The world's first lie detector suit
You may have seen people being put up against a lie detector machine known as a 'polygraph test' to determine whether they are telling the truth. But is it really that simple in real life? How much can our subconscious behaviour really give us away? Dr Sophie Van der Zee from the Computer laboratory at Cambridge university explains to Chris Smith how she has been using motion detection equipment to see if someone is seeking to deceive...
Chris - Let's just describe what we're seeing here. You have this gentleman who is wearing his everyday clothes but strapped onto him are orange boxes - one in the middle underneath his neck, one on the top of each arm, one on each wrist, one on each hand, on his feet, knees. What's in these boxes?
Sophie - These are called inertial sensors and what they basically do is together, they send a 3D image of how he looks like. As you can see here on the screen, he looks like a skeleton man via these receivers to the laptop.
Chris - So, this gadgetry that he's wearing is beaming movements that are being picked up by these orange boxes back to your laptop and it's then superimposing those movements onto this mannequin that we're seeing on the screen behind us.
Sophie - Yes, that's correct.
Chris - Well, let's see this in action then Sophie. So, what we've got, we've put Nick on our virtual dancefloor. Projected onto a very large screen behind us is a mannequin like you would draw in art classes which is showing all the bits of his body. Tell us what we're actually seeing projected here.
Sophie - Yeah. Well, you can see a skeleton man. You can see that you have 17 sensors on your body. So, what this sort of does is it transmits all the information about where he is in the room and where every sensor is as opposed to all the other sensors. So, that's global and local position. And together, that sort of gives us an image of the skeleton man in a 3D space. If you would be working in Hollywood, you would not be really interested in the skeleton man. You want to see Gollum up here. so okay, we now know how it looks like in person, we now know how it looks like on the screen, sort of a skeleton that moves around with 5 green dots. The question of course now is, how well can you dance? Okay, wow, very well. Okay, like a Saturday Night Fever is happening right here. Very impressive. Are you guys proud of your father? Maybe we should ask how your children are feeling right now.
Male - Yeah, he likes going on the stage quite a lot.
Sophie - Wow! Do you recognise him in here?
Male - Not really. He looks like in a film where they'd put maybe a Spiderman suit on him and he could fly.
Sophie - Wow! So I think you're your son's new hero. Okay, well thank you very much.
Chris - So, how is this going to flush out if this bloke did the deed?
Sophie - We did some experimental research where we had a lot of different suspects stealing some money and we found that people who actually stole the money moved about twice as much as people who didn't.
Chris - They were literally kind of a bit anxious dancing on the spot, were they?
Sophie - Yeah, but without necessarily seeing it because I calculated the behavioural differences between truth tellers and liars and I've gone back to the video data to see if I actually saw that liars move more. But I didn't really see that even when I...
Chris - What sorts of movements?
Sophie - It was actually very consistent across all the body parts so for the head, the arms, the trunk and the legs. For example, when we were talking during the interview about one of the two tasks, the people who were lying literally moved twice as much than the people who were telling the truth. And it was about 12 cm per second over their entire body.
Chris - Is this just more stereotypical things like crossing and uncrossing your legs, shrugging your shoulders or is it more subtle fine movements that you wouldn't necessarily pick up on?
Sophie - It's any type of movement and I think this is why we've been finding these results. So far, most psychologists have actually looked at movements by coding video data. Whenever you code, you usually say, "Did someone move his right hand, yes or no?" but you don't include how big this movement is and you also don't take into account all the small movements people make all the time. so, we thought, what if we literally move every millimetre that people move around. If we measured that and then see how much people actually move because that's been a very big question in deception research for decades.
Chris - The idea then, you would bring in the possible perpetrator, wire them up like this, and what do you do? Do you just observe them when you're just talking to them and ask them random questions - how old they are, where they live and that kind of thing where they're not lying and then compare the number of movements then with the number of movements when you then start probing them about the crime? I mean, how do you establish a baseline for what's normal for them? How do you know if someone isn't just very jittery?
Sophie - So, what you've just been describing is what happens with the polygraph. I have to say because we've only done a first few sets of experiments. We haven't actually tried to figure out how we can maximise our results on an individual level. So, what type of questions to ask is something we're actually going to investigate over the next couple of months. So, what you've been describing definitely works. I don't know if it works for this technique or maybe we have to do something else.
Chris - Would this be practical in a crime detection situation? Do you bring all your suspects in and make them look like this gentleman? What's your name by the way?
Nick - Nick.
Chris - Did you expect this Nick when you turned up here this evening?
Nick - Not exactly, but it's very fashionable.
Chris - He looks pretty good. Certainly, a talking point would be a conversation start or a party wouldn't it? But is this how you'd see this being used?
Sophie - No. this is an inertial system that works with sensors. But of course, if you ever want to use this in practice, you can't actually ask a suspect to put the suit on because A. no one want to and B. it takes a lot of time. So, what we're currently doing in the lab here in Cambridge is investigating if there's any alternative way of measuring movement without having to sort of touch someone so unobtrusively. We got some equipment from all kinds of BA, the army. We have radars, we have time of flight cameras and we're now creating one big test. We're to find out how else we can measure movements.
Chris - Literally, watch what a person does and use a computer to log that rather than wire them up in this way. And then they wouldn't even know that they're being effectively monitored, would they?
Sophie - Yeah.
Chris - But returning to my point, if someone is just nervous because they've got someone grilling them, like Alan is a bit "blokey," he looks pretty scary. If he's grilling someone, how do you dissect away the nervous effects of them being grilled by a police officer from just the fact that they're telling porkies?
Sophie - I hope that with our technology, that actually isn't an issue because we've been finding that although anxiety might affect how someone behaves, that's not what's causing the difference between truth tellers and liars. But it is one of the main criticisms about using the polygraph where you're comparing sort of control questions with critical or target questions that of course, they're quite good at detecting deception because whenever you're lying, you usually get a bit more anxious. But then some people who are actually innocent can become very anxious as well because someone is just asking you about if you've murdered this person. So any normal person will get a bit sort of anxious about that. So that actually is a problem. But because our movements were not actually caused by anxiety, I'm hoping that we can protect innocent people by using this method.
Chris - Could I train myself not to move more often so I could become really good at lying?
Sophie - So, with non-verbal cues to deception, behavioural control is definitely one of the things that makes our job very difficult because some people might find it very hard to lie and experience a lot of cognitive loads whilst other people get a bit more emotional or anxious. But some people can sort of step over that and control their behaviour and they'll try to appear as honest as they can. We haven't tried it yet with this suit, but one of the benefits of getting some media attention is that I actually have newspaper articles about a research. So, one of the things I want to do next - but anyone who will be my participant, please don't listen - is, give half the participants the newspaper article about a research that says that liars move more and the other half, we won't give this newspaper article and then see what they do because I think what they will do is try to sort of beat the system and I think that they'll try to move less. But now, the question is, if we can detect if then liars moving less as a countermeasure differs from actual truth-telling behaviour. So we were hoping to create a software that gives sort of a greenlight for whenever someone seems to be telling the truth, a red light for whenever they seem to be lying and then sort of an orange amber for whenever they're a bit suspicious because they're moving way too much or way too little to sort of fit the system so, that means you need to keep talking.
Chris - What do the police tell you about this? Do they seem enthusiastic or are they a bit sceptical or wouldn't they tell you the truth?
Sophie - No. I've actually had quite positive responses from the police. The local police here also offered us access to a sex offender sample so we can actually start to submit not just on students and general public. One of the criticisms you always get when you do this type of research is, what happens when you test the type of people that are usually interviewed by police officers? So hopefully, that's also one of the things we'll be doing this year