Paradigm Shifts in the Neurobiology of Perception

Abstract

In cognitive science, we are currently witness to a fundamental paradigm shift [for review see, e. g., Varela et al. (1991) and Bechtel and Abrahamsen (1991)]. In the early days of artificial intelligence and cognitive psychology, cognitive processes were assumed to be based on algorithmic computations controlled by formalizable rules which act upon quasi-propositional knowledge about the external world. In recent years, however, more and more cognitive scientists have converted to the idea that functions like perception, problemsolving, or memory emerge from complex interactions in highly distributed neuronal networks which, unlike conventional information-processing systems, are shaped by learning and experience-dependent plasticity. In such networks, information processing does not follow explicit rules, but is based on the self-organization of patterns of activity. This paradigm shift is motivated by the apparent inability of classical models to account for many facets of cognitive processes and, in addition, by their lack of biological plausibility. In the present contribution, we describe this paradigm shift as it occurs in the neurobiology of perception, with particular reference to the visual system. In addition to demonstrating crucial differences, we elaborate basic conceptual assumptions which are common to both the classical neurobiological view on perception and to the more recent connectionist framework.