Wheeler: Coherence

by Rush Stewart

News from Jeff:

I want to call your attention to the following “brown bag lunch” talk by Greg Wheeler, Senior Research Scientist and Executive Board Member at the Center for Artificial Intelligence (CENTRIA) in Lisbon. The talk should be of interest to researchers from a variety of fields that make use of Bayesian methods. The abstract for Greg’s talk is as follows:

Coherence at last!
Gregory Wheeler (joint work with Richard Scheines, Carnegie Mellon)

Abstract. Coherentism maintains that coherent beliefs are more likely to be true than incoherent beliefs, and that coherent evidence provides more confirmation of a hypothesis when the evidence is made coherent by the explanation provided by that hypothesis. Although probabilistic models of credence ought to be well-suited to justifying such claims, negative results from Bayesian epistemology have suggested otherwise. In this essay we argue that the connection between coherence and confirmation should be understood as a relation mediated by the causal relationships among the evidence and a hypothesis, and we offer a framework for doing so by fitting together probabilistic models of coherence, confirmation, and causation. We show that the causal structure among the evidence and hypothesis is sometimes enough to determine whether the coherence of the evidence boosts confirmation of the hypothesis, makes no difference to it, or even reduces it.  We also show that, ceteris paribus, it is not the coherence of the evidence that boosts confirmation, but rather the ratio of the coherence of the evidence to the coherence of the evidence conditional on a hypothesis.