Title: Relative Maximum Likelihood Updating of Ambiguous Beliefs

Abstract: When ambiguous beliefs are represented by multiple priors, updating such a belief may also include refining of the initial belief by discarding some of the priors. Maximum Likelihood (ML) and Full Bayesian (FB) are the two extremes of refining beliefs based on likelihood. To capture behaviors between these two extremes, this paper proposes a new updating rule: Relative Maximum Likelihood (RML) updating. This rule updates an intermediate set of priors that is a linear contraction of the initial set with respect to the set of priors attaining maximum likelihood. The linear contraction parameter captures the DM's relative inclination towards ML with respect to FB, hence ML and FB are the two extreme special cases of RML.
More importantly, characterization of conditional preferences given by RML updating is provided when the preferences are represented by the Maxmin Expected Utility. The axiomatization relies on weakening the axioms that characterize FB and ML. In particular, the axiom that characterizes ML is identified in this paper, fixing an existing problem in the literature.