Abstract
Scenes are visual structures that are inherently arranged in a 3 dimensional space, but little is known about how the volume and depth information of complex 3d scenes is processed at the beginning of the visual processing. This study investigated how duration and masking procedures influence the perception of volume information of 3-dimensional scenes. Eighty participants were shown high-resolution stereoscopic images of scenes, for various durations (from 50 msec to 800 msec). The images were either presented with or without a 3d mask (100 msec). Stimuli were complex computer-generated scene images (e.g. kitchen, living-room, garden, patio) with a variety of spatial layout arrangements. Two sets of volumetric scenes were created: a correct 3d image (as in a normal stereoscopic viewing condition) and a flat 3d image (generated by displacing one image a constant number of pixels, producing an image out of the plane of the monitor). Both sets of images had the same disparity, on average over the whole image. The task was to discriminate between true and false 3d images. The results showed how performances of discriminability evolve with increasing presentation time (from d′∼1 at 50 msec to d′∼2 at 800 msec). The presence of a mask impaired 3d discrimination only at short duration (50 and 100 msec). In addition, a stimulus analysis identified the scene spatial layout most suitable to elicit true dimensional perception, and a low false alarm rate.
Acknowledgement: C.L. is supported by the National Funds For Scientific Research (Belgium). A.O. was awarded a NIMH grant (1R03MH068322-1).