An evidence-based imperative to videorecord eyewitness lineups.

DNA-exoneration cases reveal the unreliability of witnesses’ in-court behaviors; despite being mistaken, these witnesses delivered confident and compelling testimonies (Garrett, 2011). But testimonies are retrospective and subject to contamination. The legal system should focus instead on how the witness behaved at the time of the initial lineup. Unfortunately, few law enforcement agencies videorecord lineups, which makes this practice impossible. In a large-scale experiment, we videorecorded witness-participants (N = 1,496) completing lineups, scored them on numerous manifestations of recognition strength and fluency, and used machine learning analyses to determine whether these behaviors discriminated between accurate and inaccurate decisions. Not only did witness behaviors discriminate between accurate and inaccurate decisions, but they also improved classification performance over and above confidence and decision speed. In fact, the behavior of a witness during a lineup can be used to diagnose high-confidence mistaken identifications. These findings establish an evidence-based imperative to videorecord lineup procedures. (PsycInfo Database Record (c) 2026 APA, all rights reserved)