Theoria: An International Journal for Theory, History and Foundations of Science

Volume 5, Issue 1/2, Noviembre 1990

Cognitive Science and Semantic Representations

The main task of Cognitive Science is to construct concepts and models that would be superordinate to knowledge in the various particular cognitive sciences. In particular, one major objective is to formulate a hypergeneral description of representations that could encompass all descriptions given in subordinate domains. A first basic distinction is between natural and rational representations, i.e. given mental entities and representations that are governed by prescriptive rules coming from logical or scientific thought. In addition, representations must be described in respect to several sources of variability, which are tentatively listed here. Description of natural representations is based on a distinction between taken representations, which are mental events, and type representations, which are lasting structures registered in memory. The connection between them can be modelled through the concept of activation. One advantage of activation models is their large compatibilty, not only with experimental evidence in cognitive psychology, but also with facts and hypotheses in neurosciences, and programming modes or requirements in artificial intelligence. Comprehension of natural language is a highly representative domain in this respect, which exemplifies the power of these concepts.