The neurotransmitter dopamine has been found to play a crucial role in choice, learning, and belief formation. The best-developed current theory of dopaminergic function is the “reward prediction error” hypothesis—that dopamine encodes the difference between the experienced and predicted “reward” of an event. We provide axiomatic foundations for this hypothesis to help bridge the current conceptual gap between neuroscience and economics. Continued research in this area of overlap between social and natural science promises to overhaul our understanding of how beliefs and preferences are formed, how they evolve, and how they play out in the act of choice.

Apparently mistaken decisions are ubiquitous. To what extent does this reflect irrationality, as opposed to a rational trade-off between the costs of information acquisition and the expected benefits of learning? We develop a revealed preference test that characterizes all patterns of choice “mistakes” consistent with a general model of optimal costly information acquisition and identify the extent to which information costs can be recovered from choice data. (JEL D11, D81, D83)

How and to what extent “neuroeconomic” data (broadly interpreted as data other than standard choice data) should be used in advancing economic theory is open to question. Several authors have attempted to make use of such nonstandard data to shed light on the process of economic decision making. John W. Payne, James R. Bettman, and Eric. J. Johnson (1993), Miguel Costa Gomes, Vincent P. Crawford, and Bruno Broseta (2001), and Xavier Gabaix et al. (2006) have used MouseLab software in order to determine the manner in which people use information. Joseph Wang, Michael Spezio, and Colin Camerer (2006) make use of eye-tracking data for the same purpose. More dramatically, researchers such as Paul William Glimcher, Joseph Kable, and Kenway Louie (2007) are using brain-scanning data in an attempt to constrain economic models of discounting and time preference. Camerer (forthcoming) presents an excellent review of economic research involving nonstandard data. In opposition to this trend, Faruk Gul and Wolfgang Pesendorfer (forthcoming) present a strong critique of the use of nonchoice data within economics. They put forward two specific arguments that users of “neuroeconomic” data must refute if their work is to be taken seriously. First—economic models were designed only to explain choices. Thus, nonchoice data can be used neither to confirm nor deny a particular economic model. Second, it is by and large true that economists are interested in choice behavior. Any two models will either make different predictions for choice, in which case they can be differentiated by standard choice data, or they will not, in which case an economist will not be interested in differentiating between them. Economic Insights from “Neuroeconomic” Data

Paul Glimcher (2003) and Colin Camerer, George Loewenstein, and Drazen Prelec (2005) make powerful cases in favor of neuroeconomic research. Yet in their equally powerful defense of standard “Mindless Economics,” Faruk Gul and Wolfgang Pesendorfer (forthcoming) point to the profound language gap between the two contributing disciplines. For example, for an economist, risk aversion captures preferences among wealth lotteries. From the neuroscientific viewpoint, it is a broader concept related to fear responses and the amygdala. Furthermore, as economic models make no predictions concerning brain activity, neurological data can neither support nor refute these models. Rather than looking to connect such distinct abstractions, Gul and Pesendorfer (forthcoming) argue for explicit separation: “The requirement that economic theories simultaneously account for economic data and brain imaging data places an unreasonable burden on economic theories.” We share the conviction of Glimcher (2003) and Camerer, Loewenstein, and Prelec (2005) concerning the potential value of neuroeconomics, yet we believe that the field will live up to its potential only if a common conceptual language can be agreed upon. Hence, we face the Gul and Pesendorfer challenge head on by developing theories that simultaneously account for behavioral and brain imaging data. The principal innovation lies in our use of the decision theorists’ standard axiomatic methodology in this highly nonstandard setting. This removes any linguistic confusion by defining concepts directly in terms of their empirical counterparts. It also allows us to pinpoint how to design experiments directed to the central tenets of the theory, rather than to particular parametrizations. If these experimental tests reveal the theory to be wanting, then knowing which axiom is The Neuroeconomic Theory of Learning

We introduce a new measure of how close a set of choices is to satisfying the observable implications of rationality and apply it to a large, balanced panel of household level consumption data. This new measure, the minimum cost index, is the minimum cost of breaking all revealed preference cycles found in choices from budget sets. Unlike existing measures of rationality, it responds to both the number and severity of revealed preference violations.

Many everyday decisions are made without full examination of all available options, and, as a result, the best available option may be missed. We develop a search-theoretic choice experiment to study the impact of incomplete consideration on the quality of choices. We find that many decisions can be understood using the satisficing model of Herbert Simon (1955): most subjects search sequentially, stopping when a “satisficing” level of reservation utility is realized. We find that reservation utilities and search order respond systematically to changes in the decision making environment. (JEL D03, D12, D83)

We unite two basic approaches to modelling limited attention in choice by showing that the rational inattention model implies the formation of consideration sets—only a subset of the available alternatives will be considered for choice. We provide necessary and sufficient conditions for rationally inattentive behaviour which allow the identification of consideration sets. In simple settings, chosen options are those that are best on a stand-alone basis. In richer settings, the consideration set can only be identified holistically. In addition to payoffs, prior beliefs impact consideration sets. Linear inequalities identify all priors consistent with each possible consideration set.

Progress on child mortality requires better curative care, but common policies to improve access risk reducing underuse incompletely or creating overuse. In an RCT of 1,768 children in Mali we analyze how subsidized care and community health worker (CHW) visits affect the targeting of acute care, using nine weeks of daily health data to measure demand conditional on need for care per WHO standards. Parents are five times more likely to seek care when medically indicated. Subsidies increase utilization by over 250%, significantly reducing underuse with moderate effects on overuse. CHW do not improve efficient health care utilization on average.

We introduce three new classes of attention cost functions: posterior separable, uniformly posterior separable, and invariant posterior separable. As with the Shannon cost function, all can be solved using Lagrangian methods. Uniformly posterior-separable cost functions capture many forms of sequential learning and hence play a key role in many applications. Invariant posterior-separable cost functions make learning strategies depend exclusively on payoff uncertainty. We introduce two behavioral axioms, Locally Invariant Posteriors and Only Payoffs Matter, which identify posterior-separable functions as uniformly and invariant posterior separable, respectively. In combination, they pinpoint the Shannon cost function.