Researchers Can Identify You By Your Brain Waves With 100% Accuracy (business-standard.com) 89
An anonymous reader writes: Scientists have developed a new system that can identify people using their brain waves or 'brainprint' with 100% accuracy, an advance that may be useful in high-security applications. Researchers at Binghamton University in U.S. recorded the brain activity of 50 people wearing an electroencephalogram (EEG) headset while they looked at a series of 500 images designed specifically to elicit unique responses from person to person -- e.g., a slice of pizza, a boat, or the word "conundrum." They found that participants' brains reacted differently to each image, enough that a computer system was able to identify each volunteer's 'brainprint' with 100% accuracy. "When you take hundreds of these images, where every person is going to feel differently about each individual one, then you can be really accurate in identifying which person it was who looked at them just by their brain activity," said Assistant Professor Sarah Laszlo. One thing the paper doesn't talk about is the effect of time on the accuracy of the system. People may perceive different things when looking at the same picture a year later, for instance.
Buying Reynolds Stock (Score:5, Funny)
Re: (Score:1)
This is why you never see people wearing tinfoil hats. Why can't sellers sell them at street level like normal businesses?
100% accuracy...with 50 people (Score:5, Insightful)
Re:100% accuracy...with 50 people (Score:4, Informative)
Indeed. My statistics are rusty, but with only 50 participants it seems unlikely you could legitimately claim a general accuracy above maybe a mid-to-high 90 percentile. Even then you'd be talking about the general accuracy within a 50-person group. Take a 500 person group and the average difference between two sets of brainwaves would naively be expected to be about 1/10th the size, and the minimum difference would likely fall far faster than that. Take a more generally useful set size, like say the population of a major city, and I seriously doubt they could uniquely identify anyone but the most abnormal.
Re: (Score:2)
The counter-argument is that you increase the number of stimulus pictures until you regain the desired accuracy.
The counter-counter arguments are manifold: pictures -> blind subjects? Auditory responses are likely much more problematic to read and distinguish (audible signal response swamping any individual cognitive responses). 500 pictures, really? Who has time for that? How accurately do the electrodes need to be placed every time the test is run? Don't you have better things to do with your time
Re: (Score:2)
Re: (Score:3)
More like sqrt(1/10) the size.
Re: (Score:2)
How do you arrive at that?
As I see it, if you have P data points ("mind prints") roughly uniformly distributed within an N-dimensional measurement space (the optimal situation for discretely identifying them between measurements) of N-volume V, the average N-volume around each point will be roughly V/P. Double the number of points, and you halve the volume per point.
I suppose that would mean that the average linear distance between adjacent points will fall with the Nth root of P, but unless the "margin of
Re: (Score:2)
I assumed a Gaussian distribution, not a uniform distribution.
Re: (Score:2)
That only makes things worse, as the average distance between adjacent points is much smaller due to most points being concentrated within the peak. I don't think it appreciably changes the rate at which that distance scales as the number of points increases though - in the one-dimensional case you should still get 10 times as many points crammed into the same range, so that on average each point can only exclusively claim 1/10th the range.
Re: (Score:2)
Indeed, you would be correct under an assumption that the samples are equally biased/unbiased. I would actually want to see more than just one test. I would want to see them doing the similar research on 25 identical twins from various region as well. That would at least address whether their method is accurate and how well it does.
Re:100% accuracy...with 50 people (Score:4, Insightful)
I'm wondering if it suffers from some of the other problems that plague biometrics - is the "brainprint" unhashable, and will it change with age?
Re:100% accuracy...with 50 people (Score:4, Insightful)
Or mood. Or sobriety.
Re: (Score:2, Insightful)
Or drugs, both illegal and "prescribed". Both voluntary and forced.
--sf
Re: (Score:1)
So what they're saying is that this is pointless and only certain people will end up using it? The article points out only "high security" locations such as the "Pentagon" and "other high security" locations. So...what they're going to use EEG's as people walk in and out of a building?
It only works on gradual students. (Score:2)
It only works on gradual students.
You have to be educated to a certain level, and then you have to be so poor that you volunteer for this type of crap. And then it works, 100% of the time.
Re: (Score:1)
I have a much simpler system that just uses an ordinary weight scale and Can Identify You By Your Weight With 100% Accuracy! I tested it here at home on our family of four, and it correctly identified each of us every single time!
False negatives and false positives (Score:5, Interesting)
100% accuracy can be misleading. Are they talking 0% false negatives, 0% false positives, or both?
I could easily see a situation where it has 0% false positives, but a high false negative rate.
That is, I could claim that my "Presidential Identification" is 100% right if I said no one I met was the President of the US. No false positives because I never said anyone was President.
Similarly, someone could do it the other way around, claiming everyone was the President and have 100% accuracy because I identified every single President in the sample.
Re: (Score:2)
It is just a 50 person sample. They are talking perfect accuracy for that sample, but that does not mean a lot, the sample is far too small for that.
Re: (Score:2)
It would be dishonest to claim 100% accuracy for anything other than 0% false negatives AND 0% false positives.
Re: (Score:3)
I don't have the journal article itself, but the way I read the abstract, they mean 0% false negatives and 0% false positives.
