Browse our site
About
People
Research Areas
Projects
Publications
Seminars
Future
Past
News
You are here:
Home
Seminars
Past
Reconsidering Coherence Reasoning
CENTRIA seminar: Reconsidering Coherence Reasoning
Main information
By:
Gregory Wheeler
(
CENTRIA
)
Date:
Wednesday, 21st of January 2009, 14h00
Location:
FCT/UNL, Seminar Room (Ed. II)
Abstract
(joint work with Richard Scheines, Carnegie Mellon University)
We hear about coherent evidence for some hypothesis, but what does `coherent evidence' mean? Intuitively, the evidence `fits together', lending support' for the truth of the hypothesis. Surprisingly little progress has been made spelling out the details of coherence justification since C.I. Lewis' speculative work on the epistemology of memory, in 1946. Generations of coherence theorists have developed the general idea of coherentism, but the idea has remained largely confined to philosophy because of a missing theory *of* coherence.
What's changed recently are so-called impossibility results (Bovens and Hartmann 2003, 2005; Olsson 2005) purporting to show that one cannot formulate a core idea behind coherence reasoning using the machinery of probability. We've been waiting 60+ years for good reason, these results tell us: the coherence idea is incoherent! However, I have shown (Wheeler 2009) that the impossibility result are a product of the independence assumptions used in these theorems, and I've shown that once these conditions are relaxed, then there is a measure where correlation (coherence) tracks increases in posterior probability (positive confirmation) just fine. Working with Richard Scheines, we have shown this result is very robust across a variety of common confirmation measures, and we have developed a causal account to explain why the (Wheeler 2009) measure works as well as it does: when the association of evidence caused by the truth of the hypothesis is greater than the association of the evidence alone, then the posterior probability of the hypothesis will be positive.
This talk will give an overview of coherence reasoning, the nature of the problem we solved, and some remarks on future work.
Short-bio
Gregory Wheeler is a Senior Research Scientist at CENTRIA, The Center for Artificial Intelligence Research at the New University of Lisbon, where is also Lecturer in Computational Logic. He holds a joint PhD in Philosophy and Computer Science from the University of Rochester. He works primarily on uncertainty frameworks, probabilistic logic, and foundations of knowledge representation (a.k.a. formal epistemology).
Departamento de Informática, FCT/UNL
Quinta da Torre 2829-516 CAPARICA - Portugal
Tel. (+351) 21 294 8536 FAX (+351) 21 294 8541
