Read a book on this and this seems to corroborate much of what I read. One of the most interesting and simple techniques discussed was the order in which people recounted their stories.
The example given is an employee who is consistently late for work. If you ask them to recount why they are late one day, a liar will tell you the story linearly: "I woke up, ate breakfast, hopped in the car, was on my way, minding my own business, then someone hit my car. I got out to trade insurance ... [etc]".
A truth teller jumps around, usually starting with the climax "Someone hit my car on the way in. I then realized my insurance was expired. I had just been going through my bills the previous night ...".
This is easily ascribed to the fact that the liar is either making the story up as they go or are repeating a rehearsed lie. A truthful person can jump around easily because they are recounting distinct memories.
> This is easily ascribed to the fact that the liar is either making the story up as they go or are repeating a rehearsed lie. A truthful person can jump around easily because they are recounting distinct memories.
Alternatively, the liar's experienced, good at his craft, and more capable of jumping around.
The best liars use the exhonerative voice to avoid telling a story about themselves at all:
"Bags are subject to search" means "We will search your bags". "A shooting involving Tacoma County Police left two suspects dead" means "Tacoma County Police shot two suspects."
Huh? I'm not disputing that institutions carefully use the passive voice to clean their prose of negative connotations...but the two examples you mention are not "lies"
1. I've seen the "bags are subject to search" phrase many times, including the public library. I cannot remember the last time my bag was searched, even at the airport when the search is differentiated from the usual scan. So that phrase is definitely not a cover-up for "We will"
2. Again, not a lie. And I don't mean in the "not technically a lie". There are many cases where there is a shootout involving police and the cause of the suspect's death is ultimately determined as suicide. So the "A shooting involving..." phrase is perfectly acceptable in a breaking news update when no determination has yet been made.
> I'm not disputing that institutions carefully use the passive voice to clean their prose of negative connotations...but the two examples you mention are not "lies"
Clippy says: "Hey, it looks like you're trying to start a semantic argument! Can I help with that?"
> I've seen the "bags are subject to search" phrase many times, including the public library. I cannot remember the last time my bag was searched, even at the airport when the search is differentiated from the usual scan. So that phrase is definitely not a cover-up for "We will"
Congratulations on being middle class and white.
> There are many cases where there is a shootout involving police and the cause of the suspect's death is ultimately determined as suicide. So the "A shooting involving..." phrase is perfectly acceptable in a breaking news update when no determination has yet been made.
There are many cases where this isn't the case, too.
By using the exonerative voice, a liar avoids committing to a particular narrative. They can easily back away from a statement by saying "I only said baggage is subject to being searched. I didn't say they were searched."
That said, I don't think the exonerated voice is a particularly skillful one. A careful listener can easily pick up on that.
Principle of Charity, my friend. I'm not really sure where you got a negative vibe from his comment. It was more of a comment on how effective word-choice is on someone's symamantic tree.
But those aren't lies, just a mixture of omission and clever wording to reframe the facts. They're used in unidirectional communication (a sign at the airport, a television broadcast), but don't hold up well when there are follow-up questions.
Only if they are aware of this technique and purposefully and skillfully prepare to defeat it. That turns out to be harder than you'd expect for methods like this. The methods used by Israeli security agents to profile travelers at airports have been well known for decades, but even so are still very effective.
If you catch X people trying to blow up a plane and no planes get blown up, that's fairly good evidence that your method of catching people trying to blow up planes has been 100% successful. So far.
What's wrong with it? He's not making the usual "tiger-proof rock" argument used to refer to the TSA, after all. The TSA doesn't, as a rule, catch people who were actually planning to cause trouble. They either catch people who didn't know/care how much toothpaste they were allowed to smuggle onboard, or they catch idiots who forgot they were carrying a .50 caliber Desert Eagle in their shaving kit. This is a different argument altogether.
The difference is that there are many groups who have publicly declared their intent to blow up Israeli airliners, so it's not as if you're defending against a non-existent threat.
No, but if my security guards keep turning away actual elephants at the border and I find no actual elephants nearby then I can conclude that my anti-elephant security works.
You raise an interesting conundrum, but you can tell of a certain success rate by the number of liars who eventually confess to lying or are found to have been lying.
