Competing explanations : exclusion and importance in historical accounts

This thesis is a philosophical examination of the nature of explanatory competition between historical accounts. It is usual for a philosophy of explanation to attempt an analysis of explanation, singular. The focus of this work is on relations between two or more explanations. In particular, I investigate explanatory exclusion and explanatory importance. The methodology of the thesis is built upon a conception of descriptive philosophy. I believe that when attempting to philosophise about a practice such as history, we should pay detailed attention to existing good practice. To this end, I develop my conclusions in conjunction with an examination of eight differing explanations of the French Revolution. Explanatory exclusion should be analysed in terms of incompatibility between explanations. Explananda, explanans, or relevance claims may be incompatible. Exclusion of the last type requires a commitment to explanatory realism, which holds that explanatory relationships mirror appropriate ontic - paradigmatically, causal - relationships. There are different types of historical explanation, yet all make reference to the causal history of the French Revolution, and therefore are candidates for explanatory exclusion. Causal attribution is necessary for historical explanation, but not sufficient. Causes must also be described in the `correct' manner, and differing conceptions of `correctness' lead to division between `types' of explanation. Historical explanations may compete, even where they do not exclude. One significant dimension of competition concerns attribution of historical importance. In order to allow substantive explanatory competition over this feature, I develop a realist analysis of historical importance. A more important cause is one which made more of a difference to the effect. In explicating this counterfactual claim, I defend an account of counterfactual decidability based on the idea of counterfactuals as implied experiments. This account is shown to have advantages over traditional metalinguistic and possible worlds analyses of counterfactuals.