Behavior and the Development of Attachment: Innovating Assessment and Theory Building

Important breakthroughs in understanding the role of attachment for social functioning and wellbeing have always gone hand in hand with the construction of new measurement paradigms. In her 1986 book on five methods of assessment, Mary Main concluded that experiences and representations of attachment shape general, stable, and coherent categories of mental organization. The rules for classifying these categories were derived from data drawn from new, experimentally controlled measures. Taking inspiration from Main’s approach, the current paper argues that a data-driven approach that leverages advances in naturalistic recording and computation may go further and also deliver a detailed understanding of how and when patterns of attachment behavior are formed during the first year of life. This paper describes how tools for long-term, fine-grained, multimodal recording can be embedded in daily family life and combined with high-throughput processing and supervised machine learning. Predictive models of dynamic interactions between features of infants’ and caregivers’ behaviors can be explored and interpreted to arrive at testable theory. By casting theory in the form of agent-based models, we can iteratively rebuild Bowlby’s original conceptualization of behavioral control systems for attachment and caregiving, which provided the starting point of Main’s empirical forays. We describe the theoretical gaps in current attachment research as well as the opportunities for multidisciplinary research that could be addressed by an approach that builds on Main’s abductive logic.