Representação no conexionismo: uma investigação filosófica

Abstract

The aim is to philosophically investigate what the representations of connectionist networks are about, and how they can shed light on questions about how the mind fits into a purely natural universe. Specifically, we will investigate the distributed connectionist representation, which is the representation in connectionist systems based on learning, distributed along the artificial neurons of the network. We will then discuss how these representations relate to the other explanations provided by classical computationalism. We will present two positions, ontological conservatism, which defends a possible compatibility between the foundations of representation provided by classical computationalism and the concept of representation of distributed connectionism. And ontological radicalism, which argues that the possibility of representation in distributed connectionist systems means the elimination of the foundations previously understood in classical computationalism. Finally, we will present an argument in defense of the position of ontological conservatism.