08/14/2014 11:48 pm ET Updated Oct 14, 2014

Catching Liars

Adam Berry via Getty Images

From a distance of 20 feet, without your knowledge or consent, the surveillance cameras, microphones and physiological sensors apply computer algorithms to process your gait, facial expression, gestures, voice, posture, heart rate, blood pressure and skin temperature -- determining within seconds if you are lying or telling the truth.

"Diogenes," the lie-catching machine in the sci-fi screenplay that a friend and I wrote 20 years ago, made no mistakes. And its mere existence changed the way people interacted: Bargainers would no longer agree to meet in person, juries became superfluous, salesmen were forced to be honest in their claims, and suspicions of marital infidelity were resolved in an instant, for a fee.

In real life we aren't quite there yet, but our recent research on truth and falsehood is getting closer than I ever expected it would. I've spent nearly 40 years studying how changes in demeanor (face, body and voice) might betray a lie. I focused on serious lies in which life, freedom, reputation, or the continuation of an important relationship were at stake, rather than the white lies of everyday life such as politeness, flattery or exaggeration. The original impetus was to help doctors evaluate whether psychiatric inpatients claiming to no longer feel depressed were lying so that they could commit suicide free of the hospital's supervision. Responding to interest from law-enforcement and intelligence agencies, my associate (the late professor Maureen O'Sullivan) and I extended our focus to lies about taking money, or false claims about a strongly held political opinion. Since 9/11 our experiments have focused on the lies told by political extremists, in hope that our results will have relevance to anti-terrorism. We set the rewards for success and the punishment for failure as high as ethics committees permit.

We have not found the modern equivalent of Pinocchio's nose, nothing in face, body, voice, speech or physiology that is unique to lying and never present when someone is worried, thoughtful, cautious, perplexed or nervous. While some still pursue that goal (and a few claim to have found it), I doubt that such a silver bullet exists. Instead, our measurements of facial muscular movement, gestures, voice and speech uncover what I call "hot spots" -- signs that something is amiss, that the full story is not being told.

The typical hot spot is a momentary conflict between the words spoken and the sound of the voice, the gesture, or the facial expression. Just as important are very brief microexpressions that can flash across a person's face in one 25th of a second. A microexpression looks exactly like a normal facial expression, except it happens so quickly that most people don't see it. It always is a sign of a concealed emotion -- sometimes deliberately concealed, sometimes just a repressed emotion. Just as important is a slight edge in the voice that doesn't fit calm words.

In our experiments in which there are only two possibilities -- someone is either lying or telling the truth -- hot spots allow high accuracy in distinguishing one from the other. In real life hot spots occur for many reasons, such as remembering an argument at breakfast with your spouse, worrying about missing a flight, or annoyance at the screening process at an airport. Lying to conceal malicious intent or past wrongdoing is only one, and not the most frequent, reason that hot spots occur where terrorists might lurk.

Nevertheless, learning how to identify hot spots can be useful in figuring out where to probe further in an interview, or whom among the millions each day who wait in line in our airports to ask a few questions about the purpose of their trip. We are training law-enforcement and national-security officials, here and in England, to identify hot spots, emphasizing that they are not signs of lying, only signs that something might be amiss. People can learn to recognize the microexpressions in an hour, and the benefit lasts. We don't know how long it takes to learn to recognize the many other hot spots, who learns the most and the least, or when a refresher course is needed.

Another line of active research is trying to develop the modern equivalent of our sci-fi lie-catching machine, Diogenes. The hardware and software that identifies hot spots in real time isn't ready for prime time now, but it is progressing. Soon automated hot-spot detectors could evaluate facial expressions and bodily physiology instantly, from a distance. Before it is deployed as a substitute for a highly trained human observer, it is essential to determine whether it is as accurate as such an observer, and to ensure that, if it were to be used as an aid rather than a substitute, it doesn't distract and lower observer performance.

The ACLU has complained that recognizing hot spots leads to the apprehension of not just terrorists but wanted felons, illegal immigrants, and smugglers. I am afraid that there is no terrorist-specific hot spot, but I am not personally convinced that it is bad to identify others who might be breaking the law. Another criticism is the possibility that members of minority groups, especially those whose names or appearance suggest that they might be Arabs, may be more uncomfortable in places such as airports and, for that reason, may show up more often as suspicious. That is possible, and even warning about it during training may not be sufficient to avoid such mistakes. However, I favor deploying an independent organization to check periodically on the performance of those doing hot-spot detection, to make certain that they are not slipping into racial profiling.

Another concern is what happens to the information picked up by an automatic hot-spot detector. Suppose the heart-rate and blood-pressure readings strongly suggest that someone is on the verge of a heart attack. After a warning that an emergency room might be a smarter destination than an airplane for such a person, what will happen to this private medical information? Regulations need to be developed to insure that it is erased rather than secretly sold to employers and insurance companies.

People around the world are already using and teaching these new approaches to identifying people who might intend harm. There is no putting this genie back in the bottle. The issue is how to use these new methods wisely, cautiously, and with safeguards for privacy and civil liberties.

This post was updated from an article by Paul Ekman for the Washington Post in 2006.

Paul Ekman, a retired professor of psychology at the University of California San Francisco, has been studying facial expressions and gestures for more than 40 years. He is the author of many books. His most recent book, Moving Towards Global Compassion, is available as an e-book at

To receive updates on news regarding facial-recognition technology, "like" and follow the Paul Ekman Group on Facebook and Twitter.