Abstract

A recent philosophic trend has recognized that empirical findings in brain research may have important implications concerning the limitations and structure of human knowledge. In keeping with this naturalization of epistemology, this paper focuses on limitations on possible visual experience implied by contemporary vision research. Discussion centers on an example from depth perception, where it can be shown mathematically that the presence of only two disparate monocular views of the world fails to provide sufficient information for the visual reconstruction of the third dimension: a vast number of possible arrangements of objects are generally compatible with the available retinal information. Current work, such as that of Marr and Poggio, suggests that the visual system reconstructs the third dimension by imposing a constraint of the form: objects are bounded by opaque surfaces that vary smoothly and minimally in depth from point to point. It will be argued that such constraints are both a priori and synthetic and that they are closely related to Kantian a priori synthetic propositions.

© 1988 Optical Society of America