To catch a liar: the effects of truthful and deceptive testimony on inferential learning

Abstract

Much of what people learn is based on the testimony of others, but not all testimony is helpful. This study explores how people deceive and how they deal with deceptive information in the context of a conceptual learning task. Participants play a game in which a learner infers the location of a rectangle based on the testimony of an informant, who is either helpful or deceptive. We investigate the behavior of both informants and learners in this scenario. On the informant level, we demonstrate that people provide different information depending on whether they are helpful or deceptive. Although deceptive informants do lie outright, they more often opt to mislead. From the learner's perspective, we show that people do choose to verify information but no more often when the informant is deceptive. Despite this, we also find that learners are capable of accurately identifying who is deceptive and who is helpful. We conclude by examining common strategies used in the two conditions and their implications in real-world settings.