A Reconsideration of an Argument against Compatibilism