Truth Serums and False Confessions

[Music] How do you get information from someone who wants to keep it from you? Somewhere locked inside their brain could be the truth about a crime, or the plan for a terrorist attack, or the password to a bank account, or nuclear codes. To get information, we often resort to violence, but that often doesn't work and can cause a person to close off even more. But that's not to say there aren't other ways of getting people to talk, or even of getting their brains to talk for them.

In this episode, I'm going to be injected with a truth serum. I'm going to coerce people into giving false confessions, and I'm going to try to hide the truth from a mind-reading machine. What's the best way to get the truth or detect a lie? Let's see if anybody can dig out the secrets I've got locked up in my head. [Music]

It's a classic scene in spy movies: someone captures someone they want information from, they inject them with truth serum, and the person spills all their secrets. The serum used in almost all of these movies is sodium pentathol, which is an actual drug that inhibits brain activity. It was first used as a painkiller but was found to work better as an anti-anxiety drug. It dulls your thoughts, making it harder to perform cognitive tasks, including the task of making up a lie.

The use of truth serums began in 1915 when Dr. Robert House, a physician in Texas, realized that scopolamine, a drug given to women during childbirth, had the effect of getting them to talk without reservation. Another drug, midazolam, has been used to treat Iraq War burn victims who suffer from PTSD. Midazolam helps reduce their inhibitions, allowing them to talk more freely about their experiences. Midazolam is also one of the drugs used in lethal injections, including several botched ones. I arranged to be injected with midazolam to see if it could make me spill my guts. [Music]

This is the midazolam, the medicine we are going to use. This is enough to put an elephant to sleep, so we'll have to make it very, very, very dilute and then give you very, very small amounts. Delicious? Please, yes. You would be comfortable and happy. It's euphoric; it will feel very cool. But I also want to keep some secrets—that is to remain to be determined.

Okay, Michael, I will introduce you to Dr. Pavlov. Hi, nice to meet you. How are you doing? Good. Dr. Pavlov is a police psychologist and an expert in interrogation. Have you started the drip yet? We would like to start with the amount of one milligram.

Dr. Pavlov's objective was to get me to admit to certain information: one, that my sister's name is Melissa, and two, that my job is hosting Minefield and Vsauce. My objective? Not to admit the truth. Are you feeling more relaxed in your body? I'm feeling warm all over. It's amazing how quickly this came into effect.

Let me just ask you a couple of basic questions. I'd love to. Okay, where are you? What town? What's the city? How old are you? 31. Do you have any siblings? Yeah. How old is your sister? Melissa? That—that's not her name. What do you do for a living? I am a choreographer by trade.

What might be something I'm familiar with that you did? Have you heard of, um, The Lion King? Yes, on Broadway. I have. I think he needs an extra dose. Absolutely. That means I got it right—they are too sharp. How much have I had total so far? Two milligrams. Question for you: how do you feel now? I feel like in order to feel like this, I would have had to have had six drinks.

Tell me about Minefield. Minefield? Yeah, my Minefield—they're terrible. They're places that we've got mines still littered throughout the land: Cambodia, Vietnam. You know, I think you misinterpreted my question: the YouTube show Minefield that you host. Tell me about that. Scientific; I am somewhat familiar with it. Did he get another dose?

Okay, we're just going to get you even more relaxed. Tell me your occupation. I'm a choreographer. And siblings? I have a sibling in Colorado. What's your favorite food? Rotisserie Greek döner. Does her sister Melissa like that too? She'll eat the meat by itself.

Trying to get Michael to admit that he has a sister Melissa was no easy feat. Luckily, he was anesthetized enough and did admit that he had a sister named Melissa. You ever think about being an uncle or a father yourself? Do you want children? Oh, I would love to have children. Yeah, my wife and I are trying, but I'm always doing stuff like this, you know? Yeah, it's like, "Hey honey, do you want to make a little baby?" and I'm like, "I gotta go get interrogated and pumped full of midazolam."

Are you going to give him any more? Yeah, because he's not sedated enough. I could do another one; this would probably be the last I want to give. Yes, do most people at this dose fall asleep? Oh yeah. How you doing, Michael? Go ahead. Feeling a little bit more tired now? Yeah, and heavy.

