Can Attachment Patterns Change in Adulthood?

What longitudinal research says about stability, life events, relationships and therapy, and why “attachment can change” is true without meaning that every change is fast, permanent or caused by the right partner.

Attachment is stable enough to matter and flexible enough to change

Two claims circulate online.

Your attachment style was created in childhood and follows you for life.

and:

The right relationship can completely heal your attachment style.

Both contain part of the truth.

Neither is a good summary of the evidence.

Adult attachment shows meaningful stability.

People do not usually wake up every month with an entirely new relational orientation.

But attachment is not perfectly fixed.

It can fluctuate.

It can change gradually.

Some life events produce temporary shifts.

Some changes endure.

Age, relationships, therapy and major experiences can all be associated with movement.

The scientifically interesting answer is therefore:

Yes, attachment patterns can change in adulthood, but change is variable, often modest, and not always permanent.

First, define what is changing

This question becomes confusing because adult attachment is measured in several ways.

Some researchers use self report questionnaires that measure romantic attachment anxiety and avoidance.

Others use interviews such as the Adult Attachment Interview, which assess the organization and coherence of attachment related narratives about earlier caregiving.

These methods overlap much less than many people assume.

A meta analytic comparison found only trivial to small associations between AAI security and self reported romantic attachment dimensions.

So “attachment changed” can mean different things depending on the measure.

A person might become less anxious with a partner without showing the same change on an interview measure of attachment representation.

Method matters.

Stability is real

A meta analysis of attachment stability from infancy into early adulthood included more than 21,000 attachment observations.

The overall stability coefficient was moderate, around r = .39.

Stability was weaker over longer intervals.

No significant stability was found across intervals greater than fifteen years in that analysis.

This is a useful middle position.

Attachment is not random.

Earlier patterns have predictive value.

But decades leave room for enormous experience.

A pattern that matters developmentally is not necessarily a permanent identity.

Across adulthood, average insecurity can decrease

A longitudinal study combined three samples and followed attachment orientation across as much as 59 years of life, from adolescence into older adulthood.

Attachment anxiety declined on average, particularly in middle and later adulthood.

Attachment avoidance also declined across the life course.

Being in a romantic relationship was associated with lower anxiety and avoidance.

These are population level trends.

They do not mean age automatically makes everyone secure.

Some individuals become more insecure.

Some remain stable.

But the average trajectory does not support the idea of complete lifelong rigidity.

Life events can move attachment, but many effects fade

One of the strongest naturalistic studies followed more than 4,000 people through life events such as starting relationships, changing jobs and other major experiences.

About half of the events studied were associated with immediate changes in attachment.

But on average, people tended to move back toward the level of security expected from their previous trajectory.

In other words:

change often happened, but regression toward the old pattern was common.

Still, around a quarter of the events were associated with enduring average change.

And even when the average effect did not persist, individuals varied.

Some people became durably more secure.

Others became durably less secure.

This is one of the clearest answers to the question.

Adult attachment is plastic.

Plasticity is not the same as instability.

Relationship specific attachment can fluctuate even when trait levels are relatively stable

A multi study longitudinal program examined within person fluctuations in attachment security toward a specific romantic partner.

Most people showed some fluctuation over time.

Those fluctuations were associated with relationship well being.

This shows why a single questionnaire result should not be treated as an eternal classification.

Attachment has both trait like and state like features.

A person may have a general orientation and still move around that baseline depending on what is happening in a particular bond.

This resembles many psychological traits.

Temperament can be stable while mood changes.

Attachment can be stable while relationship specific security moves.

Does a good partner make you secure?

Sometimes people describe “earned security” in romantic relationships as if a consistently loving partner can retrain the attachment system.

There is some reason to think supportive relationships can matter.

Attachment theory itself predicts that repeated experiences of responsiveness can update expectations.

But the empirical picture is not simple.

A 2025 longitudinal dyadic study followed 1,036 established couples across twenty months.

Attachment anxiety showed a small decrease over time for women.

Avoidance was largely stable.

The measured relationship characteristics, including satisfaction, commitment, support, closeness, responsiveness and disclosure, explained little of the changes overall.

This is a valuable corrective.

A good relationship may support security.

It is not a guaranteed attachment treatment.

Relationship duration matters, but not always in the direction people expect

Some longitudinal work has found that romantic partners can become increasingly important attachment figures as relationships progress.

Avoidance toward the romantic partner can decrease as the bond develops.

At the same time, a meta analysis found that the negative associations between insecurity and relationship satisfaction or commitment tended to become stronger in longer relationships.

