Using network analysis to capture developmental change: An illustration from infants' postural transitions

Network analysis is a tool typically used to assess interrelationships between social entities in a system. In this methodological report, we introduce how concepts from network analysis can be utilized to capture, condense, and extract complex developmental changes in individual behaviors over time. Using infant postural-locomotor development as an example, we demonstrate how network analysis principles can be applied to rich empirical data. We used existing free-play data from 13 infants followed longitudinally as they progressed from sitting to walking. We documented the range of postures adopted during play, how often infants transitioned between postures in their postural networks, and derived parameters of density and centrality. Analysis revealed that posture network density increased after infants learned to crawl and gained crawling experience as one might expect, but density did not further expand with gains in upright locomotion. Certain postures held different roles in the overall posture network displayed by an infant, and these centrality patterns depended on the time period involved. More central postures in the network were not always postures in which infants spent the most time. We discuss how network analysis might be utilized to better understand infant behaviors in other contexts (e.g., problem-solving, interventions, humanoid robotics).