Schneier on Security
A blog covering security and security technology.
« Cell Phone Surveillance System |
| DARPA Cyber Colloquium »
November 1, 2011
The Economist on Lying
Two articles. And this is the cited work.
Posted on November 1, 2011 at 6:14 AM
• 28 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
nice squid reference in the second article.
Hmm they all relie on having prior knowledge to compare "known patters" with the test subjects "current pattern".
It is without doubt a remarkable step forward in a relativly short time but, we still have a long way to go.
Oh and another issue not mentioned, whilst fMRI requires you to be in a closed machine (due to magnetic field) the IR sensing can be done at a distance admitedly extreamly crudely currently.
But I expect that technology will vastly improve within a relativly short time with the right drivers (DHS dosh anybody?).
Thus we could be looking at mind reading from a distance at some point in the not to distant future.
However one thing that has always worried me about the brain sensing experiments, they usually have as a basic requirment "no left handers".
On enquiring about this from somebody I know they joked "The trouble with you lefties is your brain is not wired right"...
@Clive: That makes me wonder if we can figure out a way to "encrypt" our minds. Our brains are remarkably "plastic" already, so why not?
@Natanale "can figure out a way to "encrypt" our minds"
This is probably a current research topic for the intelligence services. If they can't create a ghost to fool these things their effectiveness will be degraded.
Probably have to go back to recruiting or creating psychopaths. But maybe training will include yoga and meditation.
The more I look into the mirror the more I see Kurzweil looking back at me.
Even assuming this stuff gets accurate enough to be used in the real world, it will (as described) only be able to catch a certain limited class of lies. If you say you have never seen an object or person before, or that you're not currently thinking about some particular thing, sure. But that leaves a huge range of wiggle room that experienced liars will have no problem exploiting.
@paul: Also, is a brain scan enough to prove that the person *really* were there/heard that/saw it?
Just considering hallucinations, schizofrenia, etc...
To not forget classical "honeytokens" (false information that the subject never knew was false). And why not use immersive 3D tech? Might not do the job for everything, though.
If all else fails, go for 1984 style doublethink.
occlumency lessons not just for adolescent wizards in children's books
"But that leaves a huge range of wiggle room that experienced liars will have no problem exploiting."
Firstly never answer questions if you can avoid it no matter how good a liar you are, if you don't talk you cannot make mistakes...
But if you have to, try to get a second person in you can trust such as a lawyer who is known personaly to you. Don't go for an appointed representative, as the saying goes "you get what you pay for" and if appointed they will do the minimum they can to get rid of you. Which as many in US prisons know includes selling you down the river.
If you have to put up with an appointed individual, each time a question is asked, ask to speak privatly with the legal representative away from the questioners and any microphones etc. Don't stop no matter how pissed off the appointed representative gets, it's in your interests to say nothing, if the representative then "leads you" you have a better case to retract anything you are forced to say.
If you cann't have a representative of some type and have to talk (unfortunatly the case in the UK these days), this is where you will have to start lying.
And as I have observed on this blog a number of times the best liars don't lie.
What they do is take the facts as known and put a little twist on them, but in slightly random ways to shift the perception of the listener to their way of thinking (the way a group of colaborators often gets caught out is their rehersed story is to much the same in detail and lacks "colour and variance").
Further a good liar knows how to deal with such things as "Did you do this?" style yes/no answer questions they simply react with some feeling with another question such as "What on earth makes you think I would do that?", or "What sort of sick mind could possibly think that?" then goes into a long moral tirade and refusess to come back on topic and gets steadily more animated and calls the questioner's own morals into question untill the questioner either gives up or backs down.
A simple technique to show outrage non verbaly is to tighten your stomach muscles to make your face go red and open your eyes wide and stare directly at them right in the eyes. Leaning forward and make your voice first loud then low and quiet, if you can spit words out, it will make the questioner move backwards, clenching and unclenching your fists will make the questioner uneasy. Which is the point a nervous person is entering into the "flight" mode as you appear to rouse up to "fight" mode, and is nolonger in control of the situation.
