Abstract
A number of fascinating inhibitory effects of inducer grouping on illusory-contour formation have recently been described by Gillam (1987) and Gillam & Chan (2002). They show that illusory contours are stronger when inducers and potential illusory contours have a ‘random’ relative spatial arrangement, than when they have a more regular arrangement. They suggest that their results are consistent with the ecology of biological vision: An occluder is usually physically independent of the background objects that it partly occludes, and thus the geometry of their contours are likely to be independent. Similarly, contours of individual background objects are likely to be independent. Thus, image regularities among these contours could be taken as evidence against the presence of a camouflaged occluder (the potential illusory figure). However, I show that strong illusory contours can occur in patterns possessing a high degree of regularity among the inducers and the potential illusory contours. I suggest that interactions between nearby, approximately parallel visual contours might help to explain many of the effects described by Gillam and colleagues. Nearby, approximately parallel visual contours tend to group with one another, and also tend to mutually inhibit one another (also see Li, 2001). I propose that these processes affect illusory-contour formation in two ways: 1) When the far ends of line-inducers (the ends not adjacent to the target illusory contour) terminate along a (virtual) contour that is roughly parallel to the target illusory contour, then the target illusory contour is weakened. 2) When inducing lines are roughly parallel to one another they mutually weaken their neural representations, resulting in reduced “end-cutting” responses to the inducing line-endings, and consequently, weaker illusory contours.