Wouldn't that be just the same case as with the caught liars but extended to a group of people who despite being successful at the interview failed fulfilling the task? There is still an unknown group of "liars" who passed the test and never confessed.
We're talking about legal contexts here. Almost no one, unless coerced/brutalized by police/DAs, admits to crimes they did not commit. And the you have the thing where you can verify their admission of guilt with corroborating details.
There have many many many cases of individuals getting behind bars because they have been coerced to plead guilty and confess to crimes that they never committed.
"unless coerced/brutalized by police/DAs" is not really exceptional, once in a decade, type case, sadly it is rather routine.
I know personally a person who plea-bargained, which means admitting to a crime they didn't commit, because the likelihood of being convicted (and the penalty of conviction) was so great a threat. He pled to a lesser crime (that he also didn't commit) and served 5 years.
Sadly, there is a systematic problem in which many must plead guilty because they will be punished beyond measure if they do not. It is structural and endemic in American justice:
General level of terror in Israel? It is absurdly low compared to the threats that surround them (lets not go into if such threats are warranted or not)
For sure. I assume the detection technique works pretty well, even against "better" liars. That said, I've encountered folks that are so good at being slippery that you almost have to admire their skill.
That's the problem with this technique. The false positive rate is likely pretty high, and without knowing the prior probability that the person is lying, the technique could be worse than useless.
If the false positive rate is too high and the prior probability a given person being interviewed actually lied, then the test is actively harmful.
I.e., if a positive result is more likely a false positive, then you end up wasting investigation resources. In this case you'd have been better off just flipping a coin to decide if someone warranted more investigation.
>There will be truth-tellers who fail this test, but you find nothing conviction worthy. They'll be fine.
Tell that to all the people who lost jobs, or spent months of their life under investigation because of failed polygraph tests.
I should of course mention that not everyone uses these tools correctly. I suppose that could be harmful. But then again, if an investigator is simply looking for a reason to find someone guilty, then the accused is eff'ed either way.
I won't oversell it and tell you that the book turns you into a human polygraph. It simply surveys what is known about deception and human behavior and how you can apply that to your interactions with others.
Well, at an airport security check I got interviewed before boarding and the guy asked me how I got here. I thought it was a weird question to ask and I answered I got here by airplane (I was in a foreign country visiting).
He was totally non-plussed by the answer and said, "No, how did you get _here_?"... I was actually confused by the way he said it and he added, "What did you do this morning to get _HERE_?". I said I got up, packed, got on the train, transferred to the airport bus, and got here. He just paused and moved on to a few innocuous questions. It was kind of comical. I was waiting to be led to some back room for further interrogation, but in the end he just said 'enjoy your trip". :)
Probably the point of the question was to be confusing: it strikes me that if you give the brain something to work on - parsing and responding to a confusing question - it could make responding with false information more difficult.
Imagine you were trying to remember the details of a false passport and an alibi for what you'd be doing - you get a weird question and your brain is then suddenly blank when you try to present your false identity.
This could also be an alternate way to ask "Did you pack your bags yourself? Did you leave them unattended?" to which everyone invariable answers 'yes'.
First off, I think this technique is completely true, in my own experience. However, I think this isn't the end-all lie detector, in that you have to know what to have the subject tell a story about.
Take the issue of national security (in the USA). To get a national security job, you'll have to go through a lie detector test[0], where they'll ask you a ton of questions. Say I'm lying about my name. Do you ask a story about my name? Say you ask about my childhood instead. If I was lying about my childhood, I'd tell stories that were as closely aligned with real life as possible, and change as few details as I needed to. A story about playing in the park in the summer with my parents doesn't change a whole lot if it's in North America or Europe.
Bottom line is, I think this method works, with the caveat that you have to know the event or thing that they would be lying about. Trying to find out if a spouse killed their significant other? Check, you can ask the stories around their alibi. Trying to figure out if James Bond is going to sell national secrets? What story are you going to ask about?
> Bottom line is, I think this method works, with the caveat that you have to know the event or thing that they would be lying about.
