Collective decisions are pervasive in every aspect of our society. As shown by Brexit and other recent episodes in European national politics, the way we shape our collective decisions can have far-reaching implications. However, even though there is a vast literature on the aggregation and revision of preferences in and by groups, there are no exact mathematical models of the way groups construct and revise a given decision problem – a process I call constructive deliberation. Getting grip on this political “logic of discovery” is a necessary step towards a suitable combination of information technology with various forms of direct democracy. In this project, I will develop a formal theory of constructive deliberation. Relying on my background in defeasible reasoning and the expertise at the host and secondment institution, I will create mathematical models of individual agents that reason and share information about the decision problem at hand, and show how and when this interaction results in group consensus. This way, the project will complement existing work on collective decision-making, which is in turn necessary for the improvement of democratic processes. In combination with specific training, dissemination, and communication actions, the project will also contribute to my personal development, serving my long-term aim to obtain a permanent position as a researcher (cf. §1.2-1.4).