The Perceiving Brain

Abstract

According to the standard model of perception, the brain is a Turing Machine; its software matches `mental representations' of the outside world with its reflections on the retina. According to more recent computational models, the brain computes retinal images; `mental representations' are superfluous. According to the new biological theory of perception (Calvin, 1989; Edelman, 1989, 1992), the brain needs neither `mental representations' nor retinal images; it is a different digital computer, or Darwin Machine, which unconsciously perceives objects as they are -three-dimensional, colored and moving-because during the process of ontogenesis neuronal populations are selected by the environment to respond to the lines of different orientation, movement and color, and thereafter continue to do so habitually. The destructive part of this paper traces theoretical foundations of computational models of perception and demonstrates why they cannot explain human perception. The constructive part sketches a theory of human perception which utilizes the ontogenetic selection to explain recognition as the multimodal perception and only form of consciousness common to the mammalian brain. Contrary to common opinion, recognition is fully accounted for by the brain's analogues of gestalts; what needs to be explained is how the human brain shifts to unimodal perception and then to perceiving a single aspect/feature.