Abstract
A single light disk within an intersection in a grid of dark tetragons is especially difficult to detect (the “vanishing disk”). Detection is made even more difficult by curving the alleys defining the grid. In previous work (VSS 2008), we explored temporal aspects of the curvature effect. High contrast light disks were detected during the initial part of each trial, regardless of whether the alleys started straight and became curved or vice versa. Dark disks, on the other hand, had higher thresholds when synchronous with the curved alleys. This suggested that light disks and dark disks were affected by different mechanisms.
If the obscuring effects occur in primary cortex, orientation selectivity is expected. We therefore measured thresholds for pairs of parallel lines in intersections. Thresholds for light vertical or horizontal lines were significantly higher than for diagonal lines when the alleys were straight. With increasing curvature of the alleys, thresholds for diagonal lines increased while those for vertical or horizontal lines remained essentially unchanged; with sufficient curvature, the thresholds converged. This is consistent with components of the grid affecting targets with similarly oriented components.
When this experiment was repeated with dark lines (comparable to the dark disks that do not suffer the vanishing disk illusion), no orientation selectivity was found. Thresholds for all orientations were essentially identical, increasing with increasing curvature. As was found for disks, there is a difference between light and dark lines. This may be interpreted as different modes for the illusory effect of the vanishing disk versus the simple obscuring of disks by complexity. Alternatively, there could be a difference between the system detecting increments and that detecting decrements, or an interaction between the two when the polarities of target and grid differ. We will present preliminary results bearing on these possibilities.