In Multiple Criteria Decision Analysis, the preferences of the decision maker regarding each criterion are classically modelled by a utility function depending on one single variable representing the point of view attached to the criteria. There are situations in which the preferences regarding an attribute depend on more variables. In this case, we propose an extended model taking into account all contributing variables. The main asset of our approach is that it does not requires much more data from the decision maker than the classical model. We also propose an explanation framework.