Filling-in the gaps in models of completion

Abstract

Kogo and Wagemans' appropriately link completion phenomena to figure-ground computations. I argue that this link can be strengthened by considering the ecological conditions that give rise to completion phenomena. However, despite their polemics, the model that they offer can be viewed as an elaboration of the "borderline completion plus filling-in" model they eschew. Finally, I argue that it is unclear whether their model can: (1) Explain how surface structure and/or border ownership modulate the shape of interpolated contours; or (2) give meaningful outputs for images of natural scenes that contain a variety of different edge types.