Back

Lie Detectors for Face Recognition

Meytlis, M.

2025-06-24 neuroscience
10.1101/2025.06.22.660938 bioRxiv
Show abstract

1Lie detection is important for government law enforcement. Current lie detection methods such as the polygraph test have been found to be unreliable (Meijer et al. 2017). New lie detection technology is currently arising that is based on fMRI; however, single subject tests have only been successful in detecting lies 88% of the time (Langleben et al., 2005; Wild, 2005). One of the main problems with most fMRI-based approaches is that they assume that various acts of deception involve common brain regions, (Ganis et al., 2003). In this work I propose a much more accurate fMRI lie detection method that does not make this assumption and is domain based. In my investigation, rather than trying to localize brain regions that are indicative of lying in general, I localize brain regions that indicate lying specifically about face recognition. In criminal investigations one frequently needs to establish familiar relations between a suspect, victim and/or witness. This type of information can be used as circumstantial evidence in a crime. In this work I propose to use fMRI to detect whether a suspect has any familiarity with an individual face. I find that activation in the left inferior frontal gyrus was a reliable discriminator for face familiarity.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.