Modulation of multi-level evolutionary strategies for artificial cognition

There are several theories of cognition, each taking a different position on the nature of cognition, what a cognitive system should do, and how a cognitive system should be analyzed and synthesized. From these, it is possible to discern three broad classes: the cognitivist approach based on symboli...

Descripción completa

Detalles Bibliográficos
Autor Principal: Romero López, Oscar Javier; Antonio Jiménez, Angélica de
Formato: Artículo (Article)
Lenguaje:Desconocido (Unknown)
Publicado: 2009
Materias:
Acceso en línea:http://babel.banrepcultural.org/cdm/ref/collection/p17054coll23/id/744
Descripción
Sumario:There are several theories of cognition, each taking a different position on the nature of cognition, what a cognitive system should do, and how a cognitive system should be analyzed and synthesized. From these, it is possible to discern three broad classes: the cognitivist approach based on symbolic information processing representational systems; the emergent systems approach embracing connectionist systems, dynamical systems, and enactive systems, all based on a lesser or greater extent of principles of self-organization, and the hybrid approach which combine the best of the emergent systems and cognitivist systems. Our research focuses on implementing a hybrid architecture for cognitive agents supported by both cognitivist and emergent approaches. On the one hand, the cognitivist approach provides an explicit knowledge representation through the use of symbolic AI techniques. On the other hand, the emergent approach defines three evolutionary strategies as observed in nature: Epigenesis, Ontogenesis, and Phylogenesis, endowing the architecture with implicit knowledge learning, sub-symbolic representations, and emergent behavior guided by bio-inspired computational intelligence techniques.