They did not, however, appear to test people from outside the group. That is, if I were to show up, it's unclear if it would identify me as not one of the sample. That still leaves a pretty substantial room for error, but it's a very good starting point.
Problem in search of Morty (Score:2)
If it's use for authentication (Score:2)
Totalitarian Wet Dream (Score:1)
This is just another of the many advances that will enable the creation of a complete totalitarian state sometime in this century.
Re: (Score:2)
Yay (Score:3)
Re: (Score:1)
does it work if you have 7+ billion people? (Score:2)
Ok. So you can uniquely identify one out of fifty. Let's see how that works with over 7 billion people.
Also, would they be able to pull the 50 people out of a crowd of others of unknown scans?
Re: (Score:2)
Doesn't matter. If your brainscan matches criminal Y or terrorist Z, you must also be a criminal/terrorist. Time to get shipped to a hole. Welcome to the future.
Note, access to money may require a brainscan as well, but also other verification. After all, there can be no chances with money on the line.
The signature will likely change over time (Score:1)
I'm guessing that these signature series they are developing to uniquely identify people will slowly diverge from what the person's actual brain pattern is on observing the images over a period of time.
no need to scan my brain: pineapple pizza is gross (Score:2)
Researchers... recorded the brain activity of 50 people wearing an electroencephalogram (EEG) headset while they looked at a series of 500 images designed specifically to elicit unique responses from person to person -- e.g., a slice of pizza...
How long before the local pizzeria makes you sign a Terms of Service agreement that says they're allowed to scan your brainwaves while you're looking at a slice of pizza, and then sell that data to the highest bidder?
On a more serious note, using biometrics of any kind to secure systems is kind of dumb, because they can't be changed, even when they get compromised.
Names (Score:2)
Stress & Biometric Issues (Score:5, Interesting)
Some rightly noted that it may be possible for brain waves to change over time. However, I wonder if stress can significantly change the identification? For example, merely looking at images may provide a different brain wave result if the person being examined has a gun being held to their head whilst looking at the images. In kind, what if the person just learned his/her significant other is cheating on them. (By a mattress with an app, surely)
A common biometric issue is that if the information that represents who you are is stolen (a fingerprint, iris pattern, etc...), you cannot easily change it. However, I wonder if appropriately controlled stress or mind exercises can change one's brain pattern?
Re: (Score:2)
Though it would be interesting to discover if a properly motivated infiltrator could actually
Disorders or stress (Score:1)
Things like PTSD, forms of dementia, or personality disorders come to mind. How do these affect brainwaves in regards to the test?
While I'm not sure about the latter two, I recently listened to a radio-documentary about PTSD and how one company was combining monitoring of brainwaves with various forms of therapy in order to find the most effective treatment, so it stands to reason that heavy stress and particularly PTSD may change things.
500 unique signatures, that is all. (Score:2)
How do biometrics work? (Score:2)
How is it determined that every person's fingerprints are unique, that their irises are unique, that blood vessels in the retina are unique, that brain waves are unique?
With things that can change over time, how reliable are the biometrics? Fingers can be injured, retinas can change with macular degeneration and other problems, irises can change when the eye is injured, brain waves???
Re: (Score:2)
As far as I'm aware, the assertion that all of those are unique are based entirely on a lack of contradictory evidence. How much effort has been expended to try to find contradictory evidence I don't know, though I'd expect that someone would have run a "check for apparent dupes" process on existing fingerprint databases by now.
Re: (Score:1)
Re: (Score:2)
fingerprinting and forensically supplied prints, supposedly absolutely identifying a suspect for much of the time since 1911, now looking less and less reliable as time goes on:
http://www.forensicmag.com/new... [forensicmag.com]
What if I have a hangover ... (Score:2)
from the night before, or have drugs (legal or otherwise) in my system ? How does it perform ? It probably does not matter in the instance of allowing people into a sensitive area - might even be a benefit: do you want someone with a sore head near the big red buttons of a nuclear missile ?
Mmmmm... pizza..... (Score:2)
Pizza hut (Score:1)
I used to like Pizza until I tasted one at Pizza Hut recently. Now I almost puke at the sight of a Pizza.
Time is not your friend (Score:2)
CAPTCHA 2019 (Score:2)
Re: (Score:1)
Maybe not for eternity, but many people are unchanging for dozens of years, and I can guarantee you that there are plenty of people that are effectively unchanged from their late teens until death.
The sad truth is that a very large number of people can be characterized as, such and such behavior has gotten me this far in life, and is therefore good enough for me to continue doing. Change in behavior only occurs when people recognize that their old behavior is not good enough; merely being not good enough is
How long does it take? (Score:2)
They said the participants "looked at a series of 500 images". If they have to look at the whole set for the identification to be accurate that's a lot.
At 10 seconds per picture that would be over one hour and a half. Even at 1 second per picture, and I'd be surprised it could go any faster, this would be over 8 minutes. Even in situations "like ensuring the person going into the Pentagon or the nuclear launch bay is the right person", waiting at the door for over 8 minutes watching the same stupid picture
Use case (Score:2)
Wants to look up time on phone.
Drags out EEG headset....