Abstract

My dissertation contributes to the study of scientific theories and models by using a speech-act-theoretic framework to investigate the discursive aspect of theorizing and modeling practices. In the philosophical study of science some of the central questions concern the nature of theories and models and how they are used by scientists. There is debate about whether theories are best understood as representations or as tools useful for non-representational purposes, and among those that think they are representations there is disagreement about what kinds of representations they are. I argue here that a systematic investigation of the discursive aspect of theorizing and modeling practices can be useful to this debate: it offers a rich source of evidence which different accounts of theories and models can be evaluated against. Theorizing and modeling often involves the presentation of a series of sentences and equations. In these cases the uttered expressions are constitutive parts of a discourse within which theories and models are presented and used. I consider the influential work of Cartwright, French, Giere, van Fraassen and others, to show that different accounts of what theories and models are have implications for what function expressions can be expected to play in discourse. By observing how sentences and equations are actually used by scientists I argue we can determine which accounts best capture scientific practice. Here speech act theory provides a framework for systematically identifying features of discourse that indicate how expressions are used by scientists on a case-by-case basis. I also defend the view that theories and models are often linguistic representations in which sentences and equations are used by scientists to say things directly about the systems they study. This runs contrary both to instrumentalist views, which take theories to be non-representational, and to the semantic view, which takes theories and models to be non-linguistic. I defend my view by showing the discursive data drawn from a canonical example, the London model of superconductivity, unambiguously favours my account. On the basis of this evidence and additional considerations I conclude that many other theories and models are likely linguistic representations too.