An experimental test of epistemic vigilance: Competitive incentives increase dishonesty and reduce social influence.

Abstract

Cultural evolutionary theory has shown that social learning is adaptive across a broad range of conditions. While existing theory can account for why some social information is ignored, humans frequently under-utilise beneficial social information in experimental settings. One account of this is epistemic vigilance, whereby individuals avoid social information that is likely to be untrustworthy, though few experiments have directly tested this. We addressed this using a two-player online experiment where participants completed the same task in series. Player one provided social information for player two in the form of freely offered advice or their actual answer (termed "spying"). We manipulated the payoff structure of the task such that it had either a cooperative, competitive, or neutral incentive. As predicted, we found that under a competitive payoff structure: (i) player one was more likely to provide dishonest advice; and (ii) player two reduced their use of social information. Also, (iii) spied information was more influential than advice, and (iv) player two chose to spy rather than receive advice when offered the choice. Unexpectedly, the ability to choose between advice and spied information increased social influence. Finally, exploratory analyses found that the most trusting participants preferred to receive advice, while the least trusting participants favoured receiving no social information at all. Overall, our experiment supports the hypothesis that humans both use and provide social information strategically in a manner consistent with epistemic vigilance.