“Deception is present in all walks of life, from social interactions to matters of homeland security.”
Thus the paper in Brain and Behavior begins, and it’s not a lie. We lie all the time whether or not we admit it, because the alternative is frankly obnoxious.
But there are circumstances when fibbing is appropriate (“Your new haircut looks great”) and circumstances when it is not (“What, me, COVID?”). Now researchers at Tel Aviv University say they have invented a lie detection system based on telltale twitches in the cheek and forehead.
They say their smart sticker, which contains electrodes that can monitor and measure the activity of muscles and nerves, achieved an unmatched success rate of 73 percent, albeit in a small test group of 40 people. (Two were kicked out for “never lying.”)
The study was led by Prof. Yael Hanein and Prof. Dino Levy with Dr. Anastasia Shuster, Dr. Lilach Inzelberg, Dr. Uri Ossmy and Ph.D. candidate Liz Izakon.
Do we need this? Maybe not in the living room, but present lie detection systems such as polygraphs don’t actually work well, certainly not well enough to serve in courts of law, the researchers point out. Also, it’s a conceited fallacy that we “always” know when a child, colleague, significant other or dog is lying.
Note that they didn’t invent the core sticker technology, for which a company named X-trodes gets credit. But the TAU team gets credit for exploring its effectiveness in lie detection.
- New Study Proves Unsuspected Social Ability in Cats
- Why the Rockfish Can Live for Centuries and You Can’t
- Discovery Deep in the Sea Sheds Strange Light on How Our Brains Evolved
Liar, liar, face on fire
“Many studies have shown that it’s almost impossible for us to tell when someone is lying to us,” says co-author Levy, adding that even purported experts such as cops don’t do much better than anyone else. Also, polygraph technology is based on the assumption that lying will hike one’s pulse, respiration and cause other physiological twitches, which the yogis among us can control. Furthermore, the researchers point out, polygraphs aren’t all used the same – the questions may differ, as does the interpretation of the response.
True, it is not news that our faces can be read. “It was Darwin who first noted that some emotions are too great to be fully feigned or concealed,” the researchers write. Others have tried to develop lie detection along the expressive, rather than physiological, angle, but haven’t been able to overcome the technological constraints – their machinery is too invasive, loud, not sensitive enough, and so on.
But wait. Could the hallmark facial contortions that expose us as foul liars be culturally specific? The Tel Aviv University study was done only on Israelis, most of them women. Maybe technology designed to out mendacious Israelis wouldn’t work on Polynesians or Alaskans? Is there any work on globally applicable facial expressions when we lie?
Yes, but the methodology was crude and there are no answers yet, Levy says. Generally, there is ample evidence showing that some facial expressions are different across cultures, and some are similar.
“Currently, the algorithm works only for Israelis. For extending it to different cultures, we would need to test subjects from different cultures,” he acknowledges. Hoowever, adjusted for culture if necessary, the same principles should apply.
Why were so many of the test subjects female? No good reason. “The main experimenter was a woman and, therefore, it was easier for her to recruit women,” Levy answers. And he does not rule out that men and women “lie differently,” though they didn’t detect any gender differences. It will take a lot more study to detect gender differences, if any.
What facial contortions do their stickers detect? Some (Israeli) liars are outed by twitching cheek muscles when they lie, others by their eyebrows; some twitch a cheek, then an eyebrow, and so on. Which is interesting: one way dogs captured our hearts and souls is by developing muscles that enable them to wiggle their eyebrows appealingly, it has been found. Most wolves can’t do that.
‘What a cute baby’
Another important point is that the test of the new technology was a crude lie, not some social modification such as “Oh what a cute baby!” The researchers attached the electrode-laden stickers to the subjects’ cheek muscles near the lips, and to the muscles over the eyebrows. Participants sat in pairs facing one another. One wore headphones. The researchers would say a word – either “line” or “tree.” The headphone-wearer was supposed to repeat that word, or lie and say the other one.
The other participant had to judge whether the headphone-wearer was lying or not. Then they switched roles.
Gut sense got them nowhere: It usually doesn’t when lying is involved. But the stickers identified the lies at a high rate of 73 percent, the TAU researchers say.
But lies in actual life are usually a lot more sophisticated. Who says the system will detect the white lies, the social lies, the lies that one almost believes oneself, the dog’s lies, etc.? Why use a lie that rudimentary?
True, the lie was crude and simple and had only one word, but the researchers wanted a simple, controllable stimulus, Levy explains. They had planned to run another study with bigger lies and a longer response with more words, but then COVID arrived.
Also, big lies (“I am not a terrorist”) are not the same as little white lies (“Orange looks good on you”), in his opinion, because the consequences of getting caught are not the same. Therefore, he predicts, our emotional and facial reactions would be more intense when the lie is big. “But we do not aim to identify every kind of lie. We aim to detect big, important lies with consequences,” he sums up.
So what do we have here? A very early concept, a baby step toward detecting liars by their facial contortions alone. But, as Levy points out, physiology may remain part of lie detection technology.
“We’re not saying that adding sweat and heart-rate measurements won’t improve the prediction. I would say that there are probably many physiological measurements we can add that will improve prediction. For example, body temperature,” he says. “All we’re saying is that small, indistinguishable muscle movements can be used to detect lies with high accuracy. You can always improve it with more features.”
Theoretically, anyway, the more experiments are done with lying in all sorts of contexts and places, the more likely we are to discover specific muscles, facial areas, twitches and timing that are associated with lying, and can train computer systems to zoom in on these.
And if I am the type of person who, upon encountering any form of authority, feels guilty and sweats bullets? “The fact that you sweat irrespective of your innocence is true,” he answers. “But the idea is that we can take your baseline measurements to questions that we know are true, and then check for deviations from that. This is standard even in the [not so] good old polygraph.”