Probability: The Concept and Its Rules of Use

Abstract

Over the past several centuries, progress on the applications of probability has

far surpassed our philosophical understanding of the nature of probability. This

dissertation consists of four chapters that explore the nature of our judgments about

probability.

In the first chapter, I present my guiding perspective of judgments about

probability as moves in a kind of practice. Judgments about probability should be

understood in terms of the intellectual practice of assigning probabilities, rather than in terms of any representative content; it is the dimensions of that practice that make judgments about probability count as judgments about probability. I propose that we assign probabilities as a way of distilling how we feel about the bearing of bodies of evidence on different propositions.

In the second chapter, I develop an account of the cognitive foundations of these judgments. I suggest that making a judgment about probability involves applying a special probabilistic concept to a proposition. Because judgments about probability involve our deployment of certain concepts and gradations of confidence do not, the former sorts of attitudes are more cognitively sophisticated than the latter.

In the final two chapters, I explore the sources of the most important norms that have been alleged to govern probability. In the third chapter, I offer an insubstantial explanation: those norms are partly constitutive of what it is to be an assignment of probability. We bind ourselves to the norms by intending to engage in a practice constitutively governed by those norms.

In the final chapter, I suggest that the norm of conditionalization, the cornerstone of Bayesian epistemology, is applicable in only a limited range of situations and cannot be made to do all of the work that it is traditionally expected to do. Conditionalization has normative significance because conditionalization preserves certain kinds of commitments. While these kinds of commitments are common, not every assignment of probabilities needs to rely on those kinds of commitments.