Quitting Certainties: A Bayesian Framework Modeling Degrees of Belief

by Michael G. Titelbaum

Michael G. Titelbaum presents a new Bayesian framework for modeling rational degrees of belief, called the Certainty-Loss Framework. Subjective Bayesianism is epistemologists' standard theory of how individuals should change their degrees of belief over time. But despite the theory's power, it is widely recognized to fail for situations agents face every day--cases in which agents forget information, or in which they assign degrees of belief to self-locating claims. Quitting Certainties argues that these failures stem from a common the inability of Conditionalization (Bayesianism's traditional updating rule) to model claims' going from certainty at an earlier time to less-than-certainty later on. It then presents a new Bayesian updating framework that accurately represents rational requirements on agents who undergo certainty loss.

Titelbaum develops this new framework from the ground up, assuming little technical background on the part of his reader. He interprets Bayesian theories as formal models of rational requirements, leading him to discuss both the elements that go into a formal model and the general principles that link formal systems to norms. By reinterpreting Bayesian methodology and altering the theory's updating rules, Titelbaum is able to respond to a host of challenges to Bayesianism both old and new. These responses lead in turn to deeper questions about commitment, consistency, and the nature of information.

Quitting Certainties presents the first systematic, comprehensive Bayesian framework unifying the treatment of memory loss and context-sensitivity. It develops this framework, motivates it, compares it to alternatives, then applies it to cases in epistemology, decision theory, the theory of identity, and the philosophy of quantum mechanics.

Genres: Philosophy

362 pages, Hardcover

First published December 1, 2012

About the author

Michael G. Titelbaum is a Vilas Distinguished Achievement Professor in the Department of Philosophy at the University of Wisconsin-Madison.

Reviews

[Owen · Rating 4 out of 5]

Coming from a mathematical background, I thoroughly appreciate the author's modeling approach to philosophy. Titelbaum gives a solid case for (and discussion of the limitations of) his Certainty Loss Framework. There were a couple minor spots where I thought the reasoning wasn't fully supported ("verdicts" being treated as partial results in one area and complete results in another, or transporting verdicts from a story with a reduced timeset to a story with a larger timeset, for example), but overall I thought this was an exceptionally clear and substantial contribution to the philosophy of probability.