Sara Schroer¹ (), Nathaniel Powell¹, Daniel Panfili¹, Mary Hayhoe¹; ¹University of Texas at Austin

Although much is known about visual and motor development, little work has been done on how vision guides actions in young children. Recent work in adults uses novel technology that integrates gaze, body, and terrain data (Muller et al., 2023). Similar integrated systems are now possible with children. To study the development of visually guided locomotion, children (1-to-6-years-old) wear head-mounted eye trackers (Pupil Labs Neon) while walking on various terrains, including a sidewalk and loose pebbles, while simultaneously recording with a fixed ZED 3d-scene camera to estimate terrain structure and a skeleton of the walker in the terrain. Work in adults shows that walkers are sensitive to terrain, modifying gaze distribution, step speed, and step length (Matthis, 2018). Similarly, data from an adult, 2-, 3-, and 4-year-old found that where children looked was sensitive to terrain, though only the 3-year-old changed their speed. The 4-year-old and adult in the two terrains were similar in gaze distribution and speed to previous adult data in flat terrain (1.38±0.05m/s; Matthis, 2018), looking up while walking on the sidewalk (adult 61% of the time, 4-year-old 56%) and far ahead on the ground on the pebbles (adult 78%, 4-year-old 84%). The 3-year-old looked far ahead on the ground while on the sidewalk (61%) but slowed down and looked down near their body on the pebbles (66%; 0.96m/s to 0.78m/s). The two-year-old moved slowly on both terrains (0.36m/s). Although 2-year-old’s gaze was distributed widely in the sidewalk condition, they mostly looked down near their body on the pebbles (54%). Additionally, the 2-year-old and 3-year-old showed awareness of the difficulty of the pebbles (e.g., asking for a hand). While these results are preliminary, they suggest that locomotion in young children, like adults, is controlled by complex decision mechanisms that take both costs and sensorimotor uncertainty into account.

Acknowledgements: NIH Grant EY05729