Who’s the president? President—uh, president—Trump. President Trump. Where are we? What city are we in? Pasadena. Which show do you like to do more, Minefield or Vsauce? Which one do you enjoy more? Vsauce. Yeah, why? Because they can be whatever I want them to be and there's no executive in control.

By the way, my eyes were closed, weren't they? And I didn't realize that you were still a real person asking me questions. I think had it not been for the drug, there's no way he would have admitted anything. Before, you were pretending to be a choreographer, weren't you? No, it's the drugs, but I really am a choreographer.

After a few minutes, he got right back on track. He's very strong-willed and really believes in what he's doing, and was able to maintain the charade. Okay, you're done. How do you feel? Uh-oh, I feel ready to drive—just kidding. While midazolam lowered my inhibitions and my ability to lie consistently, truth serums raise ethical concerns about one's right against self-incrimination. Furthermore, large-scale studies have shown that truth serums can cause people to reconstruct and fabricate memories, so truth serums are not very reliable.

But what about the techniques used by the police? [Music] I'm going to tear you apart if you don't give us some answers, Michael. Let me take a shot at this. Look, we're all friends here. If I was in your shoes, I'd probably do the same thing. You're being too easy on this scum; your fingerprints are all over the crime scene. You make me sick, Michael. If you get them too scared, they'll shut down. We only want to scare him enough.

That's right, and after I do that, he's going to give you relief, and then we'll do that over and over and over again. You're going to be taken on an emotional roller coaster that will deplete the finite cognitive resources you'll need to keep withholding information. All right, I confess. It really does work. You see, at one moment I'm terrified, but then I'm comforted. The constant shifting from one emotion to another fogs my critical thinking until I might confess to a crime I didn't even commit. It’s called the fear-then-relief response.

The good cop-bad cop method of interrogation evolved out of a psychological manipulation procedure called the Reid technique, developed in the 1950s and used broadly by police departments since the 1970s. Now, under the Reid technique, interrogators first assess whether a suspect is withholding the truth. If they think the suspect is lying, interrogators then move to the second stage in which they seek a confession. Now, critics of the Reid technique say that the first phase isn't reliable enough and that often someone who is completely innocent makes it to the second stage where it's assumed they’re guilty, and police begin seeking a confession, and they often get it.

But why would anyone falsely confess? Well, to find out, I decided to learn how to make someone confess to a crime they didn't commit. Dr. Melissa Roussano is an expert in investigative research, including the process of interrogation and confession. Tell me about false confessions and why people make them.

In a nutshell, people confess to crimes that they didn't commit because they come to believe that it's in their best interest to falsely confess. And that's really counterintuitive, right? Because confessing is the worst thing to do, right? Like, that's what you need to avoid at all costs. Yes, when you're not in the situation of being convinced that it's actually your best option. But that's kind of what I want to do today; I want to see how easy it is to get a false confession from someone. How do I do that?

One way is minimization techniques. You've got to get them to believe that their best option out of this situation to make this end is to confess. You can say, "Look, I really think it's in your best interest to cooperate." There are other techniques as well, like maximization, right? Okay, so they have to believe that if they don't confess, things are going to be worse—that the punishment will likely be more severe, and so their best option in the moment is to confess to you.

One other technique is befriending them, where you're lulling the suspect into a false sense of security to trust you. Now, what about making a deal? So police officers are not allowed to make explicit offers or leniency or a deal, so you cannot make any promises, but you can say, "Look, I really think it's in your best interest to cooperate." You don't have to explicitly say something for the message to be communicated.

With all these tools in mind, I was ready to try to get a false confession. All right, our subjects answered our online ad offering $75 to test a new logic exam for an educational institution. They didn't realize that the institution was fake and the other participants taking the test were actually actors working for us.

It's a pretty standard release form. Before the test, participants signed a document stating that they would be under breach of contract if they cheated on the test. When you work on the individual questions, it's very important that you work alone. Don't talk about what you're doing; don't share answers. Good luck; see you soon!

The subjects didn't know that there were several hidden cameras around and that we were watching everything unfold from another room. During the testing, the participants and our actors did not share information. At no time did cheating take place. I'm going to collect these and put them in the database, and I'll be right back to start the next phase. Okay guys, thank you, thank you!

