Talk: Rainer Hegselmann (Bayreuth)

Title:

Computational social epistemology: A case study on two-armed bandits versus inductive truth seekers and epistemic free riders

Abstract:

The talk is about a community in which individuals gather experimental evidence, modify their opinions in a social exchange process,and act on their actual opinions. An extended version of the so called bounded confidence model is the central tool of analysis.

In the extended model, a community has a choice between n strategies (n = 2), each of which leads to success with a constant but unknown probability. Such problems are known as n-armed bandit problems. One part of the community consists of truth seekers who experiment with the two strategies and evaluate their experiences. Another part of the community consists of epistemic free-riders who do not do any statistics at all. At the same time, however, both groups, the truth seekers and the free riders, are involved in a social exchange process. Overall, we are dealing with the dynamics of truth approximation in an epistemic community in which both an objective and a social component are at work.

I hope to show that the given scenario is both rich and simple enough not only to formulate but also to answer interesting efficiency questions about different epistemic policies.