Preferences of Agents in Defeasible Logic