I wore an earpiece so that Melissa could guide me during the entire process. A few minutes later, it was time to set up a false accusation. So, we actually might have a problem. Um, I'll need to speak to each of you separately. So, James, could you come with me just outside? I'll take these now; thank you. And, um, Nicole, you can just wait right here. Awesome, thank you!

You gotta give him enough time to plausibly be talking to and calling the director. Hi, Nicole. Hello! You and James both had the same wrong answer for the triangle problem. It looks to me like there was information sharing, like you guys spoke to one another about the answer on that problem.

But we didn't talk about it during the individual problems! I kept my eyes on my paper, because I knew that would probably throw off the results. If I look, if I just write out here, like, I admit that I shared answers—I don't remember talking to him at all about like what we're doing.

First, I tried to minimize the consequences of confessing. I really think it's in your best interest to sign this, stating that you admit to sharing the information. I honestly believe that that is—but I did not—look, I—I don't know what's going on here. I really—that's—I don't know; I came here to help.

I understand, and I never said, like, "Hey there's Megan!" Please let me finish because I—I don't appreciate like over-talking and stuff. I was too aggressive at first; I had a lot to learn. Next, I tried to maximize the consequences of not confessing. What we're looking at is possible breach of contract here. If he comes down here, he's going to be a lot more upset.

If the director has to come down here, he's going to be even more upset and annoyed than he already is. It did not happen. Okay, and I put that on my dad and I went down around my neck. So once it was me against her family, I lost my chance to get a false confession. I had to become the suspect's friend.

I also understand you're a nice guy, and like sharing information or helping someone with the problem is the nice thing to do. It's what we normally do, but I didn't share the information, so I can't sign that document. But I failed to convince this guy to be my friend. I struck out with all these test subjects.

But after my failed attempts, I was ready to put everything I'd learned into my next encounter. Is there a strategy I should use? I think that you should proceed with making sure that you're not overly confrontational and making sure that she sees you as someone who she can trust your advice. And so I would proceed with that kind of approach.

Okay, going in. Thanks for waiting. So I think we do have a problem. I was looking through your individual questionnaires and you guys both had the same wrong answer on the triangle problem. Really? So it was kind of this weird like, okay what's going on? Like statistically, what are the chances that two people are going to have these very unique answers?

So I called the project director to see what to do, because if you know this is a sign that information was shared during the individual section, that's a major problem for the study, right? Yeah! Look at each other's stuff! That was like—we were like on like, we didn't ask each other anything, we didn't like, you know, we didn't talk to each other the whole entire time.

And so I don't know how we got the same answer, but I'm like 100% sure that that was like the answer that I came up with. Deyja was resolute about her innocence, but I was about to maximize the consequences of cheating should she not cooperate. And that's a really big deal because through this grant, we need to have the integrity of the data.

We need to always obey the rules of the experiment, and it could even be a breach of the contract that you guys signed when you came to do this study, okay? Deyja now knew what was at stake. My job was to convince her that I had her best interest at heart.

I have to call the director back, right? And I can either tell him I don't know what's going on, we can't proceed, like you need to come down here! I don't know who he's going to involve. The easier option is to just document that, um, information was shared. Okay, so if I write out, um, you know, I admit that I shared, uh, the answers—is that crazy to say that we didn't share answers though? Because I'm like—I don't know what he's saying or if he said that he looked at my answer or—but we didn't.

Deyja wasn't quite convinced, so the next tactic was to minimize the impact of confessing. I don't know exactly how it happened, I just know that we've got to figure out the best way forward. This is so weird; I don't know exactly how it happened. It looks like information sharing! If you just tell us that that happened, then what we can do is I can call the director back and say that you guys are cooperating and I know him! That’s the best thing.

Okay, I mean, it's not true, but outside? Okay, if you sign and date it, I can go and call him back and I'll try to get you out of here. Thank you very much, Deyja! I will be—I’ll be right back. Thanks for your patience, so you did an excellent job of communicating to her that it was in her best interest.

The best way out of this situation was to confess, but it feels so bad! I mean she knows the seriousness; she knows that it's a lie. I mean, I didn't cheat! This is not cheap; those were all my answers. [Music] Hi! Okay, this concludes the study. I’m going to give you a debrief now to tell you about what we're studying.

