[e]

Jim Joyce: Decision Theory

Tutorial Abstract
Decision theory seeks to provide a normative account of rational decision making, and to determine the extent to which human agents succeed in living up to the rational ideal. These lectures will introduce some central problems of decision theory, paying special attention to those that interest philosophers. We will focus primarily on evaluating the normative status of the principle of expected utility maximization as a requirement of practical rationality. I will begin by introducing the notion of a decision problem and explaining the formulations of expected utility theory due to Leonard Savage and Richard Jeffrey.
We will then discuss the following questions:

• Are "state dependent" utilities required for a full account of rational decision making? More generally, can questions of value be separated from questions of belief in the way that expected utility theory assumes?
• What account of practical rationality is assumed in decision theory, and what is its relationship to epistemic rationality? Here one important issue will be the extent to which the tenets of epistemic rationality can be deduced from the principles of practical rationality.
• Should preferences be interpreted behaviorally? To what extent should we think of subjective degrees of belief as reducible to stable patterns of preference behavior?
• What is risk and how is it measured? Here we will discuss the Allais paradox, which seems to show that expected utility theory cannot adequately capture rational attitudes toward risk. The critical question turns out to be whether risk is a "separable" quantity that obeys the "sure thing principle".
• How should evidential ambiguity be handled by decision theory? Here we will discuss the Ellsberg paradox and raise issues about how decision theory applies in cases where decision makers have beliefs that are best represented by "imprecise" (= set valued) prior probabilities.
• Does decision theory need to explicitly appeal to an agent's beliefs about causal relationships, or are ordinary non-causal beliefs sufficient? Here we will discuss the debate between causal and evidential decision theorists.
• What is the proper role of ratifiability consideration in decision theory? More generally, to what extent should the fact that an agent knows that she will regret performing an action influence her decision making? Here we will discuss some recent work on principles of “preference reflection,” which require agents to align their current preferences to what they expect their future preferences to be.
• What should we say about "Death in Damascus" cases in which it seems that no action can be rationally chosen?
• Does it make sense for rational agents make predictions about what they are likely to do in the course of their deliberations by assigning subjective probabilities to their own acts? If so, what role should these act probabilities play?

USEFUL READING

Jeffrey, Richard. Probabilistic Thinking, Chapter 6, which can be found here

Joyce, James M. The Foundations of Causal Decision Theory. New York: Cambridge University Press, 1999. Chapter 1

_____ "Are Newcomb Problems Really Decisions?" Synthese, 156[3], 2007: 537-562.

______ "Decision Theory," in Donald Borchert, ed., Encyclopedia of Philosophy, Vol. 2. 2nd ed., Macmillan Reference, 2006: 654-661. PENULTIMATE VERSION

______ "Decision Theory," The Philosophy of Science: An Encyclopedia, Routledge, 2005: 181-188. PENULTIMATE VERSION

Shafir, Eldar and Tversky, Amos. "Decision Making," in E. Smith and D. Osherson, eds., An Invitation to Cognitive Science, Volume 3: Thinking, (2nd edition), pp. 77-100. Cambridge: MIT Press, 1995.

Biographical Note.

Jim Joyce is Professor of Philosophy and of Statistics and Chair of the Department of Philosophy of University of Michigan.

Professor Joyce's research concerns the philosophical aspects of probability theory, epistemology, the theory of rational choice, and the general philosophy of science. His book The Foundations of Causal Decision Theory was published by Cambridge University Press in 1999. His recent articles include, "A Nonpragmatic Vindication of Probabilism," Philosophy of Science 65 (December 1998); "Why We Still Need the Logic of Decision" Philosophy of Science 67 (June 2000), "Levi on Causal Decision Theory and the Possibility of Predicting One's Own Actions", Philosophical Studies 110 (2002); "Bayes Theorem", The Stanford Encyclopedia of Philosophy (Winter 2003 Edition); "Bayesianism", in A. Mele and P. Rawling, eds., The Oxford Handbook of Rationality, Oxford University Press, (2004); "Williamson on Knowledge and Evidence", Philosophical Books 45 (2004); "How Degrees of Belief Reflect Evidence", Philosophical Perspectives 19 (2005), "Are Newcomb Problems Really Decisions?", in Synthese.