Preferences can be expressed as wishes, constraints or both. Generally, wishes and constraints do not complement each other. A different reasoning principle is applied to rank-order the set of options depending on whether preferences refer to wishes or constraints. Consequently, these two types of preferences have been characterized by two separate sets of postulates offering a normative view of the preferences. This paper provides a complementary study. In particular an empirical analysis has been conducted in order to assess human endorsement of the normative view of wishes and constraints. Results showed that single postulates are highly endorsed by humans. However the latter showed more or less strong endorsement w.r.t. patterns of postulates. US laypersons and French participants in the experiment showed different behaviors.

This paper investigates the psychological plausibility of the bipolarity concept, i.e., that positive and negative kinds of information are treated differently. Sections 2 and 3 review relevant investigations of the representational and affective systems in the experimental psychology literature. Section 4 provides new data supporting the idea that even when considering how affective changes occur, a certain level of independence exists between the positive and negative sides of affect. Together the studies reported here strongly support the psychological plausibility of bipolarity: Positive and negative kinds of information are not processed in the same way whichever domain is considered, preferences (affect) or beliefs (mental categories). © 2008 Wiley Periodicals Inc.

This paper investigates the statistical properties of the Choquet and Sugeno integrals, used as multiattribute models. The investigation is done on an empirical basis, and focuses on two topics: the distribution of the output of these integrals when the input is corrupted with noise, and the robustness of these models when they are identified using some set of learning data through some learning procedure.