Shake, Rattle, and Roll: Movement and Infant Cognition

Developmental research has long relied on measures of infant looking to yield insights into infant language acquisition.  Recent advances in software for coding movement and for deriving motion measures non-invasively from even low-quality video open up a rich new source of information about infant behavior and its relation to cognitive development:  their movement. We investigated the movement patterns of infants participating in speech perception studies employing both observer-coded and automatically derived approaches to measuring motion.  In the first study, we used Anvil, a software tool for annotating digital video on temporally ordered, hierarchical tiers, to examine systematic patterns in the movement of 6-month-old infants listening to linguistic and music stimuli.  In the second study, we automatically derived measures of movement magnitude using optical flow analysis of infant video to investigate the correlation of the movement, looking times and attention of 14-month-old infants in a word-object association task. Hand-coded data from the first study revealed that the 6-month-old infants showed differential movement patterns to language and music, including the possible underpinnings of adult-like head motion during vocalization.  Directional movements of head and torso also suggested that infants expected an interactional partner in the language, but not the music, condition.   In the second study, the motion patterns of 14-month-olds correlated with looking time results, and, in addition, captured nuances of their understanding of word-object association that looking time measures do not.  In particular, success in the task was mediated by infants’ mutual gaze with an Experimenter, and movement measures revealed that patterns of stillness associated with attention to the Experimenter were earmarks of success. The two studies we describe here showcase the successful application of two complementary, novel, and revealing methods for analyzing infant movement.  The use of measures of movement, whether observer-coded or automatically derived from video, allows fresh insights into the cognitive processing of infants involved in language learning tasks, enriching the dimensionality of our understanding of infant cognition.