The Adult Attachment Interview
Historical and Current Perspectives
In 1985, the publication of a monograph entitled “Growing Points of Attachment Theory and Research” (Bretherton & Waters, 1985) marked a major turning point for the direction of the field. Here, Main, Kaplan, and Cassidy (1985) reported that an interview-based method of classifying a parent’s state of mind with respect to attachment was strongly associated with the infant’s behavior toward that parent during Ainsworth’s strange situation procedure conducted 5 years previously (Ainsworth, Blehar, Waters, & Wall, 1978). At the same time, infant attachment classification with the mother was found predictive of verbatim transcripts of children’s responses to Kaplan’s version of the Separation Anxiety Test at age 6 (see also Kaplan, 1987), and both mother–child and father–child discourse patterns were found sharply predicted by strange situation behavior towards the same parent in infancy (see also Strage & Main, 1985, and Main, 1995). Taken together, the above discoveries led these authors to appropriately subtitle this publication “A Move to the Level of Representation” in attachment research.
Until that time, research in attachment had focused almost exclusively upon nonverbal behavior as observed in or found to be correlated with the Ainsworth strange situation. This structured laboratory separation and reunion procedure yields three traditional categories of infant attachment with respect to a particular parent (secure, avoidant, and resistant or ambivalent; a fourth category, disorganized/disoriented, has subsequently been added). Studies of nonverbal behavior as related to these categories centered primarily upon (1) home observations of mother–infant interactions (see Belsky, Chapter 12, this volume) and (2) follow-up investigations examining corresponding differences in preschool and kindergarten behavior. In this latter context, children judged secure with their mothers during infancy were repeatedly found to enjoy more favorable outcomes (see Weinfield, Sroufe, Egeland, & Carlson, Chapter 4, this volume). It was not until the advent of the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1984, 1985, 1996), however, that representational processes as the likely mediator of differences in parental caregiving behavior were fully comprehended and made accessible to investigation.
The AAI protocol was developed in the early 1980s, as was an accompanying system for scoring and classification (Main & Goldwyn, 1984a, 1998a). Main and Goldwyn’s initial analysis showed that several continuous rating scales appearing to reflect a parent’s current state of mind with respect to his or her own attachment experiences were substantially related to aspects of the infant’s behavior toward that parent in the