Topics in scientific realism

Abstract

In this thesis scientific realism is presented as a view comprising different kinds of realism: metaphysical realism; axiological realism; epistemological realism and semantic realism. There are different arguments to the effect that there is no single conception of metaphysical realism. That of global response-dependence is singled out. It is argued that realists ought not to be worried about global sceptical arguments of this kind. Two formulations of a localised anti-realist argument, the pessimistic meta-induction argument is presented. It is argued that a formulation of entity realism, like that proposed by Devitt (1997) is not adequate for defending realism against this argument. Psillos’ (1999) defence of realism regarding entities against the pessimistic meta-induction argument is then examined and it is argued that his defence fails. Worrall’s account of structuralism is then examined. It is argued that Worrall’s account still faces problems posed by Psillos (1999). These problems can be reformulated to accommodate Worrall’s most recent formulation of structuralism. Ladyman’s ontic structuralist position is then considered and it is argued that no coherent account of structures is provided on this view. It is then suggested that we may develop a version of semirealism, put forth by Chakravartty (1998) as a way of being realists while still answering the pessimistic meta-induction argument. Semirealism can be understood in terms of dispositions and it is argued that we ought to be realist about the causal bases of dispositions. This view allows us to be realists about both entities and structures. Various formulations of inference to the best explanation are then presented. It is argued that this principle ought to be construed such that the principle argues to its conclusions having prima facie warrant. It is then argued that realism is better off than constructive empiricism with regard to answering a weak version of the underdetermination of theory by data argument. Finally it is argued that the no miracles argument for scientific realism (NMA) is much more limited in its scope than is often suggested. It is shown how a recent formulation of NMA suggested by Psillos (1999) relies on a traditional formulation of NMA.