Both findings can coexist.

Long term relationships create opportunities for greater security.

They also expose regulatory patterns repeatedly.

Time alone does not resolve insecurity.

Repeated relational experience matters more than simply remaining together.

Therapy can be associated with increased security

A meta analysis of 36 psychotherapy studies involving 3,158 patients found that people entering therapy with greater attachment security tended to have better outcomes.

More relevant to change, increases in attachment security during therapy were associated with better treatment outcome.

This does not prove that psychotherapy universally changes attachment.

The studies varied in diagnoses, methods and attachment measures.

But it is evidence that security is not completely static in treatment contexts.

A 2024 study of therapeutic attachment similarly examined how clients can develop greater secure attachment to a therapist and how this relates to growing comfort with closeness or autonomy.

Again, therapy can provide attachment relevant experience.

It does not guarantee a global permanent shift.

Security priming shows short term malleability, not permanent healing

Researchers have also experimentally activated mental representations of security.

A 2022 meta analysis included 120 studies and nearly 19,000 participants.

Attachment security priming produced moderate positive effects across affective, cognitive and behavioral outcomes.

This is interesting evidence for momentary malleability.

It should not be oversold.

If imagining or activating a secure relationship representation changes behavior for a period, that shows the system can move.

It does not prove a person’s long term attachment orientation has been permanently rewritten.

State change and trait change are different claims.

Major transitions can create both stability and change

The transition to marriage, parenthood, separation, loss and other major relational events can activate attachment systems strongly.

A 2026 study following new mothers from pregnancy through two years postpartum found both stability and change in attachment anxiety and avoidance, with predictors varying across dimensions.

Earlier marriage transition research found most people retained their broad AAI classification across assessment points, while some changed toward greater security.

The repeated lesson is the same:

attachment is constrained by history but responsive to experience.

Why old patterns return under stress

Someone may feel secure for months.

Then a serious conflict activates familiar reactions.

They think:

“Nothing changed. I am back to my old attachment style.”

Not necessarily.

Learning is rarely erased completely.

Newer patterns can coexist with older responses.

Under high stress, well practiced strategies may become more accessible.

Change is better evaluated through:

• frequency;
• intensity;
• recovery speed;
• behavioral choice;
• ability to repair;
• flexibility across contexts.

The question is not whether the old response ever appears again.

It is whether it still controls the system in the same way.

The partner cannot do the work for you

The idea that “a secure partner heals an insecure person” can create dangerous relationship expectations.

One person becomes responsible for providing endless reassurance.

Or for tolerating withdrawal.

Or for proving that love is safe.

Partners can contribute to security.

They can also contribute to insecurity.

But attachment change is not a service one person performs on another.

Adult change involves multiple processes:

• repeated experience;
• self observation;
• emotion regulation;
• communication;
• corrective relationships;
• boundaries;
• sometimes psychotherapy;
• changed interpretations;
• changes in life context.

No partner can guarantee the outcome.

The DarkBrain model: baseline, activation, repetition, consolidation

A useful way to think about attachment change is through four layers.

Baseline

What is the person’s general level of anxiety or avoidance across time?

Activation

What situations temporarily move the system?

Separation? Commitment? Conflict? Loss? Support?

Repetition

Are new responses occurring repeatedly enough to become familiar?

Consolidation

Does the change survive new situations and later stress?

This framework separates momentary calm from durable change.

A person may feel secure on holiday.

That is not the same as being able to stay engaged during conflict six months later.

What the evidence does not support

The evidence does not support these universal claims:

• attachment style never changes after childhood;
• everyone becomes more secure with age;
• a secure partner will heal an insecure partner;
• one breakup permanently makes someone avoidant;
• therapy always creates earned security;
• a short term security exercise rewires attachment permanently;
• every change in a questionnaire score is a deep structural change.

Change exists.

Its depth has to be demonstrated over time.

The answer

Can attachment patterns change in adulthood?

Yes.

Some changes are temporary.

Some are relationship specific.

Some are gradual age related shifts.

Some follow major life events.

Some occur during therapy.

Some endure.

But stability remains real enough that old regulatory strategies can reappear, especially under stress.

The mature position is neither fatalism nor magical optimism.

It is:

history creates probability, not destiny.

Continue exploring

Next: Earned Security: What the Concept Means

The final article examines one of the most misunderstood terms in attachment language and why earned security is a specific research concept, not simply a synonym for “I used to be anxious and now I feel secure.”