A Framework for Comparing Agent Architectures
Aaron Sloman, Matthias J. Scheutz · 2002
Research on algorithms and representations once dominated AI. Recently the importance of investigation of architectures has been acknowledged. However different researchers use different objectives, presuppositions and conceptual frameworks, and this, together with the fact that many of the colloquial words and phrases referring to mental phenomena are ambiguous, leads to much argumentation at cross purposes, reinvention of wheels and fragmentation of the research community. This paper proposes a framework, the CogAff architecture schema, which should facilitate greater communication and integration by identifying dimensions for comparing architectures. This should also explain some of the existing conceptual confusion and diversity of views. Keywords: AI architectures, autonomous agents, cognitive modelling, philosophical foundations, software agents. 1 Introduction AI has always been concerned with algorithms and representations, but we also need to understand how t...