WEBVTT 00:00:00.000 --> 00:00:04.600 You're watching TV, you're 45 minutes into a tense crime drama, and the cops have their 00:00:04.600 --> 00:00:08.560 primary suspect wired up to a lie detector machine. 00:00:08.560 --> 00:00:17.360 Finally the truth, or maybe not. In fiction and on the news, polygraph tests are often depicted as basically infallible, 00:00:17.360 --> 00:00:21.480 unless you're a super genius serial killer with the ability to control your heartbeat 00:00:21.480 --> 00:00:27.880 in every drop of sweat. But in real life, there are plenty of cases where relatively normal, non-serial killers 00:00:27.880 --> 00:00:34.240 have beaten the test. On top of that, the polygraph isn't even measuring lies strictly speaking. 00:00:34.240 --> 00:00:39.440 Lie detector is an effective, but actually misleading, little bit of branding. 00:00:39.440 --> 00:00:44.500 What a polygraph test actually measures are a number of physiological responses that typically 00:00:44.500 --> 00:00:50.640 correlate with stress, respiration, perspiration, heart-rateration, heart rate, and blood pressure. 00:00:50.640 --> 00:00:55.360 Now lying is typically stressful, even when we don't feel guilty because deception takes 00:00:55.360 --> 00:01:01.960 effort and there are consequences to getting caught. To take all those readings, a polygraph machine has several key components. 00:01:01.960 --> 00:01:06.080 First is the pneumograph, a device that measures a subject's breathing. 00:01:06.080 --> 00:01:12.500 There are two major variants of pneumograph. The first uses two sets of rubber tubes, one that circles the chest and one that circles 00:01:12.500 --> 00:01:17.880 the abdomen. These loops stretch when the subject breathes, thereby recording the rate of respiration. 00:01:17.880 --> 00:01:21.600 The second type of pneumograph uses electrical impedance to measure breathing. 00:01:21.600 --> 00:01:26.160 So the test administrator places two pairs of electrodes on either side of the chest. 00:01:26.160 --> 00:01:33.520 Because nipples never lie, I guess? They then run a high-frequency, low-amplitude current through the subject's chest cavity. 00:01:33.520 --> 00:01:38.160 And because air is a better electrical insulator than human flesh, the chest blocks more of 00:01:38.160 --> 00:01:44.800 the current as the lungs expand. To measure how sweaty the subject gets when getting grilled, a mouse is placed on a desk 00:01:44.800 --> 00:01:50.200 in front of them and they game. No, I'm just kidding. A galvanograph is attached to their fingertips. 00:01:50.200 --> 00:01:54.820 This uses electrical current to measure the conductivity of the skin, which increases 00:01:54.820 --> 00:02:00.040 when damp. Next, there's the cardiospygmograph, which uses an inflatable blood pressure cuff to 00:02:00.040 --> 00:02:04.160 amplify the sound of the subject's blood flow, and therefore heartbeat, recording blood 00:02:04.160 --> 00:02:11.080 pressure and heart rate. Finally, there's the data acquisition system, or DAS, which collects data from each of these 00:02:11.080 --> 00:02:19.040 subsystems and converts it into a readable form. A bunch of squiggly lines on a paper readout, or, well, more likely these days, a screen. 00:02:19.040 --> 00:02:23.520 Modern DAS can even assist administrators through simultaneous computer analysis of 00:02:23.520 --> 00:02:29.340 the results. Now, that's a lot of delicious, quantifiable data that's being collected, but does it 00:02:29.340 --> 00:02:32.660 really mean that we can measure lies and solve crime? 00:02:32.660 --> 00:02:37.460 One of the most difficult aspects of applied science is just how many potential confounding 00:02:37.460 --> 00:02:42.660 variables there are in the real world, with the messiest being the human element. 00:02:42.660 --> 00:02:45.820 Polygraphs measure breathing, sweating, heart rate, and blood pressure. 00:02:45.820 --> 00:02:50.540 All physiological symptoms of stress, but there are tons of causes of stress that have 00:02:50.560 --> 00:02:55.320 nothing to do with lying to the authorities because you just murdered your spouse. 00:02:55.320 --> 00:03:00.260 For instance, having the authorities accuse you of murdering your spouse might be incredibly 00:03:00.260 --> 00:03:05.860 stressful. For that matter, so is jogging, drinking coffee, watching a scary movie, performing on stage, 00:03:05.860 --> 00:03:08.980 falling down the stairs, getting married, or being punched. 00:03:08.980 --> 00:03:14.220 And the human body doesn't really differentiate much between these various sources of stress. 