MoL-2017-05: Manipulating the Manipulators: Richer Models of Strategic Behavior in Judgment Aggregation

MoL-2017-05: Terzopoulou, Zoi (2017) Manipulating the Manipulators: Richer Models of Strategic Behavior in Judgment Aggregation. [Report]

[thumbnail of MoL-2017-05.text.pdf]
Preview
Text
MoL-2017-05.text.pdf

Download (715kB) | Preview

Abstract

Judgment Aggregation is a formal framework of collective decision making. When agents that belong to a group express their individual opinions on a set of logically interconnected issues, a good rule is required in order to combine these opinions and induce a representative collective judgment for the group. However, it is often the case that some agent may find a way to achieve a preferable outcome for herself, by reporting a dishonest opinion. This kind of strategic behavior, namely manipulation, constitutes the heart of this thesis. Until now, researchers have been making two very strong assumptions in the context of Judgment Aggregation: first, that all the agents fully know the truthful opinions of all the other members of their group; and second, that every agent thinks that everyone else is always truthful. In this thesis we start with enriching the existing model of Judgment Aggregation in a twofold manner: we account for partially informed agents who behave strategically under various types of uncertainty, as well as for higher-level strategic reasoners, who reflect on the reasoning of their peers. We employ analytical methods, and we explore how the aforementioned assumptions affect the agents’ choices and insincere acts in Judgment Aggregation. Moreover, after investigating in depth single-round aggregation processes, a model of Iterative Judgment Aggregation is developed. In our third model, the agents have the possibility to change their initially submitted judgments in a sequential random order, while they are (maybe partially) observing the acts of their peers. We study paq whether common aggregation rules in iteration reach equilibria states, pbq how fast they do, and pcq what the potential benefits of strategic behavior are, for a group of agents en masse.

Item Type: Report
Report Nr: MoL-2017-05
Series Name: Master of Logic Thesis (MoL) Series
Year: 2017
Subjects: Language
Logic
Depositing User: Dr Marco Vervoort
Date Deposited: 29 Jun 2017 13:38
Last Modified: 13 Jul 2017 13:27
URI: https://eprints.illc.uva.nl/id/eprint/1541

Actions (login required)

View Item View Item