Skip to ContentSkip to Navigation
Over onsFaculteit WijsbegeerteNieuwsAgenda

Formal Epistemology Colloquium

The Department of Theoretical Philosophy in Groningen hosts a colloquium on Formal Epistemology with talks by Kenny Easwaran (Texas A&M) and Gerhard Schurz (Düsseldorf).

Measuring Confidence  -  Kenny Easwaran (Texas A&M)

One objection many philosophers have raised to Bayesianism is the claim that it is implausible that degrees of belief could be so finely grained as to adequately be represented by precise real numbers. I will consider analogies to other numerical quantities and use them to argue that the precision implicit in Bayesianism is not in fact unrealistic.

The No Free Lunch Theorem and the A Priori Advantages of Regret-Based Meta-Induction  -  Gerhard Schurz (Düsseldorf)

Recently a new account to the problem of induction has been developed (Schurz 2008, Schurz and Thorn 2016) based on a priori advantages of regret-based meta-induction (RW) in online learning (Cesa-Bianchi and Lugosi 2006). The claimed a priori advantages seem to contradict the no free lunch (NFL) theorem (Wolpert 1996), which asserts that relative to a uniform prior distribution (UPD) over possible worlds all (non-clairvoyant) prediction methods have the same expected predictive success. It is argued that the NFL challenge cannot be `solved' by computing expected success relative to the `actual' instead of some prior distribution (as frequently argued) because this idea is viciously circular. A new way of dissolving the NFL challenge is developed, leading to four novel results:

  • RW enjoys free lunches, i.e., its predictive long-run success dominates that of other prediction strategies.
  • Yet the NFL theorem applies to iterated online prediction tasks provided the prior distribution is a UPD.
  • The UPD is maximally induction-hostile and assigns a probability of zero to all possible worlds in which RW enjoys free lunches. This dissolves the apparent conflict with the NFL.
  • The a priori advantages of RW can be demonstrated even under the assumption of a UPD. Further advantages become apparent when a frequency-uniform distribution is considered.
When & where?
  • Wednesday, June 7 from 13:00 until 17:00
  • Room Omega of the Faculty of Philosophy, Groningen
Last modified:26 May 2017 6.39 p.m.