Abstract
In the Poggendorff illusion an oblique line is interrupted by a pair of parallel lines or a rectangle, resulting in an apparent misalignment of the visible ends of the oblique line. From a century and a half of experiments, it is clear that multiple factors may contribute to the phenomenon. Here we examine the role of the junctions and the interpretation of occlusion in the Poggendorff effect. In the first experiment, observers viewed a vertical rectangle (1.75 x 8 deg) and oblique line (acute angle: 30 deg) through LCD shutter glasses (frame rate: 120 Hz). On different trials, the oblique line was tilted left or right and at -3, 0, or 3 arc min of disparity with respect to the rectangle which was in the plane of the display. Observers reported whether the oblique line ends appeared aligned or not on each trial. There are two main results: i. observers were most likely to report misalignment when the oblique was in the plane of the object; ii. there was a near/far asymmetry – oblique in front of the rectangle tends to be seen as aligned, but not when behind the plane. We conclude that an important aspect of the illusion is whether the junction is interpreted as a join or not. In a second experiment we created Poggendorff variants involving intersecting rectangles and triangles, or long ellipses intersecting long ellipses. Shadows were added to one or another part of the figure to imply a depth order of the two objects. In this experiment, interrupted contours appear aligned if they are implied to be in front, but not if behind. We conclude that contour extrapolation processes work correctly unless vetoed by evidence for a join between objects. A model explaining the mechanism of junction bias will be described.
Meeting abstract presented at VSS 2017