Schneier on Security
A blog covering security and security technology.
« The Psychology of Security |
| Corsham Bunker »
February 7, 2007
Eyewitness Identification Reform
According to this article, "Mistaken eyewitness identification is the leading cause of wrongful convictions." Given what I've been reading recently about memory and the brain, this does not surprise me at all.
New Mexico is currently debating a bill reforming eyewitness identification procedures:
Under the proposed regulations, an eyewitness must provide a written description before a lineup takes place; there must be at least six individuals in a live lineup and 10 photos in a photographic line-up; and the members of the lineup must be shown sequentially rather than simultaneously.
The bill would also restrict the amount of time in which law enforcement could bring a suspect by for a physical identification by a victim or witness to within one hour after the crime was reported. Anything beyond one hour would require a lineup with multiple photos or people.
I don't have access to any of the psychological or criminology studies that back these reforms up, but the bill is being supported by the right sorts of people.
Posted on February 7, 2007 at 6:38 AM
• 34 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
The most powerful evidence on this I saw was a program a few years ago on (UK) Channel 4 program on memory.
A woman who's mistaken, but powerful testimony had put someone in prison for around 3 or 4 years (When the mistake was realised) was being interviewed for the program. She said even though the rationally compleatly accepted that this guy wasn't her attacker and compleatly accepted his innocence, she still "remembers" his face attacking her and cannot see him or his picture without reliving the feelings of terror associated with it.
I had a dude try to climb in the window while I was in the house. At the time there was a serial rapist going around the area striking every few days or so. Anyway they didn't catch the guy that day.
But about 6 months later the cops come into work and showed my about 15 mug shots. The then asked if any of these guys was him. Well he wasn;t there and i said so.
That was the wrong answer apprently. the reply was "what about this guy, he matches your discription", they all matched my rather vauge discription. They contunued "look it was this guy. Just say so and you can get back to work". they keep this up for 20 mins, I was almost shouting at them that the guy who climb through my window wansn't there. I'm pretty confident that I'm right and I'm dam sure that I didn't want to be "sure" of something that I'm not sure of.
Some people can be convinced after the fact that something/someone did it and after a whlie they cannot distinuish this implanted fact and real life. This is particualr true for abuse cases with young victims.
In NZ there was at least one case where the mum overracted about something else. And then ask the chlid (8 years old) if the dad "did this to you". The Child is simply not able, after a bit of repeated questioning to tell the difference, they can even suffer the tramua of the imaged event. It was proved latter that the events never happend (physical examination when the child was older amoung other things- glab we don't have the death pently eh). There was a review on the use of young witnesses. I didn't follow what happened. Some studies have suggested that this type of thing can happen even in children as old as 14-15 (with supporting evidance).
Its not a easy problem to deal with. We want to listen to the victium, but can we trust the testomony?
Eyewitness testimony is based on the completely believable assumption that memory works like film. Linear, photographic, stored in complete - yet distinct - frames of information. Memory is nothing like; eyesight isn't even analogous to film beyond the lensatic function of the eyeball. The brain only sees a very limited area of information, and extrapolates a running partial stream into a convincing full-view world. It doesn't even read all of the information that falls on the retina. And that information is stored disassembled.
Recall is a process of reassembling pieces of a memory based on the gestalt impression. "Try really hard to remember" something, and your brain will helpfully supply information that fits the shape of the hole, whether it belonged there or not. You want a face? Your brain only knows the most poignant elements of the particular face you saw, based on what you noticed under emotional stress. Stare long enough at those, and you start to shuffle features around them to try and lock down what the rest of the face was like. Your brain doesn't remember the perp's face as such, but it will work with the pieces of faces it has to try to map that one face out of its difference to your absolute face.
Reminds me of this bit of legal wisdom:
"Anyone who says that there are always two sides to any story has never heard three eye-witnesses to an accident give testimony."
Reforming eye-witness interview procedures is helpful, but at the end of the day, the problem is with the jury. Juries give way too much weight to eye-witness testimony.
Related to Bruce's Psychology of Security article, that's probably built into our brains as well. If one of our tribe-mate relates his eye-witness account of how the sabre-tooth tiger ate his brother, I'll believe it and avoid the tiger. Good survial strategy. Not that good as a justice strategy.
An interesting approach would be to require education of juries on how memories can be mistaken, manipulated, fail, etc. so that the jury can rationally weigh the evidence. But that's not likely to happen.
A few years ago a rape victim gave an excellent description of her attacker, so good that someone quickly recognized him from being on local television. However, his alibi was unimpeachable: at the time of the rape, he was in a television studio being prepared for a live on-air interview to follow, with the crew, the host, and the other guests as witnesses. When they went live on the air, the victim watched the show at home.
Somehow, her memory got altered, replacing the rapist's face with the guest's face. How? It mystified her as well.
Memory is malleable. It is an active process. Interrogation techniques, among other things, can change memory.
The book _Blink_ by Malcolm Gladwell has some interesting things to say on this subject.
