SOPhiA 2022

Salzburgiense Concilium Omnibus Philosophis Analyticis

SOPhiA ToolsDE-pageEN-page

Programme - Talk

The epistemic-aleatoric uncertainty distinction for formal epistemology and suspension of judgment
(Epistemology, English)

In formal epistemology we are concerned with providing frameworks that describe our belief states and rules for how we ought to rationally possess and change those belief states. Many of these formal frameworks are graded. They describe to what degree a subject believes a certain proposition rather then describing solely whether a subject believes a proposition or not. In doing so, these frameworks represent the subject's uncertainties about the truth of propositions. It was recognized early in the history of philosophy and probability theory that uncertainties can be of at least two kinds: chance-based and epistemological. Still, modern formal epistemology (in particular Bayesianism) neglects the difference, treating both uncertainties within the same framework. This is in contrast to a recent development in engineering and machine learning research that has drawn increasing attention to the distinction that is called the distinction between aleatoric and epistemic uncertainty. In this paper, I argue that it is fruitful to apply this distinction to formal epistemological frameworks, too. Although the distinction can already help to provide a better view on issues in formal epistemology concerning belief, the need for taking the distinction seriously becomes even more pressing ever since epistemological research has focused to some extent on the notion of suspension of judgment. All formal frameworks should be able to describe the whole doxastic household of a rational subject, including belief, disbelief and suspension. However, it has been noted that traditional formal frameworks such as Bayesian epistemology are not capable of describing suspension properly. I will argue that many problems within the challenge to properly represent suspension arise from not distinguishing aleatoric and epistemic uncertainty. I will briefly show how a Bayesian framework can be adopted to accommodate the distinction. For this, I will introduce a two-dimensional framework that separates the two forms of uncertainty and I will show how rationality demands from Bayesianism can be transferred to the proposed framework.

Chair: Niklas Gärtner
Time: 16:50-17:20, 07 September 2022 (Wednesday)
Location: SR 1.003
Remark: CHANGE. The talk is cancelled!

Daniela Schuster 
(Universität Konstanz, Germany)



Testability and Meaning deco