Abstract

I develop a new theory of theoretical progress or 'truthlikeness'. Unlike previous theories, my approach focuses on the sets of models of scientific theories, rather than their linguistic formulations. Such an approach, I argue, avoids several long-standing problems in the philosophy of theoretical progress. I find in Chapter One that the most prominent schools of twentieth century philosophy of science have all failed to account for theoretical progress. I further argue that such an account is an essential element of the best argument for the doctrine of scientific realism. Chapter Two offers a taxonomy of recent theories of verisimilitude. A four-part classification is proposed according to where the theories stand on each of two basic meta-theoretical dichotomies: the structure of scientific theories, the measure of theoretical progress. My own approach is based on a semantic view of theories and a similarity-based measure of progress. I argue in Chapter Three that such an approach avoids the problems of language-dependence and incommensurability; it also explains the relation between truth and truthlikeness. In Chapter Four, I develop my own theory in detail. I propose that one scientific theory is closer to the truth than another just in case its family of models is more structurally similar to the actual world. Chapter Five is an historical illustration of my theory: the theory of atomic structure, 1867-1913.