Robust Probability Updating

TL;DR

This paper introduces a robust probability updating method that does not rely on assumptions like CAR, providing a general framework for updating probabilities in complex scenarios with arbitrary distributions, messages, and loss functions.

Contribution

It generalizes probability updating beyond traditional assumptions, characterizing optimal strategies under various loss functions, especially with the RCAR condition for logarithmic loss.

Findings

Optimality characterized by RCAR condition for logarithmic loss

Provides existence and characterization theorems for robust updating strategies

Offers a general solution for probability updates with arbitrary distributions and messages

Abstract

This paper discusses an alternative to conditioning that may be used when the probability distribution is not fully specified. It does not require any assumptions (such as CAR: coarsening at random) on the unknown distribution. The well-known Monty Hall problem is the simplest scenario where neither naive conditioning nor the CAR assumption suffice to determine an updated probability distribution. This paper thus addresses a generalization of that problem to arbitrary distributions on finite outcome spaces, arbitrary sets of `messages', and (almost) arbitrary loss functions, and provides existence and characterization theorems for robust probability updating strategies. We find that for logarithmic loss, optimality is characterized by an elegant condition, which we call RCAR (reverse coarsening at random). Under certain conditions, the same condition also characterizes optimality for a much larger class of loss functions, and we obtain an objective and general answer to how one should update probabilities in the light of new information.