Predictions and choices for others: Some insights into how and why they differ. (2023). Journal of Experimental Psychology: General.
Abstract: Being able to learn another person’s preferences and choose on their behalf are important skills. However, people often do not choose what the other would choose for themselves. Over two incentive- compatible studies, we identify how and why people choose differently for others than the others would choose for themselves. Participants observed choices made by another person and then (a) predicted what this person would choose or (b) chose for them in new decisions, while we tracked their mouse movements. Participants learned noisy human preferences as easily as they learned noiseless algorithms. Moreover, participants’ predictions of what others would choose were in line with the others’ actual choices roughly 80% of the time, regardless of whether they were paid for predicting consistently with the others’ actual choices. Thus, neither difficulty in learning noisy preferences nor motivation appear to be major factors in how people choose for others. However, participants were much less consistent with their recipients’ preferences when choosing for them. Surrogates incorporated their own preferences and tried to maximize expected value. Mouse-tracking results imply that the recipient’s preferences affect the surrogate’s decision later in the choice process when choosing (vs. predicting).
Gaze-informed modeling of preference learning and prediction. (2019). Journal of Neuroscience, Psychology, and Economics.
Abstract: Learning other people’s preferences is a basic skill required to function effectively in society. However, the process underlying this behavior has been left largely unstudied. Here we aimed to characterize this process, using eye-tracking and computational modeling to study people while they estimated another person’s film preferences. In the first half of the study, subjects received immediate feedback after their guess, whereas in the second half, subjects were presented with four random first-half outcomes to aid them with their current estimation. From a variety of learning models, we identified two that best fit subjects’ behavior and eye movements: k-nearest neighbor and beauty contest. These results indicate that although some people attempt to form a high- dimensional representation of other people’s preferences, others simply go with the average opinion. These strategies can be distinguished by looking at a person’s eye movements. The results also demonstrate subjects’ ability to appropriately weight feedback in their estimates.