Distance perception across spatial discontinuities

Our previous studies in depth perception (Meng & Sedgwick, 2001) have supported a surface representation of 3-D layout. In such a representation, space is defined by surfaces, and distance is specified along those surfaces rather than through empty space (Alhazen, 1989, p. 155; Gibson, 1950, p. 6; Sinai, Ooi, & He, 1998; reviewed in Sedgwick, 2001). There is increasing evidence that visible surfaces play an important role in providing a framework for many other aspects of visual perception in a 3-D environment as well. Studies in stereopsis (Brookes & Stevens, 1989; Gillam & Sedgwick, 1996; Glennerster & McKee, 1999; Wilcox, Chodirker, & Bray, 2000), apparent motion (He & Nakayama, 1994), visual search (He & Nakayama, 1992), and visual attention (He & Nakayama, 1995) have all provided examples in which perception is influenced by or organized along visible surfaces (see review in Nakayama, He, & Shimojo, 1995). In order to use a surface as reference, on top of which a 3-D representation is built, an object has to be related to that surface. There are many possible relations an object might have with respect to a surface. For example, an object can touch a surface, hover above a surface, be suspended from a surface, or move along a surface. We have been particularly interested in studying the most direct and obvious relation between an object and a surface, which is the perceived contact. For example, a stationary object is seen as resting on the ground plane. For objects that are not directly resting on a common surface, we used the term nested contact relations (Meng & Sedgwick, 2001; Sedgwick, 1987, 1989) to describe a series of contact relations among multiple surfaces. In our previous experiments, observers were instructed to position a marker along a track on the ground to match the distance of a cube resting on a platform raised above the ground plane. Our results show that changing the perceived contact relation between the platform and the ground alone changes the perceived distance of the cube. Thus, the cube’s perceived contact relation with the platform mediates its perceived location relative to the ground. In those experiments, because the matching task was performed using a marker and track lying on the ground plane, the observers were forced to use the ground as the reference. In the experiments reported here, we explore situations where multiple reference surfaces are available. Our computer-simulated scenes show a cube resting on a platform on the left-hand side and a marker resting on a track on a platform on the right-hand side (Figures 1 and 2). The relative depth of the cube and the marker can be compared with respect to the ground plane, to the front surfaces of the two platforms, or to the top surfaces of the two platforms. Experiment 1 was designed to create situations where there is either a displacement between the top surfaces or the front surfaces of the two platforms. In theory, if the ground plane is always used as a global reference, performance should not be influenced by such displacements.