The standard representation theorem for expected utility theory says roughly that if a subject’s preferences conform to certain  conditions, then she can be represented as maximising her expected utility given a particular set of credences and utilities—and, moreover, that having those credences and utilities is the only way that she could be an expected utility maximiser, given the facts about her preferences. These theorems are widely taken to provide the mathematical, normative, and (in some cases) conceptual basis for contemporary decision theory in a wide range of disciplines. However, the kinds of agents that the theorems seem apt to tell us anything about are highly idealised, being (amongst other things) always probabilistically coherent with infinitely precise degrees of belief and full knowledge of all epistemically necessary truths. Ordinary agents do not look very rational when compared to the angels usually talked about in decision theory. In this paper, I will outline a theorem aimed at the representation of those who are not probabilistically coherent, logically omniscient, or even very good decision-makers—i.e., agents who arefrequently irrational. The agents in question may have highly incoherent credences, limited representational capacities, and are only assumed to (i) be deductively competent with respect to obvious implications, and (ii) maximise expected utility with respect to a restricted class of relatively simple gambles.

Philosophy Seminar Series
Date: Thursday, 4 February 2016
Time: 3pm – 5pm
Venue: AS3 #05-23
Speaker: Edward Elliot
Moderator: Prof Neiladri Sinhababu

About the Speaker:

Edward Elliot presently works mainly on issues to do with the conceptual foundations of decision theory and formal epistemology, with a special interest in the nature and content of degrees of belief. His long-term research project is to understand representational phenomena of all kinds (mental, linguistic, scientific), and to situate them within the natural world. He was awarded his PhD in November 2015, from the Australian National University. His thesis was on the topic of decision-theoretic representation theorems and their connection to the characterisation and naturalisation of degrees of belief and utilities.