You also have to be careful in the other direction. False negatives are a problem but false positives are an even bigger problem because most people will be innocent. And you could be asking someone to recall a time when they were drunk or exhausted or under stress, or the truth is embarrassing or painful to think about, or the day you're asking about was entirely unremarkable, etc.
I thought this had been discussed on HN earlier this year, but searching for it has failed me. I did find https://news.ycombinator.com/item?id=9061964, though, which looks interesting and related and got no attention.
These kinds of analytical methods (i.e. word counting) work well in controlled experiments when you know you're going to be lied to by a subject, and also told the truth so you have a point of reference.
In the real world, you don't know if you're being lied to. That's the entire point! You don't have any normalized data that says this is the number of words a person uses when they're telling the truth and when not, so the detection technique will be as inaccurate as this missing calibration info. And since there are wide variances between individuals in both general memory, recall, and "talkativeness" you have to develop a sample corpus per person. Also you need to interview the person on multiple days and randomize when you ask the questions so as not to introduce bias and to protect from ordering.
It's easy to conduct a study like this and proclaim an obvious conclusion, but this means almost nothing for real-world application of the technique.
Yeah. There's a reason why the standing order for the public is "don't ever talk to the police without your lawyer in the room". Doubt or confusion about past events is often perceived as an attempt to fabricate. Add in the ability for the interviewer to feed false facts to you in an attempt to question your story, and it's tough to pass the truth as the truth.
It doesn't matter if you're actually innocent, what matters is whether they can cast doubt on your innocence. Do you think you could repeat a simple story that just happened to you recently in multiple different chronological and detail configurations/contexts without introducing some inconsistencies?
Only if you're in a hurry! The interviewer can patiently wait for the subject to become bored. Formal interviews in intimidating settings (like panel interviews for jobs) are often scheduled to be much longer than necessary just on the expectation that the first half will be nervous ice-breaking and the real insights are after everyone loosens up and is capable of acting naturally.
That's probably half of "in vino veritas": drinking is an excuse to spend a few extra hours with nothing to do but chat.
I think the point of asking about senses is that memories of feelings are more consistent than memories of detail. And if it's being done properly, the person's answers are being benchmarked against their other answers. If you're struggling to remember a particular day at all then it's not necessarily held against you, but if your recollections of what was said and done at a particular event are hazy after you've calmly explained that you know you were there until at least 8pm because you got the 8:14 train home then...
Only if this were considered conclusive (which it isn't) and if it weren't attempted first with an honest recounting of an actual memory (which they do).
I assume the true experience is done first to give a basis for comparison, since this is based on the idea that people paint pictures of their memories less reliably when they're fabricating them on the spot.
Having poorer recall actually might bias the results more in favor of a lie, if the interviewer doesn't reject the method as unsuitable or inconclusive first.
>Morgan found that the use of these mnemonic props – open-ended questions about various sensations and sequences of events – dramatically increased memory recall about what had happened. The subject’s stories consequently became more and more complex, and richer in detail. Or at least, they did when people were telling the truth. When it came to the lies, even well-rehearsed ones, the subjects tended to falter and were unable to complete the interview. According to Morgan, this was because when they were prompted to dredge up deeper memories, the liars had nothing to draw on.
Is this technique robust enough to detect implanted false memories (psychologist implants memories of child molestation) or are such recalled "memories" indistinguishable from real ones?
In an expansion[1](pdf) the classical Bugs Bunny at Disney study[2](pdf), over several leading interviews participants added more and more detail. So while it would probably work with an accidentally and freshly implanted memory; it could probably be defeated by either purposely pushing for sense recall, or perhaps accidentally over the course of many attempts to fix the "memory".
Knowing the technique is probably half way to defeating it in all honesty, as long as a person has time to prepare (or be prepared).
I'm guessing it wouldn't. A memory doesn't exist as a single entity, and a false memory is often corrupting a couple real parts of memories, either overwriting (or outright adding) a small detail or mixing together real parts into something that didn't happen.
If someone was given a false memory of a traumatic experience, they likely have still experienced something similar to the faked traumatic experience (maybe saw it in a movie) and will have an actual part of a memory to feed greater details as someone dug into the memory.
