Shape partitioning interacts with global shape integration

Objects are often identified by the shape of their profiles but complex objects are often comprised of multiple articulated components. It has been proposed that complex objects are decomposed and recognized by their component parts. This study exploits the proposition that the visual system decomposes objects at matched deep concavities on their boundaries. Rapid decreases in thresholds for detection of sinusoidal deformation of a circle's radius with number of cycles of modulation shows that shape information is integrated around radial frequency (RF) patterns. Here we merge RF patterns to form composite patterns with concavities and show that integration around the RF patterns is disrupted if the concavities are shallow but preserved if they are deep, consistent with their decomposition at matched deep concavities. Geon theory identifies complex patterns through a structural description of viewpoint invariant primitives known as geons. Geons are defined by properties on their boundaries that co-occur in a non-accidental manner across viewpoint changes rather than by reconciling metric properties such as curvature with viewpoint specific templates. Similarly, shapes of RF patterns are defined by the positions of curvature features on their boundaries. We argue that RF patterns provide flexible stimuli that might be used to study geons.