How Do Lie Detectors Work?
You’ve seen it in the movies. You’ve seen it on TV. To eschew an entire interrogation scene we just hook someone up to a bunch of machines that beep whenever they lie. But really movies like to sling around polygraphs and pretend we know what the squiggly lines mean. Then there’s all the weird truth serum shenanigans if we’re feeling like shaking things up from the usual polygraph tests. But how does the sausage get made? How do lie detectors work?
Early Lie Detection
The history of lie detection technology is unsurprisingly quite long. As social creatures, we’ve probably been trying to figure out how to catch each other in lies since Unga and Bunga realized they could lie. Heck, if crabs lie, then we definitely do.That considered, lie detection has historically been quite the oddity–we’ve been throwing things at walls to see what sticks for a while.
Given how fallible human memory and attention is, it’s safe to say that trusting humans even when they think they’re not lying is hard enough. Here’s a quick memory test and attention test you’ll probably see in every basic psychology class. If you succeed (or you already know the test), more power to you, but do know that being aware it’s a test will prime you to look for the “catch.” Don’t beat yourself up for falling for it if you do–that’s what they’re designed to do.
Some of the oldest documented lie-detectors date back to 1000 BC, and can be bought from your local grocery store. Way back in the feudal-Chinese-era, suspected liars were to stuff their mouths full of rice and made to spit it out later. Dry rice meant you were lying; anxiety was thought to reduce salivation and thus keep the rice from getting wet. It’s born of how our mouths feel kind of dry when we’re anxious.
There was also the whole divine judgement or trials by ordeal thing. Also known as “do an unpleasant thing and depending on the outcome you might be a liar.” You know, stuff like throwing women into water to see if they would float during the Salem Witch Trials. Those didn’t work out so well, and were mostly used as a way to persecute people.
You probably see the word “phrenology” and reflexively think “yikes.” Quick go around, it was that thing where the shape of your head was said to determine your mental faculties or character traits. It’s probably most well known in the cultural lexicon as the backwards way to baselessly “scientifically justify” racism. Upon its inception in 1870, Franz Joseph Gall tried to use the shapes of criminals’ skulls to determine whether or not an individual was lying. He even testified in court and traveled around to do demos.
This is probably the thing we hear about the most–as well as how inaccurate it is. The polygraph’s roots probably date back to a guy named Erasistratus in 300-250 BC. He wanted to detect lies based on heart rate.
But the earliest things we’d actually call a polygraph go back to 1881, when Cesare Lombroso created “Lombroso’s Glove.” The idea was to measure blood pressure of the accused and chart it–something that would be developed upon through WWI by William M. Marston. By that time, the lie detector had evolved to also measure changes in breathing patterns–as well as blood pressure.
Shortly afterwards, the polygraph as we probably know it came to be, now with new galvanic skin response detection! Which is basically how bioelectrically reactive the skin is.
While widespread in its use through the late 1990s, you’ve probably figured out a huge issue with the polygraph. Your blood pressure goes up, your breathing changes, and your galvanic skin responses can change for reasons other than lying. They can change when you’re flat out stressed, for example. You can bet top dollar that people are going to get stressed when they’re asked all sorts of weird control questions to get a polygraph set up–even more so when asked about something like a crime.
So you can imagine the false positives there. Of which there were quite a few–polygraph failure rates can be above 15%, not too great when the bar for criminal conviction in America is “beyond a reasonable doubt” (we’re pulling the data from Americans).
It’s also pretty well known that people can cheat polygraphs. No surprise polygraphs aren’t considered admissible evidence in America.
Neuroscience & AI
You’ve probably heard very mixed things about lie detection technology. Namely, you’ve probably heard that technology might help just a little bit–but nowhere near enough to secure any kind of convictions you might have. Either that or some headline is telling you the world is going to end because lie detection just got foolproof.
By the 1980s (and the dawn of neuroscience), we figured out that we needed to step up our lie detection game. So we started trying stuff like PET scans (positron emission tomography) and the like. Here’s some research published in 2012 that’s basically a super long way of saying these scans had a failure rate below 1%. Though the failure rate has also been shown to be as high as 10 or even 25%.
We’ve also started turning to artificial intelligence, which is pretty good at predicting us. Heck, in 2012 AI was able to put together enough random data points about a teenager’s shopping habits to figure out she was pregnant before her dad did. But even those are flawed, the art dystopian headline we linked above discusses an eye-reading program that has a failure rate around 14%.
So the pretty widespread claims that “there’s no signature for when you’re lying” are pretty solid at this time, as much as we try. So for now it’s a pretty fuzzy process of maybes, but who knows where we might end up–since Elon Musk wants to put microchips inside brains.
Anyway, here are some liar logic puzzles. See if you’re better than a polygraph.