Modelling Democratic Deliberation

Deliberative democracy is a political theory that places deliberation at the heart of political decision making. In a deliberation, people justify their preferences to one another. They are confronted with new information and new perspectives, which might lead them to change their preferences. Therefore, deliberative democracy, unlike social choice theory, takes preferences to be secondary (derived) and dynamic rather than primary and static. The first goal of this thesis is to formally model deliberation as aspired by deliberative democracy, henceforward referred to as democratic deliberation. This is done in two steps. Firstly, this thesis develops models for preference formation, since democratic deliberation is about justifying one’s preferences. These models combine multi-agent plausibility models from dynamic epistemic logic with Dietrich and List’s setting about reasons and rational choice. Combining these allows us to define the agents’ preferences in terms of (i) their knowledge and belief, (ii) their motivational state or perspective and (iii) the properties that hold of the alternatives. Secondly, we introduce a model transformer for the preference formation models that models deliberation as a process in which all agents share all their information and all their perspectives. Together, the preference formation models and the model transformer for deliberation make up our formal framework. This framework is able to model two claims that are often made in the literature on deliberative democracy, namely that deliberation might lead to preference change and to a better understanding among the agents. The second goal of this thesis is to use this formal framework to investigate the philosophical claim that deliberation provides an escape from social choice theory’s impossibility results. The main result proved in this thesis is that in cases where the issue at stake is one-dimensional, deliberation is useful because it ensures single-peaked preferences via meta-agreement, and hence helps to circumvent Arrow’s impossibility result.