Cognitive mechanisms for estimating others' preferences

Abstract

Predicting others’ preferences is considered essential for social life. Recent studies have proposed computational models, such as Bayesian inversion, that allow us to infer the preferences of others by observing their choices. However, these models don't always work when we try to predict others' preferences for novel objects, or when observations are limited, which is often the case in daily life. In this study, we investigated the cognitive mechanisms underlying such inferences by conducting an experiment where participants in pairs inferred their partner's preferences for novel face images. The results showed that they were able to infer the partner’s preferences with a small number of samples, suggesting that they exploited a sparse representation between the faces and the preferences of many people to make inferences. These findings indicate the heuristics we apply to make reasonably accurate inferences while reducing computational complexity in social situations.