Could ‘Brainprints’ Unlock Your Future Phone?
Our phones are the key to our personal digital data. How can we build a better password? Smartphones now include fingerprint scanners, and companies have been testing other biometrics—including retinal scans and heart rhythms—for the ultimate personalized password. Could “brainprints” be used as an unbreakable identification tool? A team of researchers created a system that could match EEG readings to an individual with 100 percent accuracy. Team member Psychologist Sarah Lazslo explains how these brainprints were created.
Sarah Laszlo is an Assistant Professor of Psychology at Binghamton University in
Binghamton, New York.
IRA FLATOW: These days our phones are the key to all our digital data. Crack the password, you get instant access to bank accounts, emails, all sorts of personal information. But how can we build a better password? You know, that’s what we’re talking a lot about with phones, the best kind of password.
Well, the future personalized password might be right inside your own body. Smartphones– think of it– they have a print fingerprint scanners. Companies have been looking into everything from retinal scans to heart rhythm trackers. But what about your brain? Your brain has a lot of electrical stuff going on there.
Brain prints, using your brain waves as an identifier, harder to hack than a fingerprint. And researchers at Binghamton University are studying how you could create these brain prints. My next guest is part of that team. Sarah Laszlo is an assistant professor of psychology at Binghamton University in Binghamton, New York. Welcome to Science Friday.
SARAH LASZLO: Hello.
IRA FLATOW: Wow, brain prints. How do you create a brain print?
SARAH LASZLO: Very carefully.
IRA FLATOW: You stole my line.
SARAH LASZLO: Yeah. Sorry. What we do is we bring people into our lab, and we connect a bunch of sensors to their head so we can non-invasively measure their brain activity. And then, we show them a whole bunch of images, rapid fire, one after another. And each of the images to sort of designed to elicit a very unique response from person to person.
And once we have about a half an hour worth of their brain data, we can submit that data to a computer. And then, the computer is able to sort of process the brain data and then figure out who is who.
IRA FLATOW: So if you showed me a picture, I would have the same brain print all the time?
SARAH LASZLO: We don’t use a single picture, because the response to any single image, it changes too much. And it’s too noisy. So we use an average of about between 25 to 50 pictures.
IRA FLATOW: And how it accurate has this become?
SARAH LASZLO: 100%
IRA FLATOW: Let me hear that again. I’ve never heard anything work 100%
SARAH LASZLO: I know. It’s 100% accurate in a pool of 50 people to a period of up to 10 months. And that’s the limits that we’ve pushed it too so far.
IRA FLATOW: Well, when you do have anything like this and you say it works 100%, immediately the hackers in the room are going to say I can break that.
SARAH LASZLO: Absolutely.
IRA FLATOW: So I guess part of your job then is to try to figure out the weakness.
SARAH LASZLO: Yeah. So immediately, like you said, immediately as soon as we got it to work, the very next lab meeting was, OK, we got to work. Now how do we hack into it? And so we’re trying a whole bunch of different hacks. But engineers on the team have some computational hacks they’re working on. But I’m not one of the engineers. I’m a psychologist.
So the hack that I’ve been working on is the trying to figure out how possible it is for a person, a malicious person, a hacker, to train their brain to impersonate the brain activity of someone else to sneak into a brain biometric system.
IRA FLATOW: It’s sort of like changing your fingerprints or your face or something, but you’re changing your brain.
SARAH LASZLO: Yeah. So it’s sort of the brain equivalent of maybe putting on a fake latex fingerprint to fool a fingerprint scanner.
IRA FLATOW: And how easy is it then?
SARAH LASZLO: It’s not easy. So far we have trained– we call them– our brain hackers. We’ve given them up to 12 hours of brain hack training. And we’ve only gotten them to be able to become about 10% more similar to their target than they were when they started. And they need to get to be about 40% more similar on average than they started. So 12 hours isn’t enough, at least.
IRA FLATOW: Does my brain print change over the course of the day if I’m hungry, or sleepy, or cranky, or something else?
SARAH LASZLO: It probably changes a little bit, but the computer algorithms that we use are sophisticated enough that minor fluctuations don’t break the identification system. And like I said, that is true up to a period of 10 months. So even though your brain activity obviously changes day-to-day, at least, for a time span of that long, it doesn’t change enough to break the computing system that identifies people.
IRA FLATOW: Now with 100% accuracy, where do you go from here?
SARAH LASZLO: I know. I kept telling my grad student, I don’t know what you’re going to do now, because you’re never going to be able to beat yourself. But most of it is towards trying to make the system that acquires the brain data cheaper and faster now.
IRA FLATOW: So yeah, OK. Give me a setup of how you collect. Do I have to put that cap on with a lot of electrodes or things? Just before you get that answer, let me remind everybody that I’m Ira Flatow. And this is Science Friday from PRI Public Radio International talking about brain waves and brain prints with the Sarah Laszlo. She’s a system professor– is that SUNY Binghamton, or is that the University of Binghamton?
SARAH LASZLO: It is a SUNY, but they tell us that we should call it Binghamton University.
IRA FLATOW: I want to SUNY Buffalo–
SARAH LASZLO: Oh, you did?
