Biological and theoretical relevance of some connectionist assumptions: The development of conceptual networks

For the study of psychological processes in cognitive science modelling two general approaches rule nowadays research: Artificial Intelligence (top-down) functional symbolic models, and Connectionist (bottom-up) neural networks modelling. Our goal in this paper is to show that analyzing the theoretical level of description and explanation of this models an important theoretical gap between both is found. Connectionist modelling through neural networks face at present several theoretical problems that have to be accounted in order to build realistic and feasible models of the brain. The lack of biological constraints, network stability or serial behaviour, for example, are relevant issues to bear in mind. Our proposed model, conceptual networks, can be located halfway between the traditional semantic networks and artificial neural networks modelling, and is presented as an attempt of building a necessary bridge between this levels of description, focussing on the correspondence between the functional level and the level that can be modelled by artificial neural networks. The elements and functioning of conceptual networks are based in biological and psychological constraints necessary to build realistic models of actual cognitive processes in brain functioning