Using Computational Analysis of Behavior to Discover Memory-Guided Attention Mechanisms During Naturalistic Visual Seach in Childhood

• In Event: What's that? How Visual Distraction May Help, or Hinder, Infants' and Children's Memory

Abstract

Introduction. A fundamental survival skill involves learning about space in a way that supports efficient attention and action in the real world. Studies of this skill, called memory-guided attention, offer conflicting evidence about its developmental course. Efficiency in memory-guided attention reduces the need to repeatedly expend expensive learning resources in previously visited spaces. At the same time, visual search anew can be more efficient than engaging memory and visual attention processes at once. The most adaptive strategy seems to depend on task dynamics. Naturalistic scenes and energy-costly task demands are more likely to engage memory during visual search in previously visited spaces in adults. As such, understanding the development of memory-guided attention may best be done in real world contexts. Yet, methodological restrictions have limited a fully embodied understanding of this skill in developing humans.

Methods. We built a SmartPlayroom, outfitted with an array of video and depth sensor technologies, and combined with computer vision algorithms for automatically characterizing oculomotor and locomotor behavior. We tested 4-9.5-year-old children on a naturalistic memory-guided attention visual search task. We measured fixation distribution during a visual search task using wearable eye tracking, and simultaneously recorded depth video data for each participant and used computer vision algorithms to track them during navigation. We manipulated object placement and trial order such that nearby objects would be encountered during initial search for reference objects. We used a computational model of top-down attention guidance based on reference object visual features and examined the use of this top-down attention guidance for reference objects during subsequent nearby object search.

Results and Conclusions. The data suggest that there is value for head and body movements during initial spatial learning for subsequent memory-guided attention, specifically in early childhood. Physical navigation during initial object search in the room was associated with stronger visual representations of goal reference objects as measured by top-down attention guidance models, which translated in better memory-guided attention for nearby objects. By middle childhood, it may be more adaptive to search de novo on each trial than to engage an additional memory process even in previously visited spaces. Overall, our study highlights the benefits of naturalistic experimental paradigms combined with modern computer methods for understanding mechanisms underlying the key operation of memory-guided attention.

Authors

• Dima Amso, Columbia University
  Presenting Author

• Lakshmi Govindarajan, Brown University
  Non-Presenting Author

• Thomas Serre, Brown University
  Non-Presenting Author