On the logic of theory change: Contraction functions and their associated revision functions