Abstract

A survey of the modern theories of explanation leads us to conclude that it is J. Woodward\'s conterfactual approach that seems to set out a promising agenda for setting up a satisfactory model of explanation. Such a model would need to come to terms with two stipulations, i.e. to reflect causal asymmetry and be empirically adequate. To meet the latter I modify the counterfactual model of explanation to require a class of observationally equivalent (Bayesian) causal models instead of the true causal model. For in most cases empirical adequacy of causal models, i.e. their identifiability and consistency with observed distributions, would neither grant the stronger conclusion Woodward embeds in his conterfactual account of explanation nor license metaphysical naturalism