A false memory of seeing a litter of puppies will still have plenty of real occurrences of seeing puppies to subconsciously draw upon.
When you are lying on purpose, you have to make sure whatever you say is accomplishes whatever made you lie in the first place. It has to make you look good, make it seem like something is not your fault and so on.
Therefore, you need to consciously validate whatever your brain outputs when you ask for some bullshit to "remember". Maybe this creates a delay or makes people be conservative in what they tell you for fear of accidentally saying something inconsistent.
If you had a false memory, however, your brain could make up details as it goes along and you would just say them aloud, without feeling nervous or consciously thinking about consistency and validity at all.
This matches my experience that the best lies are based in truth and experience. That is, either use something which actually happened to you and change only the most necessary bits or use something which you know well to have happened, just not necessarily to you.
I wonder if successful con artists (or novel writers) have such a vivid imagination that they would be able to pass this kind of test. If you have such a good imagination that you can convince yourself you actually are in a make-believe world, does the lie become just as convincing as the truth?
Put another way, if a novel is so richly detailed that I feel as though I'm actually in the story, would describing my experience of that novel be interpreted by this system as truth or a lie?
I'd think it depends on the amount of time you've spent imagining this experience in the novel. If you have had enough time to vividly imagine the entire experience, none of the questions should catch you by surprise and you could answer almost as if you were actually there.
Regarding con-artists, yes I suppose it comes down to whether they can be as quickly imaginative on the spot as most people who are truthfully recalling an experience. That's probably where the inaccurate 15% of results come in.
Wonder if there's a dataset for this. Maybe run a simple n-gram over the interviews and see if certain phrases show up more often for truth tellers vs liars.
If you're in the poker world, you know that physical tells _are_ a thing but only to reaffirm your read and should not be the entire basis of whether you bet/call/fold (ala Casino Royale).
Because it's so hard to place your entire reasoning because, "he's covering his mouth, so he's bluffing" and "he's shaking his hands, he must have a good hand" sometimes those are true tells, but sometimes maybe the room is just really cold and the player is trying to warm himself up. Or maybe he's throwing a reverse tell.
I remember Phil Hellmuth shilled a book about physical tells from a former FBI agent, not sure if there was any substance to that book.
Interesting how this dovetails with the art of writing -- when you're writing a book, it makes the story much more compelling if you engage all five senses. I wonder if we have some intuition about this and a story that doesn't feels "one dimensional."
I also recall hearing a funny thing on a podcast once about someone interviewing someone with a (pretty dubious even by weird paranormal stuff standards) far-out alien abduction tale. They stopped them and asked "so you were on this ship for hours... how did you go to the bathroom?" Interview was over. They didn't have that one scripted. Turns out the person was actually trying to virally market a book.
I remembered those tricks but not the movie it came from. How true is this though ? even innocent people cracks down under pressure even if they rebel against injustice at first. The amount of "intelligence" and organization to protect whatever dogma a government wants to protects scared me. Noticed the check mark when the student asked for a little humanity.
Honestly, I don't know how effective monitoring affect is per se but as a wild guess I'd say it's at least as accurate as a polygraph/coin flip.
However, I think there is more weight in the idea that a story that is repeated verbatim like that is more likely to be fabricated or at least neglecting some of the truth (which is the takeaway I got from the article).
But just like every other form of torture its efficacy is highly questionable (e.g. the inherent possibility of false confessions). I think sleep deprivation is particularly problematic because while it does seem to almost universally deplete the victim's willpower/resistance, it also causes a profound detriment to memory and other cognitive function-- meaning it is completely useless in cases where the victim isn't talking because they don't remember the information rather than because they just don't want to cooperate.
the behavioral response. Going from angry/rebel -> submissive happened with innocent people, preferring to take a penalty rather than keeping on being treated this way.
But just like every other form of torture its efficacy is highly questionable
This one truly baffles me. Man, as soon as I find myself tied on a chair, I would spill the whole bag of beans and more. Heaven forbids they pull out the pliers.
It's interesting that a low occurrence of unique words in a story can be an indication of a lie. Does that position people speaking a secondary language in an unavoidable place of distrust? Beyond body language, it's just more difficult to tell if someone's story lacks richness because they're not telling the truth, or they're simply not comfortable with the language and have a small vocabulary to draw upon.
