Explanation as contextual

Abstract

There is a view that all explanation is contextual. An explanation answers questions that are relevant in a context and that are open to solution in that context. In another context, there might be no such questions, or they might not be open to solution. Van Fraassen has used a contextual account of explanation to argue in favour of what he calls ‘constructive empiricism’ and against what he calls ‘scientific realism’. On his account, both empiricists and realists search for theories that are empirically adequate. These will explain the relevant observable phenomena, but differ on the unobservable phenomena, for example quantum states. For the realist, science aims to provide a literally true account of the unobservables. For the empiricist, science aims at no more than empirical adequacy. One argument in the realist armoury is the following. The best philosophical explanation of how the best scientific explanation does explain the observables requires that it is true about the unobservables. An empiricist response to this is that all explanation is contextual, so there is no globally best scientific explanation. The present paper explores the empiricist line by reference to formal learningtheory and a logic of questions. Van Fraassen’s contextual theory of explanation does not employ learningtheory. The present paper is a step towards a more developed theory, differingfrom van Fraassen in some respects.