New Lie Detection Software More Effective than Polygraph

Photo of author

( — December 14, 2015) — How many white lies a day do people use in order to save themselves from embarrassment or to spare somebody else from being embarrassed? What about lying while giving testimonies?

A new software being developed at the University of Michigan may put a stop to lying, because lying will easily be detected.

A polygraph test measure a person’s pulse or the rate of breathing in order to discover a lie, while this new software doesn’t have to do that. The new software considers words and gestures to decide whether a person is telling the truth or not. At the moment, it is still in the prototype phase and the way researchers have created the software, is by examining videos of court cases.

During the experimental phase the software recorded 75 percent accuracy in identification of deception when compared to trial outcomes, as opposed to human identification of deception being only 50 percent accurate.

This is how the process was carried out: 120 video clips of actual trials were studied and the researchers noted some common ways how people behave when they are lying. They presented several examples of typical gestures when people are lying, like moving their hands more, trying to look directly at the questioner, using vocal fillers very frequently (‘um’, for example) or trying to sound more sure of what they are saying, all the things that people who tell the truth don’t often use.

Rada Mihalcea, one of the authors of the study said, “In laboratory experiments, it’s difficult to create a setting that motivates people to truly lie. The stakes are not high enough. We can offer a reward if people can lie well – pay them to convince another person that something false is true. But in the real world there is true motivation to deceive”.

Since the videos include both defendants’ and witnesses’ testimonies, both sides can’t be right at the same time, so it is obvious that somebody must be lying.

In order to find out which person is lying, the videos are transcribed with the help of vocal fillers and then analyzed to determine how often certain words are used. After that, the researchers observe the gestures of their subjects and use a standard coding scheme to analyze different facial parts movements, namely head, brow, eyes, mouth and hands.

The system was found to be 75 percent accurate, 20 percent more accurate than human detection.

Mihalcea added, “People are poor lie detectors. This isn’t the kind of task we’re naturally good at. There are clues that humans give naturally when they are being deceptive, but we’re not playing close enough attention to pick them up. We’re not counting how many times a person says ‘I’ or looks up. We’re focusing on a higher level of communication”.

At the present phase, the researchers are trying to include other physiological parameters into the system, in order to get more accurate results, like heart and respiration rate and body temperature fluctuation. This will contribute to a better determination of deception together with the already present features. This new software may be very helpful to police and health professionals in determining the honesty of their subjects.