I think ideally you'd compare the story you're investigating with other, provably-truthful stories told by the secondary-language speaker. That way word reuse could be a relative value rather than an absolute one.
I think I see a very obvious problem with this: in the experiment they ran, the interviewer seems to be leading the interviewee in how to remember an episode. Which means there is no way to make sure the interviewer is not (consciously or not) planting those very same "rich details" that purportedly make the difference between a lie and a truth, into the interviewee's narration.
So I don't see that they're avoiding a "Clever Hans" situation at all. They could be leading their interviewees on, providing subtle clues to the ones that are telling the truth, but not to the ones that are lying.
And even if this was double-blinded, there's still no way to ensure the interviewer doesn't interact in a different way with different interviewees, therefore messing up the results of any experiment pretty badly.
Nothing new here, this interviewing technique by storytelling has been standard procedure for a long time with the British Police. One simple variation is to recall the story from the beginning, then recall from the end and see what doesn't match up.
I used a variation of this trick when I was a resident assistant in college. We had the unfortunate job of busting drinking parties in the dorms and part of our responsibility was to collect ID numbers or social security numbers if the person claimed to not have their IDs. People would quite happily rattle-off a fake SSN. But if you wait 60 seconds and come back to them and ask for the number in reverse, it's essentially impossible to repeat the same fake, especially if you've been drinking.
I used to be a bouncer in my college town and you'd be surprised at what lengths kids will go to pass off their fake ID's as real. they'll memorize everything on the ID, including where the city is located - "So Flagstaff, is that North or South of Phoenix? I can never remember?"
I would ask for all the information on the ID if I thought they looked young. Then once the person was confident they had passed my tests, I'd ask them what their area code was when they call home. 9/10 a fake ID holder will fail this test because its not something on the ID, but something they should know without fail.
We didn't have to. We had a slick way to find out without tipping off the ID holder.
It went something like this:
As I'm asking the kid the details on his ID, I make a hand gesture to the waitress that works the rear bar near the door. She would then have the manager call the phone at the front desk where we were. As I was talking to the kid, I would show Bouncer #2 the city and as I was talking to the kid, Bouncer #2 would ask the manager the area code over the phone in a simple coded manner.
The bouncer would give the city and state initials to the manager. So if it was Phoneix, AZ. He would say, "Yeah, its a PH, AZ." The manager would then confirm he had the right city and then relate the area code. After a few months, I had memorized a lot of the shorthand we had for cities so that it only took a few seconds to confirm. Considering most fakes come from the same states and cities, it wasn't a long list to learn.
By the time I got down stalling, the other bouncer would then ask him the question, already knowing the answer.
Knowing some bouncers and bar tenders, they tend to have a dynamic level of skepticism. If you look 25-ish, that sort of slip probably won't deny you entry. If you look 20-ish, it may be more likely.
The difficulty here is that the incentives heavily favor allowing false negatives, and denying false positives. I know that in Virginia, the bar tenders were personally responsible if they served alcohol to anyone underage. And in a college town (where I was for quite some time in grad school), there are loads of underage kids trying to get into the bars, and the authorities would sometimes hire underage kids to test the bars.
Oh yeah, I definitely don't blame them, and I understand the incentive structure. I don't even think that I necessarily disagree with it, given our laws about drinking. It's just an irk :p
> ask for the number in reverse, it's essentially impossible to repeat the same fake, especially if you've been drinking
I would find this extraordinarily difficult to do in any state of mind.
The only way I could give you the digits of my telephone number backwards would be to mentally go through my telephone number from beginning to end once for each digit of the number, and stop a digit earlier on each pass.
Whenever I give a fake phone number (which I do quite often when inspecting real estate) I always confidently reel off the number I had fifteen years ago. It still trips easily off the tongue.
I hope the poor sucker who has that number now doesn't get too many follow up calls.
Similar thing when giving your phone number to someone hitting on you. Better strategy is to swap two digits with each other, making it easier to remember and increasing the plausibility of an "honest mistake."
