Detecting deception in event narratives: the analysis of response latencies and error rates to unexpected questions.
- 1University of Padua
Abstract
Deception is a central phenomenon in the study of human behaviour, and in forensic and investigative contexts it assumes significance due to its potential consequences. Although the unexpected questions technique has shown promising results, its application to the narration of a false autobiographical event remains underexplored. This study aims to fill this gap using the unexpected questions technique in an investigative interview setting. A total of 60 participants were randomly divided into two groups: the honest group was asked to describe their most recent day trip, while the deceptive group was asked to memorise a narrative describing a day trip. Both groups underwent an audio-recorded interview containing control, expected and unexpected questions. The results showed that both groups exhibited longer response latencies and higher error and inaccuracy rates in response to unexpected questions, but these indices were significantly higher in the deceptive group (p = .022 for response latency, uncorrected for multiple comparisons; p < .001 for error and inaccuracy rates). The findings suggest that longer response times and higher error and inaccuracy rates during unexpected questions may reflect not only the difficulty of the questions, but also additional cognitive demands faced by deceptive participants. Furthermore, these indicators were used to classify a subject as deceptive or honest using machine-learning algorithms. Within the sample, the model showed excellent cross-validated discriminatory performance (AUC = 0.967), suggesting that these indices may have diagnostic potential worthy of further investigation.
Related
- Exploring asymmetry of electrical compound action potential latencies in sequential cochlear implants as a possible effect of the lateral olivocochlear bundle.
- Narratives of Aging and the Care Home in South Africa: Insights from Antjie Krog's Blood's Inner Rhyme (2025).
- The artifice of seasonality in Japan: Discursive narratives of temporal value
- Model Context Protocol: The unexpected catalyst of a bioinformatics interoperability revolution
- Unexpected detrimental role of plasmacytoid dendritic cells during viral infections of the lungs
BibTeX
@misc{rosa2026detecting,
title = {{Detecting deception in event narratives: the analysis of response latencies and error rates to unexpected questions.}},
author = {M. V. Rosa and G. Melis and G. Sartori and A. Zangrossi},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.29.762734},
url = {https://www.biorxiv.org/content/10.64898/2026.10.29.762734v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Detecting deception in event narratives: the analysis of response latencies and error rates to unexpected questions.
AU - M. V. Rosa
AU - G. Melis
AU - G. Sartori
AU - A. Zangrossi
PY - 2026
DA - 2026/10/29
PB - bioRxiv
DO - 10.64898/2026.10.29.762734
UR - https://www.biorxiv.org/content/10.64898/2026.10.29.762734v1
AB - Deception is a central phenomenon in the study of human behaviour, and in forensic and investigative contexts it assumes significance due to its potential consequences. Although the unexpected questions technique has shown promising results, its application to the narration of a false autobiographical event remains underexplored. This study aims to fill this gap using the unexpected questions technique in an investigative interview setting. A total of 60 participants were randomly divided into two groups: the honest group was asked to describe their most recent day trip, while the deceptive group was asked to memorise a narrative describing a day trip. Both groups underwent an audio-recorded interview containing control, expected and unexpected questions. The results showed that both groups exhibited longer response latencies and higher error and inaccuracy rates in response to unexpected questions, but these indices were significantly higher in the deceptive group (p = .022 for response latency, uncorrected for multiple comparisons; p < .001 for error and inaccuracy rates). The findings suggest that longer response times and higher error and inaccuracy rates during unexpected questions may reflect not only the difficulty of the questions, but also additional cognitive demands faced by deceptive participants. Furthermore, these indicators were used to classify a subject as deceptive or honest using machine-learning algorithms. Within the sample, the model showed excellent cross-validated discriminatory performance (AUC = 0.967), suggesting that these indices may have diagnostic potential worthy of further investigation.
ER -