Eye tracking lie detection system 82.5% accurate
It has been proven that there are certain changes in behavior and minute facial expressions that signal when someone is telling a lie. In fact, there are expert interrogators who are trained to identify these biological signs of deceit in criminal investigations. Research at the University of Buffalo, NY has found that there might be an even more accurate way to detect lies using eye tracking technology to identify indicative changes in eye movements.
The research sought to find out whether there were visual cues related to eye movement that signal deceit and whether or not a machine could detect them. What they found was that both were true. In fact, the automated system they developed can detect lies with 82.5% accuracy, which is extraordinary compared to the 65% average accuracy of human interrogators.
The experiment recorded eye movement of 40 subjects during regular conversation to create a baseline of “natural” behavior. Then these movements were compared to eye movements during the moment the subjects delivered the lie. The automated system was programmed to detect dramatic deviations from the baseline to predict whether the subject was lying.
The results of this study went hand-in-hand with findings of previous studies, which noted subjects’ eye contact changes when telling a high-stakes lie. Future research will include a larger sample size, as well as the incorporation of other biometrics to analyze body language in addition to eye movements. The end result could potentially be a computer-based system that can learn about human behavior well enough to assist human interrogators in challenging cases with higher accuracy.
- Improved Eye Tracking Algorithm to Be Used for Driver Fatigue Detection System
- The Engineer Behind the Eye Tracking Robotic System