Drechsler, Mareile
(2012)
Rationality, decisions and large worlds.
PhD thesis, London School of Economics and Political Science.
Abstract
Taking Savage's (1954) subjective expected utility theory as a starting point, this thesis distinguishes three types of uncertainty which are incompatible with Savage's theory for small worlds: ambiguity, option uncertainty and state space
uncertainty. Under ambiguity agents cannot form a unique and additive probability function over the state space. Option uncertainty exists when agents cannot assign unique
consequences to every state. Finally, state space uncertainty arises when the state space the agent constructs is not exhaustive, such that unforeseen contingencies
can occur. Chapter 2 explains Savage's notions of small and large worlds, and shows that ambiguity, option and state space uncertainty are incompatible with the small
world representation. The chapter examines whether it is possible to reduce these types of uncertainty to one another. Chapter 3 suggests a definition of objective ambiguity by extending Savage's framework to include an exogenous likelihood ranking over events. The definition
allows for a precise distinction between ambiguity and ambiguity attitude. The chapter argues that under objective ambiguity, ambiguity aversion is normatively permissible.
Chapter 4 gives a model of option uncertainty. Using the two weak assumptions that the status quo is not uncertain, and that agents are option uncertainty averse, we derive status quo bias, the empirical tendency for agents to choose the status quo over other available alternatives. The model can be seen as rationalising status quo bias.
Chapter 5 gives an axiomatic characterisation and corresponding representation theorem for the priority heuristic, a heuristic which predicts binary decisions be-
tween lotteries particularly well. The chapter analyses the normative implications of this descriptive model.
Chapter 6 defends the pluralist view of decision theory this thesis assumes. The chapter discusses possible applications of the types of uncertainty defined in the thesis, and concludes.
Actions (login required)
 |
Record administration - authorised staff only |
Download statistics
Download statistics