Importantly never answer a question with a yes or no only some kind of ambiquity or by asking another question. You can answer with known facts but keep them uncertain, this has a dual purpose in that it confirms you are speaking the "truth" but importantly have little memory of the issues in question. Use expressions like "It's difficult to tell because it was..." dark / the light was bad / it might have been / etc (what you call wriggle room).
When you get good at it often you can get an awful lot more information out of the questioner than they get out of you. And you can catch the questioner out when they start inventing things (ie lying) to try to get information out of you.
If they make the mistake of saying "We have evidence..." invariably what they actualy have is nothing of the sort, just some vague facts that they think tells them something. Don't hesitate to call it into question and laugh at them with "you call that evidence, what are you...?". Contary to what you see on CSI / NCSI / etc all physical evidence is at best an effect from an unknown and thus open cause, observations by witnesses are compleatly unreliable even trained observers are lucky to get close to 50% of stuff right two or three hours later unless they are making a record at the time.
The ultimate responses to some styles of questions is another type of accusation at the questioner to simply state "You are on a fishing trip stick to specifics" or simply state at someone making an accusation at you "Is why are you lying to me?" then refuse to elaborate and refuse to speak any further with such a duplicitous person.
The important point to remember if you wish to be a good liar is never never alow some one to develop real empathy with you, as that is when they effectivly control you. Likewise don't alow them to keep control of the conversation, throw in irrelevances in detail and refuse to be put back on the rails with a statment such as "As I was saying before you so rudley interupted..." and when the questioner gets going on some point interupt them with "is this nonsense relavent to anything in particular?" and if they say yes then say "In what way" and keep questioning their responses.
Oh do the oposit of what the questioner wants, so keep folding back the conversation so it does not move forward with things like "as I've already told you it went like this..." and start from an irelevant point in time long before the supposed incident. Likewise if the questioner trys to fold it back start by saying "I think you will find I've answered that, why are you wasting everybodies time?" and don't budge. Get up and move around when they try to order you to sit down refuse say you've got cramp are not feeling well or need the toilet etc.
Also refuse any food or fluid unless you personaly get it out of the tap / machine etc, if they push it at you push it back and say "You eat it" if they refuse simply say "why have you druged it", even throw it on the floor or at them if they are persistent.
Oh and remember the "two questioner" routien they are their to give each other thinking room, which you don't want to alow them. When one asks a question occasionaly speak to the other questioner with things like "Is your colleague always..." a liar / a bully / rude / moraly degenerate / etc. You need to break up any flow or patern they have not just that of the questioner.
Oh and long before you get tired start to nod off and show signs of becoming compleatly disconnected and ask questions like "what time is it", "is it raining", start long rambalings about "did you see XXX on TV the other day?"
As I was once told "A good liar can talk all day and say nothing.".
A forensic use of fMRI as a lie detector would need to sort out real-time visual and other sensory input, memory, imagination, representations of concrete and abstract patterns, and individual brain encoding differences.
Remember it doesn't have to actually work. It just has to be either accepted in court, or believed to work by security agencies - just like polygraphs.
@Natanael L: it depends on what you mean by "prove", as NobodySpecial points out. The oldest techniques, involving evoked potentials, can tell whether a scene is "familiar" or "unfamiliar", for example, but lots of pictures can appear "familiar" even to someone who has never seen that particular picture before. The newer methods, which apparently detect which of a large-but-limited set of video clips are closest to a scene firmly imprinted on recent memory, still would rely on having an appropriate set of clips to choose from (rather in the way that mug-shot identification relies on the target being in the mug-shot collection).
@ Natanael L,
"That makes me wonder if we can figure out a way to "encrypt" our minds. Our brains are remarkably "plastic" already, so why not?"
