(1), 401–422.