Spatial scale of interactions between lines and aperture edges in the barberpole illusion

Purpose. In the barberpole illusion, there is a competitive interaction between terminator and contour signals for direction. Previously (ARVO 95) we found that a luminance cue for occlusion can render terminators extrinsic and lead to a perceived direction perpendicular to the lines - the 1D direction. We investigated whether the occluding contour - the aperture's edge - has to intersect the line ending to cause the terminator to act as extrinsic, or whether it can exert its effect at a distance. Methods. Observers viewed line gratings (duty cycle 0.25) that drifted within a rectangular aperture. Terminator strength was controlled by varying the contrast of the grating within the aperture (Cgr); the contrast between aperture and background (Cbg) was fixed at 1.0. The apertures were outlined with blank regions with luminance equal to that of the interior of the aperture. The width of the outlined region was varied to separate the line terminators more or less from the aperture walls. Results. Perceived direction had a monotonic dependence on contrast: it went from the 1D direction at low grating contrast Cgr to the 2D direction at high Cgr. The grating contrast Cgr at which the perceived direction switched from 1D to 2D directions was measured for different widths of the outlined region. 2D motion was seen even at low Cgr when the outlined width was greater than 10 min arc, just as in the case of zero aperture contrast. However, the transition from 1D to 2D motion occurred at higher and higher values of Cgr as the outlined width (separation between terminator and aperture wall) was decreased below 10 min. Conclusions. This result indicates that the t-junctions formed by the terminators and aperture walls need not be actual intersections. The mechanism that senses occlusion of the line ending is activated when the terminator gets within an interaction radius of the aperture wall, even when the separation is clearly visible.