Abstract

Scientific models almost always contain idealizations, and this fact suggests methodological questions about how model building should proceed. Biologist Richard Levins addressed such questions by arguing that highly idealized models have a special role in helping to explain the behavior of populations. In When Less is More: Tradeoffs and Idealization in Model Building, I assess and partially endorse Levins' views first on their own terms and then through a novel analysis of idealization in modelling. This analysis begins with an articulation of theoretical desiderata including simplicity, precision, generality, representational capacity, and accuracy. After defining these desiderata, I evaluate them to determine any constraints that they put on the achievement of other desiderata. For example, I show that precision and generality "trade off" against one another. A full analysis of idealization requires coupling an examination of the constraints faced by modelers with an analysis of the scientific value of each desideratum. In this dissertation, I focus on the desideratum of generality and assess its importance for scientific explanation and some kinds of confirmation. I conclude When Less is More by discussing how an analysis of idealization, such as the one offered in this dissertation, can form the basis for determining a new set of methodological norms for model building