Pictures, perception, and mental models

Abstract

The main point of this paper is to consider the way space is perceived in pictures and in “reality” and the question of whether mental models are a good means in explaining how space is visually perceived. Real or physical space is presumed to be (locally) Euclidean. Some kinds of pictures—e.g. pictures in perspective—are lawfully connected to the depicted scene so that the (Euclidean) geometry of that scene is preserved in these pictures. Following Johnson-Laird, visual perception is based on the construction of a partially analogical mental model. Therefore, as I will show, the geometry of a mental model representing the spatial layout of a scene in the physical world (or of a picture of such a scene) should also be Euclidean. However, at least since the famous experiments of Blumenfeld in 1913 it seems clear that our phenomenal or visual space is not Euclidean. How does this fit together? Can it be that the different cues which are involved in the perception of spatial arrangements are not modeled in a Euclidean way, but that the model in toto is (nearly) Euclidean? Is such a model built up by using “modified weak fusion”?