Consequences for perception psychology and epistemology

Abstract

Despite arguments to the contrary in the ongoing controversy between exponents of the Gibsonian and the computational representational theory of perception, the analysis in the preceding chapter of Gibson’s and Marr’s account of vision shows that the problems to which they give rise are in principle the same. Given the similarities of conditions which apply for the two accounts—rather than the differences most often focused on—this is not surprising. Thus, it applies to both Gibson’s theory and Marr’s computational model of vision that substantial assumptions and presuppositions are made about the “capabilities” and “knowledge” on the part either of the sensory system picking up information in the ambient array, or of our visual system to process representations from information in light images. And it applies to both Gibson’s theory and Marr’s model of vision that the descriptions of “high order variables in the optic array” or of “computed structures in representations from light images”, rest on an abstraction in the sense that they rely on ordinary everyday descriptions of things in the world and how we perceive them. In this respect it makes no difference in principle whether a technical description of the information extracted in patterns of light projected on to the retina is reached by an analysis of a Gibsonian kind, and expressed in terms of gradients, high order variables, and affordances, or whether the description of virtually the same information is the outcome of complicated computations of representation, expressed in some other technical language. It applies to Gibson’s as it does to Marr’s technical descriptions that they are ways of representing reality and the objects of which the experimental scenes are made up, which rely on ordinary everyday descriptions of these objects and scenes, i.e. rooms with walls and pictures on walls, or various shaped blocks put on a surface. If not, it would not be possible to determine whether, or to what extent, or for what purpose, their technical descriptions were adequate or relevant descriptions of—precisely—these scenes.