Abstract
Many physical objects, particularly biological ones, have shapes that change due to natural movements, such as articulations. Such motion, while non-rigid, is generally constrained by underlying physical structure (e.g. a skeleton). One such constraint is the fact that contour regions that move independently of the rest of the object, such as fingers, are typically convex, while concave regions that form the boundary between convex parts (e.g. the space between fingers) do not move independently but only in conjunction with a larger—primarily convex—section of the object (e.g. the spaces between fingers moving along with the whole hand). The current study tested whether the visual system employs this regularity when assigning figure and ground in animated displays. Subjects who had to make a figural assignment to a dynamically deforming contour bounding two colored regions showed a bias for assigning figure so that the moving contour was convex rather than concave. This result held even when the dynamic cue was inconsistent with established static cues to figural assignment. This study extends earlier research showing a bias for biologically valid dynamic deformations (i.e. joint articulations; Barenholtz and Feldman, 2006) and provides evidence for a more general dynamic cue to figure-ground assignment that is based on the geometric properties of the motion of physical objects.
This research was supported by NGA Award #HM1582-04-C-0051.