Boning Yu

This paper establishes a calibration theorem for risk-weighted expected utility theory (REU). REU extends expected utility theory (EU) by introducing a risk function, which allows a decision-maker to weigh potential outcomes according to their respective risks when determining the instrumental value of an action. REU is intuitively appealing as it seems to resolve some of the classical challenges to EU, such as the Allais paradox and Rabin’s calibration result. Using only widely accepted and minimally controversial assumptions, I calibrate REU using Allais preferences and show that accommodating Allais preferences commits REU to questionable behaviors. Thus, REU resolves the Allais paradox only at the cost of becoming susceptible to a Rabin-style challenge (being calibratable). Compared to existing calibration results against REU, my approach relies on less demanding and more realistic assumptions.

I present a model of group deliberation that conceives of evidence discovery as a dynamic process, where agents’ chance of discovering high-quality evidence depends on how close their current beliefs are to the truth. Through computer simulations, I examine the impact of different practices of communication of both evidence and conclusions on groups’ performance in finding the truth. I draw three major conclusions. First, when agents have access to high-quality evidence, communication about evidence improves groups’ performance by allowing agents to accumulate a larger body of evidence; however, when agents lack access to high-quality evidence, communication about conclusions allows them to obtain relatively high-quality information by learning from others’ results. Second, for any given practice of communication about evidence, extensive communication about conclusions almost always improves a group’s ability to converge on the truth; nevertheless, when agents are communicating their conclusions, extensive communication about evidence, despite promoting diversified information, is not necessarily beneficial. Third, in scenarios where discovering the truth is challenging, the best strategy is to learn all asserted conclusions without communicating evidence; however, in scenarios where agents have decent access to high-quality evidence, combining full communication of conclusions with some degree of communication about evidence is the best practice.