Vision as a user interface

Abstract

The egg-rolling behavior of the graylag goose is an often quoted example of a fixed-action pattern. The bird will even attempt to roll a brick back to its nest! Despite excellent visual acuity it apparently “takes a brick for an egg.” Evolution optimizes utility, not veridicality. Yet textbooks take it for a fact that human vision evolved so as to approach veridical perception. How do humans manage to dodge the laws of evolution? I will show that they don’t, but that human vision is an idiosyncratic user interface. By way of an example I consider the case of pictorial perception. Gleaning information from still images is an important human ability and is likely to remain so for the foreseeable future. I will discuss a number of instances of extreme non-veridicality and huge inter-observer variability. Despite their importance in applications (information dissemination, personnel selection, . . . ) such huge effects have remained undocumented in the literature, although they can be traced to artistic conventions. The reason appears to be that conventional psychophysics—by design—fails to address the qualitative, that is the meaningful, aspects of visual awareness whereas this is the very target of the visual arts.