Using a real-effort experiment, we show that people project their current tastes onto others when forecasting others' behavior, even when their tastes are exogenously manipulated and transparently different. In the first part of our experiment, "workers" stated their willingness to continue working on a tedious task. We varied how many initial tasks workers completed before eliciting their willingness to work (WTW); some were relatively fresh when stating their WTW, while others were relatively tired. Later, a separate group of "predictors"---who also worked on the task---guessed the WTW of workers in each state. We find: (i) tired workers were less willing to work than fresh workers; (ii) predictors (in aggregate) accurately guessed the WTW of workers when they were in the same state as the workers about whom they were predicting, but, (iii) when fresh predictors were guessing about tired workers, they substantially overestimated their WTW, and (iv) when tired predictors were guessing about fresh workers, they underestimated their WTW. Using an additional treatment, we find that workers also mispredicted their own future WTW, and we compare the magnitudes of intra- and interpersonal projection bias.

We demonstrate that people's impressions of a real-effort task are shaped by the elation or disappointment they felt when first working on the task. In our experiments, participants learned from experience about one of two unfamiliar tasks, one clearly more onerous than the other. We manipulated participants' initial expectations about which task they would face: some were assigned their task by chance just prior to their initial experience, while others knew in advance which task they would face. In a second session conducted hours later, we elicited their willingness to work again on their previously assigned task. Participants assigned the less-onerous task by chance were more willing to work than those who faced it with certainty; conversely, those assigned the more-onerous task by chance were less willing to work than those who faced it with certainty. These qualitative results---and the fact that differences in willingness to work were observed hours after first impressions were formed---are consistent with a form of attribution bias wherein participants wrongly ascribed sensations of positive or negative surprise to the underlying disutility of their assigned task.

We examine errors in learning that arise when an agent who suffers attribution bias fails to account for her reference-dependent utility. Such an agent neglects how the sensation of elation or disappointment relative to expectations contributes to her overall utility, and wrongly attributes this component of her utility to the intrinsic value of an outcome. In a sequential-learning environment, this form of misattribution generates contrast effects in evaluations and induces a recency bias: the misattributor's beliefs over-weight recent experiences and under-weight earlier ones. In the long-run, a loss-averse misattributor will grow unduly pessimistic and undervalue prospects in proportion to their variability. Both the short and long-run properties of beliefs under misattribution suggest a tendency to abandon worthwhile prospects when learning from experience. We additionally show how misattribution introduces incentives for familiar forms of expectations management.

Fixed differences loom smaller when compared to large differences. We propose a model of relative thinking where a person weighs a given change along a consumption dimension by less when it is compared to bigger changes along that dimension. In deterministic settings, the model predicts context effects such as the attraction effect, but predicts meaningful bounds on such effects driven by the intrinsic utility for the choices. In risky environments, a person is less likely to sacrifice utility on one dimension to gain utility on another that is made riskier. For example, a person is less likely to exert effort for a fixed monetary return if there is greater overall income uncertainty. We design and run experiments to test basic model predictions, and find support for these predictions.

We propose a neurocomputational model of altruistic choice and test it using behavioral and fMRI data from a task in which subjects make choices between real monetary prizes for themselves and another. We show that a multi-attribute drift-diffusion model, in which choice results from accumulation of a relative value signal that linearly weights payoffs for self and other, captures key patterns of choice, reaction time, and neural response in ventral striatum, temporoparietal junction, and ventromedial prefrontal cortex. The model generates several novel insights into the nature of altruism. It explains when and why generous choices are slower or faster than selfish choices, and why they produce greater response in TPJ and vmPFC, without invoking competition between automatic and deliberative processes or reward value for generosity. It also predicts that when one’s own payoffs are valued more than others’, some generous acts may reflect mistakes rather than genuinely pro-social preferences.

This paper describes a series of laboratory experiments studying whether the form in which items are displayed at the time of decision affects the dollar value that subjects place on them. Using a Becker-DeGroot auction under three different conditions—(i) text displays, (ii) image displays, and (iii) displays of the actual items—we find that subjects' willingness-to-pay is 40–61 percent larger in the real than in the image and text displays. Furthermore, follow-up experiments suggest the presence of the real item triggers preprogrammed consummatory Pavlovian processes that promote behaviors that lead to contact with appetitive items whenever they are available.