The perils of methods that detect lies 70% of the time: A reply to Ben-Shakhar and Verschuere (2024)

Abstract

Ben-Shakhar and Verschuere (2024) commented on Brennen and Magnussen's (2023) review of the literature on lie detection. They argued that the review's conclusion that such methods are not ready for forensic use is too pessimistic and that they already are in use worldwide. While we agree that many jurisdictions do in fact use such methods, we see this as cause for alarm rather than confirmation of their utility. It is documented that several lie detection methods distinguish to a statistically significant degree between deceitful and truthful statements, but they also have substantial error rates. This means that applying them at the level of the individual statement will frequently be misleading, making them unsuited to guiding police investigations.