A systematic bias in the perceived location of a triangle’s occluded vertex.

Understanding how we perceive partially occluded objects has been a central focus of vision science. Here we report novel observations about partially occluded triangles. Specifically we document a systematic bias in the perceived location of a triangle’s occluded vertex. We report the results of four psychophysical experiments, in which participants explicitly specify where they believe an occluded triangle’s vertex is. The results demonstrate that across a variety of configurations there exists a systematic underestimation of the vertex’s would-be location, leading to the triangle to appear less elongated than it really is. This bias persists in the absence of an explicit occluder, a configuration that lacks t-junctions or other explicit cues to occlusion. Partially occluded triangle displays have been suggested to be particularly informative for probing models of amodal completion (Boselie & Wouterlood, 1992; Carrington, Palmer & Kellman, 2016), and our observations provide new constraints that such models need to account for.