An axiomatic approach to \(\varepsilon\)-contamination (Q2580963)

The paper deals with decision-making in which the uncertainty regards not only a single probability measure but a set of them. Namely, it is assumed that an economic agent is almost certain (say, (\(1-\varepsilon\)).\(100\%\) certain) that the uncertainty he faces is characterized by a particular probability measure, but there is a minor fear (say, \(\varepsilon\dot 100\%\)) that it is not so, and he is left completely ignorant about the true measure. That model is called \(\varepsilon\)-contamination of confidence. The aim of the paper is to provide a simple set of behavioral axioms under which the decision-maker's preference is represented by the Choquet expected utility with the considered contamination of confidence.