As far as I remember the human brain is most plastic around the age of two and then goes down hill from there. Around about 21 it's down to something less than 10% of it's ability to absorb raw information in a way that modifies it's structur in significant way. Thus adults tend to to be a lot less able to adapt to the loss of limbs etc than children.
Also our ability to learn significantly new methods or tools tends to go down with age, admitadly a large part of this is laziness, in that people don't want to significantly "slow down" to learn and then come to speed with a new tool, unless the motivation is significant (see London Ambulance and their sacking of staff that could not learn to type quickly enough when they went to a computerised system).
Objectivly I can see no reason why people could not be taught to thing in obsficated ways, but how would they learn, normaly learning progressses as a feedback process, without suitable feedback they could end up doing the opposit.
Where to start?
Claims involving fMRI should be taken with a grain of salt.
So should studies with extremely small numbers of subjects (3 in the second, 9 in the third).
So should studies where you only hear about the number of successes and not the failures (the first study succeeded once with one volunteer and twice with another; how many failed?).
The recreation depicted is not "surprisingly accurate" or a "reasonable simulacrum". Come on, a black T-shirt vs. Steve Martin in a uniform?
The more I read about this line of research, the more it sounds like the "remote viewing" saga all over again.
IMO, this is just a high-tech polygraph and probably suffers the same weakness.
Lies by basically honest people will be detected at about the same level they will be detected by simple observation.
Those of us who are basically socio/psychopaths will easily beat this just like we currently beat polygraphs and interogators. See Clive R's excellent instructions above.
@echowit - that's not the problem, the problem is that since this is a big expensive box it must be correct.
So simply scan each applicant for a job requiring security clearance and don't hire those who show a red light.
Then start checking airline passengers/potential terrorists and those whose brain shows nervous activity when you show them 9/11 footage you put on a watch list.
Then you check all teachers/scout masters/etc, show them child pornography and black list those that show brain activity.
No need for this ever to be tested in court, no need for any scientific facts to be involved - just like polygraphs.
And of course it must be accurate - look how big and expensive it is!
Well I'm very happy about the two articles and the cited work.
@Clive Robinson: Show the test subjects their brain scan live while sitting in the machine. Do this hundreds of times. Ask questions, let them choose if they'll lie or not, and train the computer to detect lies ("pattern detection" in scans for true and false statements), and tell the subject every time what the computer guessed.
If the subjects successfully manage to modify their thinking so that they the computer no longer can find relevant patterns (so that true/false guesses hits ~50% accuracy, even with the best computer models) we have succeded.
When a number of subjects consistently have fooled the computer, we ask them to tell what they did. We then tell this to new subjects who don't get feedback. If they succed too, we've got working general instructions! Yay!
If not, more research! Can it work if we let these new subjects who get no feedback talk to the subjects who succeded? If then, can we modify/create new general instructions? If yes and they work, success!
(FYI: Mythbusters tried this once. Grant fooled the machine while the two others failed.)
@ Natanael L,
"We then tell this to new subjects who don't get feedback"
It's getting to this point that is going to be very very expensive and potentialy lethal.
fMRI is expensive not just to buy but to use, and I suspect the pattern matching system likewise (though I suspect you could build it in with the MRI processing electronics with little difficulty).
But it gets worse as those who have had to be in one know the process is unnerving and most certainly not conducive to thoughtfull activity. But what about the health issues? so little is currently known about the effects of the large magnets and high powered RF fields in the longterm.
With most non pasive body scaners you dump energy into the body in pulses in frequency ranges at which various parts of the body are effectivly transparent. At the very least when the energy is nonionizing you are going to expect heating effects.
Most of this energy is going to be focused on the brain eyes ears nose etc. Now I don't know about you but I would be reluctant to have any of those "cooked" as I need them all for living in our modern world.
