A 700-person longitudinal study tracked children from infancy to age 18 and found that sensitive caregiving in the first 3 years still predicted adult attachment security 18 years later (Fraley, Roisman, Booth-LaForce, Owen & Holland, 2013).
Bowlby called this template the internal working model. It is the silent operating system of every relationship you have.
The template
A baby has no built-in expectations about how relationships work. It watches.
- Did crying bring comfort, or didn’t it?
- Did reaching out get met, or ignored?
- Did caregivers come back?
Across thousands of repetitions, a prediction system gets built. The brain compresses the data into a single model: “this is how love works.”
It is not a memory. It is a forecast.
What the model actually does
Three jobs, all automatic:
- Filters perception. You notice signs that match the model. You miss signs that don’t.
- Generates expectations. Before the other person speaks, you’ve already predicted what they’ll do.
- Triggers behavior. You react to the prediction, not the action.
This is why two people can read the same neutral text (“running late, see you in an hour”) and one feels comforted, the other panics. Same words. Different model.
How long it lasts
The data behind this finding (the NICHD Study of Early Child Care: a 700-person observational study that followed children from birth to age 18) is unusually clean:
- Early sensitive caregiving moderately predicted adult attachment security
- The effect was real but modest: r ≈ 0.10 to 0.25 across the lifespan
- About 35 to 45% of adult attachment variance is heritable
- The rest is shaped by current relationships as much as past ones
The template is moderate, not deterministic. It is bias, not destiny.
Why “secure / anxious / avoidant” is misleading
- The original 3-category attachment styles are statistical fictions (Fraley & Waller, 1998)
- Real attachment lives on two continuous dimensions: anxiety and avoidance
- You are not a “type.” You are a coordinate.
- The coordinate moves over time, with effort
Why this matters
- Your default reactions are predictions, not perceptions. You react to the forecast before the moment arrives.
- The model is moderate, not fixed. Real change happens. Effect sizes leave room for it.
- Current relationships rewrite the model. A consistent partner can reshape an insecure template.
- The model isn’t your fault. No one chooses their first three years.
- The model is your responsibility now. You can notice when you’re reacting to the template instead of the present.
You can’t choose what you learned. You can choose whether to update it.