Abstract
In traditional figure-ground displays, an edge marks a border between two adjacent regions. One region is typically seen as a shaped entity and the other appears shapeless near the edge as if it continues behind the other region. Previous research has established that familiarity and convexity are among the configural cues determining which of two adjacent regions appears to be the shaped, near entity; 2-D displays were used in most of that research. Do configural cues influence perceived depth from binocular disparity? We used 3-D displays in which two adjacent, opaque, high-contrast regions densely covered in random dots were separated by an edge suggestive of a familiar shape (a face) on one side. Binocular disparity specified that the edge and the dots from one region lay at one viewing distance, while the dots from the other region lay at another. Either the face side was closer (consistent condition) or the non-face side was closer (inconsistent condition). Observers were instructed to attend only to the perceived separation in depth. Their task was to decide which of two successive displays contained greater separation in depth between the two surfaces (2AFC). We found that perceived depth between the two surfaces was greater in the consistent condition than in the inconsistent condition: PSEs were greater for inconsistent than for consistent comparison stimuli compared to consistent or inconsistent standards. Thus, perceived depth from binocular disparity is influenced by configural cues, despite instructions to attend only to relative depth. Implications for the manner in which configural cues and depth cues are combined will be discussed.
Thanks to Simon Watt, Martin Banks, and Dhanraj Vishwanath for their helpful suggestions and generous donation of time.