Prospect theory is currently the main descriptive theory of decision under uncertainty. It generalizes expected utility by introducing nonlinear decision weighting and loss aversion. A difficulty in the study of multiattribute utility under prospect theory is to determine when an attribute yields a gain or a loss. One possibility, which has been adopted in the existing theoretical literature on multiattribute utility under prospect theory, is to assume that a decision maker views the complete outcome as a gain or a loss. In this holistic approach, decision
weighting and loss aversion are general and attribute-independent. Another possibility, more common in the empirical literature, is to assume that a decision maker has a reference point for each attribute. We give preference foundations for this segregated approach in which decision weighting and loss aversion are attribute-specific.