The basis of the polygraph or lie detector test was researched by William Marston, along with his wife and their friend. This is a device that records some body responses like blood pressure (pulse), respiration rate, and electric conductivity on the skin. And it acts on the principle that when a person is lying, that person will perspire more, their heart rate and pulse might go up, their rate of breathing might change. In other words, the body responds when a person is lying. And those measurements don’t lie! So if there are changes in these metrics, that means the person being questioned is fibbing. This research was then commercialised in the form of a lie detector instrument and test that police officers and other law enforcement personnel have used for years to help them with their jobs.
Fun Fact: William Marston invented…. comic book heroine Wonder Woman. And what is her weapon? Why, it’s the lasso of truth!
How does the lie detector test work? The person taking the test is hooked up to a machine, and someone asks them questions – things they know to be right, at first, to establish a ‘baseline’ for that person’s physical responses. Then they will ask a bunch of questions and measure the body responses while the person answers these questions. The lie detector produces a graph or a line that is then interpreted by the person administering the test, such that spikes in these rates indicate that a person is lying.
Is this accurate? Apparently, not really! Experts argue that a change in breathing rate or blood pressure don’t necessarily correlate with deception. For example, a person might be upset or in a bad mood in any case, and we all know what that feels like! Our heart rates can increase, we may be so frustrated that we cry, our breathing rate may change. In addition, some people can control their reactions very well. That’s why while the polygraph is used as an indicator, its results are not usually allowed to be used during a court trial. It simply is not accurate enough.
The principle that the body responds differently when telling the truth and when lying, has been taken a step further in this age of AI. AI has been made to analyse the faces of a number of people who are known to be telling the truth and who are lying. And compares and analyses facial expressions and mini expressions and their changes, and studies the relationships between these changes. One of the first such systems has been shown to be about 80% accurate in terms of telling who is lying. But that’s not amazing. That means that it is wrong 20% of the time! It’s wrong for 20 people out of every 100! That’s pretty dangerous if it is used to exclude people from getting justice or being eligible for certain jobs, or to stop them from entering countries.
Are new and more reliable systems going to be developed? We hope so, but in the meantime, take those lie detector reading with a pinch of salt!
This article was adapted from an article on the MIT Technology Review by Sunaina Murthy