IRA FLATOW: Yes. So I can call it SUNY Binghamton, if you don’t mind.
SARAH LASZLO: You can call it whatever you want, but I will get in trouble if I call it anything but Binghamton University.
IRA FLATOW: Yeah. I know that do the same thing in Buffalo. So tell me is it a cap, one of these typical things with all the wires coming out that we see?
SARAH LASZLO: Yeah. Right now it is. And so one of the challenges is to make it not like that. Because of course, nobody wants to put that on every time they want to get into their computer. So one of our prototype devices that we’re testing has the sensors embedded in Google Glass. So you just put the Google Glass on, and it has the sensors in it. And then it does the identification in the on-board processor or on the smartphone that the Google Glass is paired with.
IRA FLATOW: Just with the Google glasses?
SARAH LASZLO: Yeah, with the Google glasses.
IRA FLATOW: So you don’t need the whole brain, you just need a part of the brain to sense it?
SARAH LASZLO: That’s right. We even with the data we collected with the head worth of sensors, we found that we only needed about five of them in order to get the 100% accuracy.
IRA FLATOW: And which part of the brain to stick this Google glasses around?
SARAH LASZLO: So the Google glasses just go on like Google glasses normally would. So when we do that, the active sensors are sort of over the temples. The sensors that we used in a published study with 100% accuracy where all placed over the back of the head. So that’s one of the challenges with the Google Glass is that it only gives us access to the front of head, and that’s not where the best sort of classification data came from.
IRA FLATOW: But you see what you’ve done here is you’ve given Google Glass a reason to be reinvigorated.
SARAH LASZLO: A reason to live.
IRA FLATOW: Yes. You should be hearing from Google like in five minutes, I would think.
SARAH LASZLO: I was at Google in September or October actually.
IRA FLATOW: You’re a way ahead of me on this one.
SARAH LASZLO: Or they are. They’re pretty smart over there.
IRA FLATOW: Yeah. So I mean, if you can get Google Glass to come up a way of identifying [INAUDIBLE].
SARAH LASZLO: That would be the greatest miracle is finding something to do that’s useful with Google Glass.
IRA FLATOW: And have you started a company or anything? If you’re going to start talking businesses, is SUNY going to be involved in this?
SARAH LASZLO: We do now that we have one patent that exists. And we have another one that’s pending. But we haven’t started a company yet. We’re scientists, and so we don’t know how to do that.
IRA FLATOW: That’s where the lawyers come in.
SARAH LASZLO: Yeah. And fortunately, SUNY does have a lot of good lawyers that are there to help us out. And they’ve been very helpful.
IRA FLATOW: So we might see this bearing fruition not to– you must have competitors, right?
SARAH LASZLO: No. We’re the first lab worldwide to accomplish 100% brand biometric accuracy with the particular technique that we use. So not really, which is nice.
IRA FLATOW: So you’re just sitting back and waiting for something new to happen now?
SARAH LASZLO: Yeah. We’re sitting back and waiting to find out when whether the NSF is going to give us more money. That’s what we’re really waiting for.
IRA FLATOW: That would be small box compared to what Google could.
SARAH LASZLO: That’s true.
IRA FLATOW: I can see why you’re going out there. Is there anything, you know, physically that you couldn’t hack into, the Google Glass if not the brain?
SARAH LASZLO: Yeah. So you could hack into, of course, the computer that is holding, say, the reference brain prints, and you could alter, say, the target’s reference to be your own reference. Or you could steal the reference so that you can feed it back into the computer. Those kinds of hacks are the things that my collaborators and engineering are working on. And I only know what they’ve told me about that.
But apparently, we have a paper that they’re preparing, where they’ve used some techniques from image processing, in particular steganography to be able to identify when a sort of reference brain print has been stolen and fed back into the system. And there about, last time I checked, there are about 70%, 80% successful in identifying that kind of attack.
IRA FLATOW: And so you’ve got a paper that’s coming out being published?
SARAH LASZLO: That one, yeah, the steganography paper is in preparation right now.
IRA FLATOW: OK. This is quite exciting, Dr. Laszlo.
SARAH LASZLO: For me, yeah.
IRA FLATOW: Yeah, well–
SARAH LASZLO: Thank you.
IRA FLATOW: –for always who have their Google Glass lying in the drawer somewhere.
SARAH LASZLO: Yeah, I know. That’s where ours lays most of the time.
IRA FLATOW: All right. We’re all in this together. Thank you, Dr. Laszlo for taking time to be with us today.
SARAH LASZLO: Thank you so much for having me.
IRA FLATOW: And good luck. That was quite interesting. Sarah Laszlo is persistent professor of psychology at the State University of New York at Binghamton, or as they like to say, Binghamton University. That’s about all the time we have for today. BJ Leiderman composed our theme music or our thanks to our production partners at the studios of the City University of New York.
And on the web this week, did you know that the US has a chief data scientist, the first one ever? And we talked with the head honcho himself, DJ Patil. Check them out that sciencefriday.com/datascientist. You can email us at firstname.lastname@example.org. Have a great weekend. Happy Passover, if you’re celebrating. I’m Ira Flatow in New York.