Contour interpolation utilizes contour relations to decide which visible fragments belong to the same object. In so doing, interpolation produces a phenomenological presence of the contours and surfaces that fail to project to the retina. This is called filling-in. A functional consequence is that the non-visible connecting regions are treated as if they too were visible. The goal of the dissertation is to explore how, why, and under what circumstances filling-in regions affect interpolated shape.

In the first chapter, my co-authors and I aimed to see whether filling-in regions affect the perception of interpolated shape when fragments were separated in both space and time. We had participants repeatedly discriminate "fat" and "thin" real, illusory or fragmented figures, where different inducing edges appeared at different times. The figures were presented in luminance noise and correlations were calculated between each pixel of noise and observer response. Resulting classification images (CIs), showed that noise pixels near interpolated contours influenced observer response similarly to when real contours appeared, and not at all when such contours were removed. Because our CIs resembled those generated with static displays, we suggested that spatial interpolation, which spans only gaps in space, may be a special case of a more general spatiotemporal interpolation process.

In the next chapter, we explain how and why filling-in regions affect shape perception by explaining how and why CI features emerge in the fat/thin task. Our data reveal that fat/thin CI features strongly depend on lightness induction. In other words, the kind of shape that we perceive depends on the lightness of that shape's surface. Our explanation also offers a novel interpretation of the CIs that emerge during shape discrimination.

In the last chapter, 1 examined whether cognitive strategy could alter filling-in. Participants discriminated fat and thin shapes, the visible portions of which could either form or not form illusory contours. Participants were instructed to treat the visible fragments as distinct entities, or as visible parts of a larger camouflaged object. Distractor lines occasionally appeared near the unviewed boundaries of the shape to potentially disrupt the filling-in process. It was found that—regardless of observer strategy—distractor lines hurt performance only when placed near illusory contours, indicating that filling-in is not readily altered by top-down influence.