So first of all, you're not in trouble; the other guy here works for us. This is not a study of logic problems; it’s a study of interrogation and false confessions. You falsely confessed today. I see that. How do you feel? I feel mad at myself because I'm usually the one who like, you know, just sticks to what you know is right. I tried to at least.

So what happened today? It was just scary, I guess, because it's an environment I've never been in before. And then just like all the pressure—it was just like, oh, well, I guess I should just sign it and say that I did it. Despite knowing she was innocent and knowing that she was admitting to breaching a contract, Deyja signed a confession that in a true criminal case could be used as evidence in a trial against her.

Imagine if there were consequences and this was a real crime! How easy it is to get someone to confess to something without even punching them or waterboarding them. Just being nice and telling them that it’s in their best interest. Yeah, it's messed up, actually. Isn’t it? It’s really crazy, because there are a lot of people in prison for crimes they didn't do.

The Innocence Project, an organization dedicated to uncovering miscarriages of justice, estimates that 20,000 people are currently falsely imprisoned in the U.S. The number one cause is faulty eyewitness testimony, and the number two cause is false confession. Current interrogation techniques are unreliable; they can fail to produce truthful information from suspects who are good at lying, and perhaps worse, they can manipulate innocent suspects to confess to crimes they've never committed.

But neuroscientists may have discovered a method for extracting the truth that's impervious to good liars and bad interrogators. Dr. Peter Rosenfeld and his team at Northwestern University, including PhD candidate Ann Ward, have developed a high-tech method for lie detection. But you guys brought with you today a P300-based concealed information test. How does that work and how is it different than a polygraph or traditional verbal-only interrogation techniques?

Well, we're looking at physiological responses of the brain in response to information that's presented. And if the information is meaningful, like the murder weapon that a guilty person used, he will recognize it, and there's a brain signature to it. And so the machine that we brought with us is basically an EEG machine. So the way my brain, your brain, all of our brains respond to things that are meaningful that we recognize is different than the way it responds to novel meaningless things, right?

Has this been used in a court? Has it been used by a prosecution? Not in court. There are some Fifth Amendment issues as far as betraying yourself based on your brain waves. But you could use it for witnesses or you could use it within companies that are already using polygraph methods. Okay, well I’m excited to see this in action. I’m gonna step out; you guys are gonna prepare the first test.

Okay, excellent! All right, see you soon. Okay, we're ready for the test. Thank you for joining me, Hannah from Vsauce, Wren from Corridor Digital. So one of us today is going to be a thief, per Dr. Rosenfeld's instructions. We randomly determined who would be the thief by seeing whoever drew the green chip. Three, two, one…Hannah's the thief!

Oh my god, I knew it! So, Hannah, go on in, look at the item! In this real-world scenario designed by Dr. Rosenfeld, the designated thief actually takes the item, handles it, and gets a good look at it. This activity causes the item to register in their mind so they can’t help but recognize it later. The other subjects are also required to enter the room and sign in to prove that they were there, but they never get a look at the stolen item, and of course, Dr. Rosenfeld and Ann have no idea who the thief is.

Then each of us is tested. I'm going to take a seat here, and we'll get you all set up. Basically, I'm just going to put a couple electrodes behind your ears and a couple on your face and then the EEG cap on your head. What you're going to do is you're going to use these two mice in front of you to respond to the images that you see on the screen. So you'll see an image, a string of numbers, an image, a string of numbers. Anytime you see an image, no matter what it is, you're going to press this button.

All right, are you ready? The images we were shown contained a variety of jewelry items, including the stolen one, of course. Wren and I hadn't seen any of them before. Hannah didn't consciously react to any of them, but would her brainwaves reveal the one item she recognized? It was time for the results.

So here we are. Have you reached a conclusion? Well, based on what we know—and we know that the stolen item was a watch—and so we looked at the brain waves in response to the watch in comparison to the brain waves in response to other things. And we have two average brain waves or event-related potentials in each frame. The black one is the average brainwave response to the watch; the red trace there is the response to the other stimuli all average together.

