Abstract
Perceived occlusion is here taken to be indicated by an illusory contour at an alignment. Most research on occlusion and shape has concerned the nature of amodal completion behind an occluder (are completions local, global or fuzzy?) Our question is not how or if shapes are completed but how shape determines whether an occluder is seen at all. In the first set of experiments there were different sets of 5 black identical shapes aligned at one edge. Width of the configuration at the aligned edge and the support ration were constant. Method. Subjective contour strength was measured by the method of paired comparison allowing a z score scale of contour strength to be calculated from choice probabilities. Results. Subjective contour strength was much greater for quadrilateral shapes which if continued beyond the edge could be symmetric than for those whose explicit visible shape was symmetric. It was much greater for shapes that could be rectangles if continued beyond the edge than for those explicitly rectangular. It was also found that a set of 5 half circles produced a significantly weaker subjective contour than either quarter or three quarter circles with the same support ratio. Implications for extant theories of occlusion perception will be discussed. In a second set of experiments we explored the role of familiarity and meaning on the strength of subjective contours along truncated letters using only those where truncation did not violate closure. Using paired comparisons we showed that truncation produced a strong subjective contour not only for truncated letters forming a phrase but also for reversed/jumbled letters and carefully controlled non-letters. The critical factor was found to be the presence of a junction of some kind in the centre (as for most letters) which truncation moved to a lower position. Thus letter recognition was not implicated.