Attachment theory: forty years of research and its critics

Mary Ainsworth's 1969 procedure was designed to be unobtrusive. A mother and her one-year-old play in a room. A stranger enters. The mother leaves briefly. The mother returns. The baby's response to the return — the reunion — became the most informative moment.

From thousands of observations, Ainsworth identified three patterns of attachment behavior: secure, anxious-resistant, and avoidant. Mary Main later added disorganized as a fourth category. The four-category typology has become the most-cited framework in twentieth-century developmental psychology. It has also become a fixture of dating-advice books and therapy-Instagram, often in versions that bear little resemblance to the original research.

1. What the original research actually showed

Securely attached infants greet returning mothers with brief distress that resolves quickly. Anxious-resistant infants cling, can't be soothed, alternate between approach and protest. Avoidant infants pointedly ignore the returning mother despite physiological signs of stress. Disorganized infants show contradictory, freezing, or fearful behavior with no clear strategy (Ainsworth et al., 1978; Main & Solomon, 1990).

The patterns predict things. Securely attached infants, on follow-up, tend to show better social competence in preschool and adolescence. Disorganized attachment, particularly, predicts higher risk for psychopathology (Sroufe et al., 2005).

2. The leap to adult relationships

The popularization of attachment theory for adult romantic relationships came through Hazan and Shaver's 1987 paper, which surveyed adults about their relationship history and classified them into Ainsworth's three categories (no disorganized). The proportions roughly mirrored those in infant populations: about 56% secure, 25% avoidant, 19% anxious (Hazan & Shaver, 1987).

This is where the framework left its original moorings. Adult self-report of romantic-attachment style is correlated with childhood attachment but only modestly — perhaps r = 0.30 across studies. Adult attachment is related to but not determined by childhood patterns (Fraley, 2002).

The strong version of the popular framing — "your childhood attachment is your adult dating style" — is overreaching. The weaker version — "people show consistent patterns in close relationships that often have developmental roots" — is closer to the evidence.

3. The pop-psychology drift

The version of attachment theory now circulating in dating-advice spaces has additional layers the original literature doesn't support strongly:

Discrete categories vs. continuous dimensions. Most modern attachment researchers measure two dimensions — anxiety and avoidance — rather than four categories. The categories are useful heuristics; the dimensions fit the data better (Brennan, Clark, & Shaver, 1998).

Type stability. The popular framing treats attachment style as a fixed adult trait. Longitudinal research shows substantial change over time, especially after stable relationships of several years (Davila et al., 1999).

Predictive precision. Knowing someone's attachment style does not give you precise predictions about their behavior. It gives modest probabilistic guidance.

4. The honest summary

Attachment patterns are real, measurable, and clinically informative. The childhood-to-adulthood link exists but is more variable than popular accounts suggest. Adult attachment style is partially shaped by current relationship experience, not only by childhood.

The most useful framing for non-clinicians: "anxious" and "avoidant" describe tendencies under relational stress, not identities. Most people exhibit some mix, and the mix can shift over time. The framework helps name patterns that would otherwise be invisible — it should not be used as a diagnostic identity.