Classical decision theory provides a normative theory for representing and reasoning with complex preferences. But straightforward application of this theory to construct automated decision making agents is difficult due to the high cost of eliciting utilities. People tend to express their preferences in qualitative terms and have troubles translating them into numeric utilities. This difficulty has recently sparked interests in studies of qualitative theories for decision making. In this paper I sample several such proposals which share the same focus on developing logic-oriented approaches to representing and reasoning with preferences.