We look for the peak-to-peak difference—the peak down here versus the highest peak following this peak—and we measure that, and it's rather obvious that the biggest response we found was Hannah’s. Wow! An interesting thing, Michael, is that you had a special response to the watch in comparison to the other items; that would suggest that there's something special about that stimulus, the watch—that it is meaningful to you.

Is it? I collect watches! I have about 40 or 50 of them! But Hannah might not have that kind of experience. I don't have that many watches, and out of the three of you, she definitely has the biggest response. So off to jail! The P300 concealed information test correctly identified which one of us was a thief.

But in this next test, I will be the only suspect, and they will know that I've stolen something, but the question will be, what did I steal? Inside this room are seven boxes, and inside those boxes are seven different items that I've never seen. My task is to open only one box at random and steal that item. All right, we've got a camouflage hat.

And that's not all! During this next test, I will be implementing countermeasures to try to outsmart the test. I will think a very strong thought every time something comes up that's irrelevant. Will that enable me to create brain waves powerful and dramatic enough to not be distinguishable from my response to the actual item I'm trying to conceal?

Anne, hello, welcome back! I committed a crime! Wish me luck because this time there will be punishments, like in the real world where getting caught in a lie can have consequences. And my friend Adam Savage was happy to advise me on my punishment. Adam, Michael, I am going to be taking an EEG lie detection test.

Okay, if I lose, there need to be stakes. Okay? Yeah, yeah, yeah. Naturally, I thought of Chinese water torture, which, as you know from personal experience, is when someone is restrained while water is dripped on their forehead. And then right after that, I thought, I know just the guy to talk to about this!

Okay, because I want it to be actually something that I'm fearing and dreading. So yeah, we did an episode on Chinese water torture on Mythbusters, and the creepiest thing that happened after we did this episode was that I got an email from someone from a throwaway account. He said we found that randomizing when the drops occurred was incredibly effective. That anything that happens on a regular periodicity can become a type of meditation, and you can then tune it out. If you couldn't predict it, he said we found we were able to induce a psychotic break within 20 hours.

It sounds like the stakes that I want! All right, thanks! You're welcome, sir. So now under the threat of Chinese water torture, I was extra motivated to beat the lie detector. All right, that's it! I was trying to think about all kinds of frightening scenarios.

I was trying to tense up, make sure I was really uncomfortable the way I was sitting. Well, for most tests, that would work! But we will take a look at this data and see! Yesterday? Wow! Okay! I don't think they're going to be able to figure out what item I stole. I really kept my mind active. I was going crazy with countermeasures.

I was sitting uncomfortably; I was thinking about all kinds of really crazy things. I was thinking about, like, what if I have diarrhea right now? How would that feel? How do I hold it back? And maybe I do have diarrhea! All these things that were sort of physical in nature as well! I thought a lot about death and dying and being dismembered. And if they can get it right, I would probably blame witchcraft as well as science!

Well, I would say that’s it! Yeah! Okay, we're ready to give him our best guess. Hello! Hi! Hey, you guys ready? Yep! Let me ask you this first of all—how confident do you feel? Pretty confident! What did I take? Camo hat! Camo hat! Camo hats! Yeah! That's impressive!

Did you see any evidence of countermeasures being used? Not something I would swear to. Even while I was doing the test, I thought, you know what? I'm reacting too quickly to the actual item; I should also treat it as though it's irrelevant. I should treat them all the same. If you are consistently reacting to all of the irrelevance with their specific countermeasure responses that you formed in your mind, they should produce bigger P300s.

Also, but the biggest will still be produced by the camo hat, even if you're countering it also. And that means I now have to be punished! Science is improving in its ability to read people's minds, and that raises some difficult ethical dilemmas.

We would have nothing to fear from perfect lie detectors if the people using them had perfect morals. So the prospect of a scientifically rigorous method of extracting information, while exciting, is also terrifying. As we continue to probe interrogation methods scientifically, we have to also continue to probe them ethically.

You know, Chinese water torture isn't even Chinese! Its earliest description comes from Italy in the 15th century, and the Chinese name was added later to make it sound more mysterious.

Oh man, these irregular drops are supposed to drive you slowly insane. You know, I think this is a lot less about getting information and more about punishment. And as always, thanks for watching! [Music] So [Music] [Music] you