Even if we do stay well below the cooking threshold we know next to nothing about the effects of longterm exposure to intense magnetic fields and pulsed RF seperatly let alone combined
For instance another organ adjacent to the head is the thyroid, which amongst other things is part of the immune system and has known effects on many noncommunicable diseases that significantly reduce life expectancy. Then there's the... quit a long list of other organs in the head and neck...
(Oh and having kept many different small furry creatures as pets when younger I would be reluctant to put any up for testing - yup even I have a soft spot 8)
These kinds of blog entries is what makes me think Schneier is someone that is not to be taken seriously when it comes to science and specifically the human brain.
I am concerned that he is fascinated about the notion of reading peoples minds, as if reading the mind has to be possible somehow. Succombing to the lure of the prowess of science and technology.
I can only hope that Schneier consider the possibility of not ever knowing some minute brain process, insofar as one is working with interpretations.
Is Scheier even aware of things related to postmodern thoery and specifically of the "problem of representation"?
@Curious - it's not that fMRI doesn't work - it is perfectly good at showing different regions of the brain being used. The danger is in the conclusion being drawn from the results.
There was a famous study that had white Americans pronounce African names and English names. The fMRI showed different areas of the brain were being used - hence the people were intrinsically racist.
So if you have to think differently to pronounce "Jakaya Kikwete" rather than "George Bush" the brain scanner proved you are a racist.
"I am concerned that he is fascinated about the notion of reading peoples minds, as if reading the mind has to be possible somehow. Succombing to the lure of the prowess of science and technology"
I think you are the person we should have concerns about as you don't appear to understand what the issues are.
For many people on this blog including I suspect Bruce, is the notian that the likes of the DHS and others spending tax pays money on "magic boxes" that absolve them of any responsability.
We know that various organisations especialy governmental ones put great store in the myth and magic of things like the polygraph despite all rational science simply because it gives the illusion of scientific method in determaning truth (something humans are realy quite bad at).
As long as there is money to be made by quacks from the deep pockets of (apparently) gulable politicos and their servants then people need to draw otheres attention to it, simply because for the average person it's difficult to fight what you are not aware of.
I do not have a link to the sources, but in the last two years believe I read two articles about recent scientific research that seem to complicate knowledge about "brain science", making known conclusions seem generally pretty dubious, or so was my impression:
1) A braincell has a more intricate network that was previously known. (Imagine a pattern resembling a dense and chaotic spiderweb around a cell)
2) That it has been shown that brain activity is going on in an idle brain, that previously could not be detected with older equipment.
Fine, but it is not like I am against people pointing out articles, I am merely voicing my expressed concern, because if Schneier never express his own opinions, it seem unreasonable if say I am to believe he is on top of things and not functioning like some propaganda for science.
Just my thoughts, as being a sceptical guy that is somewhat interested in philosophy and in how the brain works.
I do not intend to argue with you, but I will want to say I think my comment was fair,my points easily understandable and I think that your point about me apparantly not understanding the issues is uncalled for.
I will not pretend to understand what probably is already a variety of known problems with scienctific research and the application of science-stuff, however I am currently not terribly interested in learning as much as I can, and am content in holding a general view based on common sense. So I hope you people will excuse me for my perhaps arrogant behaviour when I voice my concern here.
"it's not that fMRI doesn't work - it is perfectly good at showing different regions of the brain being used."
No, it shows where blood flow is being directed in the brain, which more or less correlates to which parts of the brain are being used more at any given moment. If the statistical analysis step gets done right, anyway.
"that's not the problem, the problem is that since this is a big expensive box it must be correct."
Good point. And the logic goes on, "since ... it must be correct" it's OK to spend whatever on it.
The motives for advancing imperfect technology
(as may be used to sell a story by supporting it with 'evidence')
are a subject worthy of consideration.
We may need a revisiting of the Crucible.
Check or the book "what every body is saying"
By an ex FBI interviewer who used 'tell's
In body language to catch criminals lying
Schneier.com is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc.