Learning and Updating with Non-Additive Beliefs

• Applicant: Professor Dr. Jürgen Eichberger
• Subject Area: Economic Theory
• Term: from 2009 to 2010
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 119786568

Driven by experimental results on individual decision making under uncertainty, decision theory has moved away from the expected utility approach which dominated economic theory over the past decades. With new decision models, such as Choquet expected utility or the multiple prior approach, aspects of decision making like ambiguity about or confidence in a belief could be modelled as well as optimistic or pessimistic attitudes towards ambiguity.For the new models of decision-making under uncertainty, learning and adjustment of beliefs in the light of new information need to be reconsidered because the well-known methods of Bayesian updating are no longer applicable.In particular, for applications in economics and game theory it will be necessary to find suitable notions for updating beliefs if there is ambiguity or a lack of confidence. Closely related is the question of how to model beliefs about independent behaviour of the opponents in interactive situations. Progress on these issues has been slow because there are conflicting aspects of the problem which strongly influence modelling choices.With this research project we want to continue work on updating rules for capacities and multiple priors. In particular, we will focus on special cases such as neo-additive capacities, which have proved to be useful in economic applications. In addition, we want to develop a notion of a product measure for capacities which is suitable for game theoretic applications.

• DFG Programme: Research Grants
• International Connection: United Kingdom
• Participating Person: Professor Dr. David Kelsey