LEARNING TO ATTEND: 2D AND NATURALISTIC ENVIRONMENTS REQUIRE DIFFERENT LEARNING SYSTEMS

All Authors:
Dima Amso, Brown University, Providence, RI, United States (Primary Presenter)
Thomas Serre, Brown University, Providence, United States

Human adults have demonstrated sensitivity to visual context cues, exploiting implicitly learned memories of the structure of the visual environment to facilitate both spatial navigation and visual search. Studies of contextual cueing in childhood have produced mixed results. These discrepancies may derive from issues with construct validity, the use of screen-based tasks to study an inherently embodied process that depend not only on simple memory and attention, but complex interactions of these systems with bodily frames of reference. The allocentric frame involves information about objects in the environment relative to each other (2D), while the egocentric frame includes information about the viewer’s location relative to objects in the environment.  In naturalistic environments, these frames of reference are coupled in the service of spatial abilities. Here we examined developmental change in naturalistic and 2D contextual cueing in the same N=39 4-9 year-old children in two parallel tasks, one in our naturalistic SmartPlayroom space and the other using snapshots of the space in a 2D computerized task. The computational power of the SmartPlayroom allowed us to examine behaviors associated with co-occurrence learning, incidental encoding, relational encoding, and estimation of similarity in path and eye movement trajectories. Our data showed generally successful visual search and spatial attention in both 2D and naturalistic environments, but that incidentally fixating objects in the SmartPlayroom, but not the 2D task, resulted in implicit encoding of object location into memory in a way that resulted in more efficient search when that object later became the target.