Abstract
Gaze patterns differ by task, but how do task-relevant gaze patterns develop? We compared patterns of eye and head motion produced while mothers carried their infants down an indoor hallway. Methods. We analyzed data from (N=6) mother/infant dyads who both wore head-mounted eye trackers that provided a scene video and gaze position at 30 Hz. Eye position data series were converted into rotational pitch and yaw velocities. Head movement was calculated by extracting optic flow from recorded videos and self-motion from flow. We limited the analysis to frames where the focus of expansion (FOE) or the center of rotation (COR) were within two degrees of visual angle estimated from two methods (Heeger & Jepson, IJCV 7, 1992; Raudies & Neumann, Proceedigns DAGM, 2009). Circular distributions of the direction of FOE, COR, and center of fixation (COF) relative to the origin of the scene camera were calculated. Results. In infants, the FOE pointed upward, consistent with a downward-pointing head position. Mothers’ FOEs pointed to the left and right. Babies CORs were elongated along the horizontal axis; that of mothers was elongated along the vertical axis. Mothers shifted gaze left/right more often than up/down. Correlations between flow of the scene videos from babies and their mothers were higher (~70%) than those for the eye-velocities and laminar flow (~55%), and were higher than those for eye-velocities of mothers and their babies (~40%). For mothers and babies the distance between FOE and COF spreads over a large range. Conclusions. Mothers explore the scene along the horizontal axis more so than infants. Passively carried infants experience pitch rotation and generally direct their head toward the ground.
Meeting abstract presented at VSS 2012