Abstract
Substantial research over the last 30 years of vision science has illuminated how we understand a dynamic world. Studies of ensemble processing tell us what kinds of motion, like linear translation and global optic flow, can be easily used by human vision perceive coherent object trajectories and estimate time-to-contact (Whitney & Leib, 2018; Van den Berg, 1992). Studies of multiple object tracking tell us how we use information to know that a given object is the same one over space and time (Flombaum, 2009; Scimeca & Franconeri, 2014). Every one of these studies has been inspired by the dynamic natural world that the visual system evolved in. But we also spend several hours daily immersed in artificial worlds that use moving displays for information. Dynamic data visualizations, map transformations, and diagrams are used for data analysis, navigation, and education. These dynamic displays are comprised of multiple dynamic objects that, across domains, demand many of the same visual processing capacities we have long studied in the lab. Empirically determining when these dynamic displays are (and aren’t) effective allows us a fresh scenario for understanding how we process dynamic scenes, while at the same time allowing research outputs to inspire interventions to help create more effective displays for students, scientists, and the general public. We present a theory-driven analysis of these displays and the visual processing demands they invoke. We show that visual processing capacities in perception and attention (ensemble processing, tracking object individuals and groups) are what help us see patterns in artificial displays, and that we fail to see patterns when processing capacity is exceeded. Visualizations that defy categorization within this framework can uncover new basic research directions. This domain presents use-inspired basic research by vision scientists to test visual thinking in the constructed, dynamic displays of everyday visual experience.
Acknowledgement: NSF DRL 1661264