That's not exactly what's going on here. The point is to use the sensory memories to improve the interviewee's ability to recall, and then see who actually recalled better the second time around. Yes, you could probably compare the two for inconsistencies as well, but that's not what he was measuring for this—just the change in level of detail between the two versions.
That's right, and another difference is that word frequencies change. E.g., in a lie there might be many references to "the green car", while a truthful story would refer to it variously as "the old green car", "the yellowish hatchback", "that ugly old car", etc. Yes there are inconsistencies in those descriptors. That is a property of memory, not of guilt.
Which shows yet another deficiency in LEOs' standard methods. They interrogate until they get an inconsistency, which they know the courts will interpret as guilt, even though this research shows that it is no such thing. Therefore intelligent guilty people can talk to the police with no fear, while no innocent person should ever talk to them.
This reminds me of Jobs To Be Done (http://jobstobedone.org/) interviews style that dig into emotions of why customer switched to a different product.
The interview jumps around the story looking for any events that moved customer to or from new solution, and digs into the emotions associated with each event to get customers to recall all details.
That's a nice write-up but I think that experienced and well prepared liars would be able to beat that.
The best way I've found to detect lies is that truth is 'internally consistent' and lies never are, all you need to do is to focus on any inconsistencies not matter how small and then to apply pressure on them, they'll fracture the story much like a crystal with a flaw in it.
Considering that 'truth' is really our best recollection of past events, why do you say that truth is internally consistent? I mean, maybe it is, but it's not a given.
Truth is grounded in verifiable facts. You can be mistaken about facts, so that's easy enough to fix. Liars can't change their story because once they do the whole thing starts coming down, they can only keep so much of the fabricated universe in their head.
That's mostly my interpretation of it, I've met only very few people that were lying outright during the course of many DDs but when they do that's how they all - so far, and of course there could be undetected ones that made it through somehow though I highly doubt that - fell down.
From what I've seen it is very hard to keep the story straight over the course of a couple of weeks if it isn't true.
> From what I've seen it is very hard to keep the story straight over the course of a couple of weeks if it isn't true.
Yes. I would add that it can be very hard to keep the story straight over the course of a couple weeks EVEN IF IT'S TRUE. That's what I was trying to say.
Sure but the teller then has no vested interest in keeping the facts and the story from deviating from each other whereas a liar does have that interest. So the liar can't 'fix' the story but the person that tries to be truthful has no problem.
I worked with the DHS a few years back on a project trying to use ML to detect lies based on a variety of ques, including paul ekman's research into facial expressions:
Yeah, I know right? I'm sure I read on HN some time earlier that CIA agents / undercover would always make sure their lie is actually as close to the truth as possible -- if it comes down to it. (last resort type thing).
I could do this easy peasy, and then ask me to do it backwards and I'll stare at you and go "No idea, I'd have to think about it... generally don't think backwards".
The key in any case, is always... don't give them rope. Patience.
"The polygraph, for example, which measures physiological changes such as blood pressure, respiration and skin conductivity, only achieves an accuracy level of around 50%..."
Where does that statistic come from? I've definitely read that polygraphs have over 90% success as long as you account for obvious "counter measures". Basically, the person administering the test can clearly see if you are pulling tricks to defeat it.
My first thought was a motivated liar could learn to beat it... surely part of its success is from the liar being unprepared for the novel questioning technique?
Storytelling is a very effective way to interview job candidates. It's a good way to tear down inflated candidates to get to the truth, but more importantly it's a good way to uncover and find talented high potential people.
This might be a better technique if the suspect does not know what the psychologist is looking for, but probably still does not work if he does. So ultimately it still isn't very usable in the real world.
The example given is an employee who is consistently late for work. If you ask them to recount why they are late one day, a liar will tell you the story linearly: "I woke up, ate breakfast, hopped in the car, was on my way, minding my own business, then someone hit my car. I got out to trade insurance ... [etc]".
A truth teller jumps around, usually starting with the climax "Someone hit my car on the way in. I then realized my insurance was expired. I had just been going through my bills the previous night ...".
This is easily ascribed to the fact that the liar is either making the story up as they go or are repeating a rehearsed lie. A truthful person can jump around easily because they are recounting distinct memories.