Abstract
There are certain fundamental aspects of visual experience that are extraordinarily difficult to account for in neurophysiological terms. The phenomenal world is composed of solid volumes, bounded by colored surfaces, embedded in a spatial void. Every point on every visible surface like a cube, or a sphere, is perceived at an explicit location in three-dimensions, and those surfaces are perceived as a spatial continuum in a volumetric space. The perception of multiple transparent surfaces, as well as the experience of empty space between the observer and a visible surface, reveal that multiple depth values can be perceived at any spatial location. It is hard to imagine how the activation of constellations of discrete neurons could possibly account for the continuous and volumetric aspects of phenomenal vision. I propose therefore a perceptual modeling approach, i.e. to model the observed properties of phenomenal space directly, as opposed to the neurophysiological mechanism responsible for that perceptual experience. The operational principle evident in the perceptual transformation is an expansion from the two-dimensional retinal stimulus (in the monocular case) to the three-dimensional structure of the spatial percept. In other words perception resolves the inverse-optics problem, to reverse the optical projection in the eye. A computational model is developed to demonstrate how a three-dimensional spatial percept can be computed from the two-dimensional retinal stimulus by a dynamic emergent process as suggested by the Gestalt soap bubble analogy. This model therefore offers a theoretical justification for the holistic emergent aspects of perception identified by Gestalt theory, demonstrated with a specific computational mechanism. http://cns-alumni.bu.edu/pub/slehar/www/webstuff/bubw1/bubw1.html