Complementarity in vision and cognition

Abstract

In information theory there is a fundamental principle, usually referred to as the informational “uncertainty principle”, which expresses a limitation of any information processing system (or agent) in terms of a relation between the system's response property and its inherent processing capacity. From this principle, it can be argued that a salutary strategy for dealing with conflicting information processing requirements is to adopt various complementary processes (or channels). Donald M. MacKay had attempted to relate the informational uncertainty principle to spatial and temporal response properties of neurons in the mammalian visual cortex, and suggested that the spatial and the temporal aspects of such neurons are complementary. I attempt to extend his efforts and to show that the informational uncertainty principle may indeed underlie many complementary relations exhibited in human perception and cognition, such as the relation between the two principal processing streams in vision and the relation between parallel and serial processes in cognition.