Why Humans are Bad at Detecting Lies

Abstract

The volume of digital scams has been growing in recent years, resulting in significant financial losses and emotional damage. This is partially because humans are poor at detecting scams. A vital research challenge is to establish why people are ‘suboptimal lie detectors’, so to inform the development of tools for improvement. Over two studies (N=210), we show that participants’ suspicion is driven by two distinct informational sources: their own behaviour and statistical likelihood computations. Participants were more suspicious of others when they themselves lied and when they encountered unlikely events. Overreliance on one’s own behaviour led to worse discernment, while higher reliance on statistical likelihoods, on the other hand, led to better discernment. The findings have implications for the design of digital interfaces to improve fraud detection, for example by offering information that can help people assess relevant likelihoods.

Keywords: Suspicion, Honesty, Fraud, Judgement and Decision-Making