Detecting deception in event narratives: the analysis of response latencies and error rates to unexpected questions.

In a randomized investigative interview with 60 participants, deceptive narrators showed significantly longer response latencies and higher error and inaccuracy rates to unexpected questions (AUC = 0.967).

• Why it matters: Unexpected questions may help detect deception, but it is unclear how well this approach works when people must narrate a false autobiographical event.
• What they did: Participants were split into honest and deceptive groups, then interviewed with control, expected, and unexpected questions while audio was recorded and response timing and error/inaccuracy were measured.
• The result: Unexpected questions increased latency and error/inaccuracy in both groups, with larger effects for deception (p = .022; p < .001), and machine learning classified honesty versus deception with AUC = 0.967.

AI summary of the abstract · may contain errors · not medical advice · read the paper.