There are many situations in which individuals have a choice of whether or not to observe the eventual outcome. In these instances, individuals often prefer to avoid observing the outcome. The standard von Neumann-Morgenstern (vNM) Expected Utility model cannot accommodate these cases, since it does not distinguish between lotteries for which outcomes are observed by the agent and lotteries for which they are not. I develop an axiomatic model that admits preferences for observing the outcome or remaining in doubt. I then use this model to analyze the connection between the agent's attitude towards risk, doubt, and what I refer to as 'optimism'. This framework accommodates a wide array of field and experimental observations that violate the vNM model, and that may not seem related, prima facie. For instance, this framework accommodates self-handicapping, in which an agent chooses to impair his own performance. Unlike other frameworks, this model accommodates self-handicapping without using notions of self-deception, cognitive dissonance and belief manipulation. It also admits a status quo bias without having recourse to framing effects or reference points. Furthermore, this framework accommodates behavior associated with anticipated regret, the Allais paradox and preferences for smaller menus, which are all difficult to reconcile with the vNM framework. In financial settings, this model accommodates a safe allocation bias, in which agents choose neither to buy nor short sell an asset for an interval of prices; this behavior has so far been explained using ambiguity aversion, which this model does not allow. Recently, experiments have been conducted in which dictators in dictator games who seem to exhibit preferences for fairness often switch to the selfish choice if they can avoid observing the recipients allocation. While the empirical findings of these experiments are difficult to reconcile either with models of Expected Utility or models of fairness, they fit the predictions of this model well. This framework accommodates all the well-known observations mentioned here and others described in the papers with a single, natural extension of the standard vNM model, and using the same assumption on preferences throughout.
展开▼
