Sylvia Culp and Philip Kitcher
Traditional approaches to theory structure and theory change in science do not farewell when confronted with the practice of certain fields of science. We offer an account of contemporary practice in molecular biology designed to address two questions: Is theory change in this area of science gradual or saltatory? What is the relation between molecular biology and the fields of traditional biology? Our main focus is a recent episode in molecular biology, the discovery of enzymatic RNA. We argue that our reconstruction of this episode shows that traditional approaches to theory structure and theory change need considerable refinement if they are to be defended as generally applicable.
2. Practices in Contemporary Molecular Biology
3. The Discovery of Enzymatic RNA
4. Understanding the Change
5. Derivative Revolutions and Reductionism Revisited
1. Despite a plethora of critiques of what used to be known, quite aptly, as the 'received view' of scientific theories, contemporary debates about inter-theoretic relations and about theory change are still often conducted as if scientific theories could be unproblematically identified as deductively closed sets of sentences. In our judgment, the attempt to find some small number of general laws that can serve as the axioms of the scientific theory under study has proved especially baneful in the biological sciences. Even where axiomatization has been achieved-as in the careful studies of Mary Williams on the theory of natural selection [1970, 1973]-there are serious questions as to whether the axiomatization captures what is central to the theory as biologists actually understand it, use it, and develop it. Indeed, we believe that the continued smouldering problems about the 'triviality' or 'unfalsifiability' of neo-Darwinism are, in large measure, the product of thinking that there has to be some principle(or principles)of great generality that formulate the core of the theory of evolution by natural selection (see Kitcher , pp. 55-60). Similarly,in the debates about the reduction of classical genetics to molecular biology, the macro-level theory has been essentially divorced from what geneticists have been doing since approximately 1915 (let alone 1953) by proposing that classical genetics can be identified with the logical consequences of Mendel's laws (or minor adjustments of those laws). If any scientific theory worthy of the name is a deductively closed set of sentences whose axioms include generalizations that range over all the entities in the domain of the theory, then it seems that either biology has no theories or what theories it has are both trivial and irrelevant to the thinking and activity of almost all contemporary biologists.
Nor are things much better if we abandon the syntactic conception of theories in favor of its chief rival, the so-called 'semantic view of theories', according to which a theory is a class of models. Although this approach is valuable in differentiatinga theory from its formulations (Suppe ), in clarifying the relation between theory and observation (Van Fraassen), and in reconstructing parts of important biological theories (Lloyd , Lloyd [forthcoming]), we believe that it encounters problems that are parallel to those besetting the older conception when it seeks to treat theories at a high level of generality. The task of specifying the class of models that is Darwinian evolutionary theory or the class of models that is classical genetics gives us no more adequate a picture of these biological disciplines than does the task of specifying the axioms of these theories-and for the obvious reason that the new-style specifications of sets of models look remarkably like the old-style axioms (see Beatty  and Giere  for examples).Proponents of the semantic view have