Abstract

I advance a decision principle called the "weak dominance principle" (WDP) based on the interval notion of probability to deal with the Ellsberg type paradox (ETP). Given ETP, I explain three things: (i) Why WDP is a better principle than many principles e.g. Kyburg's principle and Gardenfors and Sahlin's principle, (ii) Why one should not, contrary to many principles, expect a unique solution in ETP, and (iii) What is the relationship between WDP and the principles mentioned above. I prove also that WDP induces a strict partial ordering on the intervals to which it is applied.