Abstract
Most studies of figure/ground have used methods that presume a single global figural assignment, such as asking subjects which entire region appears in front. In our study, we used the motion-probe method introduced in Kim and Feldman (2009) designed to assess figure/ground locally at arbitrary points along a boundary, seeking evidence of local propagation of border-ownership (figure/ground assignment) along the boundary. In the motion-probe method, a small spatially circumscribed motion signal is created at a point on the boundary between two coloured regions, and the subject is asked which colour appeared to move; because the figural region “owns” the boundary, the response reflects border-ownership. In our study, subjects were shown semicircular shapes, to which a bar was added in such a way that in some configurations the T-junctions induced a clear local change in figure/ground assignment (example display at http://ruccs.rutgers.edu/~jacob/Demos/figure_ground.html). We then assessed figure/ground at various other points along the border, ranging from relatively near the inducing bar to relatively far, giving us the opportunity to capture the propagation of the figural status induced by the junction cue. We found a systematic effect of probe position, with probes closer to the inducer showing an increasingly strong tendency to receive figure/ground assignment consistent with the inducer—that is, as if the figural status propagated spatially from the point of the inducer. A computational model of the propagation mechanism based on Bayesian belief networks suggests intriguing parallels to known properties of neural coding of border ownership in visual cortex.