00:03:14.220 --> 00:03:18.440 Any of them can turn you into a sweaty, stuttering mess if they're intense enough. 00:03:18.440 --> 00:03:22.980 For blood pressure readings, there is a known problem called white coat hypertension, where 00:03:22.980 --> 00:03:27.280 some patients have consistently higher blood pressure readings whenever they're tested 00:03:27.280 --> 00:03:31.920 in a medical setting like a hospital or doctor's office compared to when they're tested at 00:03:31.920 --> 00:03:35.640 home or with those automatic tests at the grocery store. 00:03:35.640 --> 00:03:39.400 This is typically because the patient finds going to the doctor stressful. 00:03:39.400 --> 00:03:43.400 Many innocent people then, even those who have never had a bad interaction with the 00:03:43.400 --> 00:03:48.000 police, find interacting with law enforcement scary, and those people are more likely to 00:03:48.000 --> 00:03:54.880 fail a polygraph. Another problem is that polygraph interpretation is, well, interpretation. 00:03:54.880 --> 00:03:58.660 No two people respond exactly the same to all stressors. 00:03:58.660 --> 00:04:06.480 And there's no hard line between pass and fail. Rather, administrators typically ask a series of irrelevant but uncomfortable questions to 00:04:06.480 --> 00:04:11.400 establish a baseline, and then they compare that to later, more loaded questions about 00:04:11.400 --> 00:04:16.200 the actual subject of the investigation. But this leaves a major opening for subversion. 00:04:16.200 --> 00:04:20.880 Others who know how the test works may intentionally speed up their breathing during the baseline 00:04:20.880 --> 00:04:25.320 questions or may slow their breathing during relevant questions. 00:04:25.320 --> 00:04:30.160 Some may take anti-anxiety medication or drink alcohol before the test to mask their stress, 00:04:30.160 --> 00:04:33.880 while others may simply walk in confident and retain their cool. 00:04:33.880 --> 00:04:39.120 Aldrich Ames, a chronic alcoholic and a spy for the Soviet Union, was given two polygraphs 00:04:39.120 --> 00:04:42.200 during his time at the CIA and passed both of them. 00:04:42.200 --> 00:04:46.440 The key advice given to him by his KGB handlers? Just relax, comrade. 00:04:46.440 --> 00:04:51.000 A major assumption of the test is that these reactions are involuntary, and to an extent 00:04:51.000 --> 00:04:56.320 they are. But almost everyone outside of the iron lung can control their breathing at least. 00:04:56.320 --> 00:05:00.360 And then from there, it takes a trivial amount of practice to intentionally speed up or slow 00:05:00.360 --> 00:05:03.840 down your heart rate, which also affects your blood pressure. 00:05:03.840 --> 00:05:08.800 In some ways, the innocent are the least prepared to take a polygraph test as habitual 00:05:08.800 --> 00:05:12.200 liars are less likely to show signs of stress when lying. 00:05:12.200 --> 00:05:16.120 What this all means, then, is that polygraph tests aren't terribly reliable, better than 00:05:16.120 --> 00:05:23.240 chance perhaps, but seriously flawed. There's no firm scientific basis for using these tools to determine guilt or innocence, 00:05:23.240 --> 00:05:27.280 and courts in most jurisdictions have ruled them inadmissible as evidence. 00:05:27.280 --> 00:05:31.160 That hasn't stopped US law enforcement from using polygraphs as part of their hiring process, 00:05:31.160 --> 00:05:36.240 though, nor has it stopped some researchers from trying to find a new 21st century equivalent 00:05:36.240 --> 00:05:41.920 to the polygraph, either through MRI scans of the brain, micro-expression tracking electrodes, 00:05:41.960 --> 00:05:46.120 eye tracking cameras, or even facial recognition technologies that's adapted to spot a lie 00:05:46.120 --> 00:05:52.160 at a distance via security cameras. It remains to be seen, though, whether these new lie detectors will be any more accurate 00:05:52.160 --> 00:05:55.920 than the old ones. If you enjoyed this video, do a polygraph. 00:05:55.920 --> 00:05:59.600 Prove it. No seriously, hit the like button, and if you didn't, you can dislike. 00:05:59.600 --> 00:06:03.960 And if you have any feedback for us, leave a comment, maybe one with a favorite outlandish 00:06:03.960 --> 00:06:10.680 lie. Here's one from us. My heart beats twice a minute like a blue whale, and I haven't blinked once this entire 00:06:10.680 --> 00:06:13.080 episode. If you don't believe me, go back and check.