Abstract:
The theory of conceptual spaces is a frequently applied framework for representing concepts. One of its central aims is to find criteria of what makes a concept natural, that is, easy to process and cognitively useful. A prominent proposal is the demand that natural concepts cover convex regions in conceptual spaces. The first aim of this talk is to analyse the convexity assumption and the arguments that have been advanced in its favour. Based on this, I argue that most supporting arguments focus on single-domain concepts (e.g., colours, smells, shapes). Unfortunately, these concepts are not the primary examples of natural concepts. In particular, unlike so-called natural kind concepts (such as the concept of apple), they cannot capture natural correlations between domains. Building on this observation, the second aim of the talk is to propose criteria of naturalness for multi-domain concepts. The representation of such concepts has two main aspects: features that are associated with the concept (e.g., the taste of the apple) and the probabilistic correlation pattern which the concept captures. Together with probabilistic considerations, conceptual spaces provide a helpful framework to study the naturalness of these concepts. With respect to feature representation, the existence of characteristic features (i.e., the fact that apples have a specific taste) is essential for natural multi-domain concepts. Moreover, these concepts capture peaks of a probability distribution over complex spaces and, in this sense, carve up nature at its joints, that is, at areas with no or low probabilistic density.
Speaker:
Corina Strößner is a postdoctoral researcher at the Ruhr University Bochum. She received a PhD from the Saarland University with a thesis on normality. Her postdoctoral research is focused on the interdisciplinary study of concepts and reasoning with concepts in which she uses formal methods as well as experimental approaches.