Does temptation decline over time? Recent studies have highlighted the importance of Pavlovian processes but less is known about how these responses change over time. In a laboratory experiment, every subject made a choice between a banana and chocolate, but a treatment group was informed in advance about the existence of the upcoming choice. These treated subjects were 28% more likely to ultimately choose a banana. Testing an alternative hypothesis of limited willpower, I find no evidence of a simple resource depletion effect using previously induced effort.

We study preferences over lotteries in which both the prize and the payment date are uncertain. In particular, a time lottery is one in which the prize is fixed but the date is random. With Expected Discounted Utility, individuals must be risk seeking over time lotteries (RSTL). In an incentivized experiment, however, we find that almost all subjects violate this property. Our main contributions are theoretical. We first show that within a very broad class of models, which includes many forms of non-Expected Utility and time discounting, it is impossible to accommodate even a single violation of RSTL without also violating a property we termed Stochastic Impatience, a risky counterpart of standard Impatience. We then present two positive results. If one wishes to maintain Stochastic Impatience, violations of RSTL can be accommodated by keeping Independence within periods while relaxing it across periods. If, instead, one is willing to forego Stochastic Impatience, violations of RSTL can be accommodated with a simple generalization of Expected Discounted Utility, obtained by imposing only the behavioral postulates of Discounted Utility and Expected Utility. Supplementary Appendix

This paper examines how past effort can impact subsequent effort, such as whether effort is reduced following an interruption. I conducted 4 incentivized real-effort experiments in which piece rates, interruption source, and leisure options were manipulated. For self-selected 'internal' interruptions, I find effort displays significant stickiness and is indicative of effort ''momentum'', rather than on-the-job learning, reciprocity, or income targeting. Five minutes after incentives return to baseline, 45% of this effort increase or decrease persists. For exogenous 'external' interruptions, however, there's no evidence of stickiness, with somewhat precise zeros.

This paper examines how risk preferences differ over goods and in-kind monetary rewards. I study an incentivized experiment in which control subjects allocate Amazon.com credit over uncertain states, whereas treated subjects allocate self-selected Amazon.com goods over uncertain states. Under a standard model with perfect information of prices, I demonstrate allocations would be identical between treatments. In practice, subjects demonstrate considerable differences across goods and monetary rewards, with credit being more evenly allocated among the uncertain states. Using an additional information treatment, I find no evidence that price or product uncertainty explains these differences. I further show that these results are not being driven by fungibility, functional form, or good discreteness.

Monetary lotteries are the overwhelmingly predominant tool for understanding decisions under risk. How- ever, many real-world decisions concern multidimensional out-comes involving different goods. Recent studies have tested whether people treat multidimensional risky choices in the same manner as unidimensional mon- etary lotteries and found that choices over consumer goods are less risk-averse and more consistent with expected utility theory than choices over monetary lotteries. While these puzzling results cannot be ex- plained by any standard model of decision making, we demonstrate that these findings are predicted by a salience-based model of category-dependent preferences that also explains the classic anomalies for choices under risk. Additionally, we experimentally verify a novel prediction of this Categorical Salience Theory. We further demonstrate that our model can explain empirical puzzles in financial markets, insurance markets, and principal agent settings, including behavior in a new portfolio choice experiment that is unexplained by expected utility theory or prospect theory.

We conducted a laboratory experiment to investigate individual ability to process conflicting social information that could be potentially irrelevant, in which each subject independently draws a ball from one of two digital urns and receives information reported by another subject who may or may not have drawn from the same urn. We find that 71% of subjects who receive new information misattribute the source of the information compared to Bayesian updating. Conflicting information is overly assumed as irrelevant, and confirming information is overly assumed as relevant. This asymmetry is robust even when allowing for subjects to perceive others as reporting non-informative signals. Attributing conflicting information as irrelevant may form the foundation of stable echo chambers or equilibria where additional information has no effect on beliefs.