A small excerpt:
[regarding the last waitress that served you in a restaurant]
"Describe her face. What color was her hair? What was she wearing? Was she wearing any jewelry? Believe it or not, you will now do a lot worse at picking that face out of a lineup. This is because the act of describing a face has the effect of impairing your otherwise effortless ability to subsequently recognized that face."
The bill seems to be opposed by those who want convictions and supported by those who are more inclined to want viable defenses. While I would prefer to err in favor of the accused, I do not like the idea of micromanaging the accuser. That seems to conflict with the idea of letting the DA or cop do their job, which includes hard to codify things like finding the hinky dude.
As long as the defense can argue that eye witness testimony is potentially flawed, especially when taken long after the event, I think the system is ok. The prosecutor can argue that the victim is sure that the accused was the attacker and the jury has to decide who they believe. It's not a great system, but I don't see a better alternative at this point. Until we have bought into the RFID tracking and ubiquitous surveillance for all, we have to settle for something less. Of course, if we get to that point, I see a lucrative business in framing people, but I digress.
There is one guy here at work that, the first time I saw him, I thought he looked a lot like someone else I know. That was my first impression and it still holds true, even after alternating which one I saw most recently. I'm pretty confident that if I told my wife to look for the guy who looks like Derek, but isn't Derek, she would be able to pick him out pretty easily. I find that kind of situation is not too uncommon, even though I'm pretty sure I would have a hard time describing a sketch artist to him or herself accurately.
Experiences like Greg's where the person is asked long after the event and nudged towards a particular answer lend support for the idea that about a quarter of people in prison are innocent, a quarter are guilty, and half are guilty of crimes other than the ones they were convicted of.
This is a good sign. There seems to have been a tendency with forensic science for the police and prosecution to rely on anything that makes convictions easier and to ignore anything that casts doubts on techniques or methods. Many of the 'infalible' techniques of forensic science look considerably weaker in the light of all the science. My prediction for the next huge hole? The 3 million+ UK DNA profile database and our old friend the birthday paradox.
@greg: "That was the wrong answer apprently. the reply was "what about this guy, he matches your discription", they all matched my rather vauge discription. They contunued "look it was this guy. Just say so and you can get back to work". they keep this up for 20 mins, I was almost shouting at them that the guy who climb through my window wansn't there. I'm pretty confident that I'm right and I'm dam sure that I didn't want to be "sure" of something that I'm not sure of."
There was a psychology study a while back that concluded that the police find a suspect and then try to prove that they are guilty. Contrast this with the scientific method that sets a hypothesis (finds a suspect) and then tries to prove it wrong (tries to prove that they are innocent).
"Some people can be convinced after the fact that something/someone did it and after a whlie they cannot distinuish this implanted fact and real life. This is particualr true for abuse cases with young victims."
A friend of mine, Gareth J Medway has a lot on this in his book debunking the Satanic Ritual Abuse cases in the UK a few years back: "Lure of the Sinister: The Unnatural History of Satanism" NYU press.
@archangel: "and your brain will helpfully supply information that fits the shape of the hole"
You mean the monsters hiding on those floral curtains in your childhood bedroom? Given a good source of white noise (train journeys work well) and a little bit of effort I can hear apparently objective external music-complete with full orchestration, the lot.
supporting UTx and UArkansas research on eyewitness testimony reliability:
It also depends on the variability of the faces shown in the line-up. As far as photo line-ups are concerned, the Duke Lacrosse player rape case featured a photo line-up presentation that only included players on the team. This not only tainted the identification but violated the NC rules for photo line-ups.
Anyone still wondering why the Durham DA no longer has this case?!?
There are instances where suspects have been badgered into confessing to crimes they didn't commit. Such instances aren't restricted to mentally challenged citizens. Police detectives are schooled in extracting confessions. Most police departments no longer video tape such interrogations, since juries have invalidated so many confessions after seeing how they developed.
And, lest we forget, what about the recovered memories debacles of the past two decades.
"but the bill is being supported by the right sorts of people."
Yeah, the criminally inclined.
Please consider this scenario: One evening you are alone at home, watching television or reading a book. The next day, police haul you into the station because someone identified you as a culprit in a crime. You have no alibi since you were watching TV alone. There is no physical evidence of your involvement in the crime other than the eyewitness.
* How do you defend yourself?
* Do the police treat you as an innocent person?
* Does the general public treat you as an innocent person?
* How does this eyewitness claim and your 'perp walk' affect your job?
* How many legal resources do you have at your disposal? Are these public defenders?
Two wrongful conviction stories from Winston-Salem
New line-up procedures and technology http://www.journalnow.com/servlet/Satellite?...
How much prison time would you accept if you were convicted of a crime and you were innocent?
I've been reading the book "Unchained Memories: True Stories of Traumatic Memories, Lost and Found" by Lenore Terr. She is a psychiatrist who studies memory, particularly cases involving childhood trauma and recovered memories. The signs and symptoms of true trauma versus suggested memory are fascinating. Specifically, see the chapter on Lua Greene's "memories" of sexual abuse.
Unfortunately, this does not address two of the biggest causes of misidentification: the implication that the person you're looking for must be one of the people included in the lineup, and memory alterations caused by sloppy police questioning.
Most people are not trained observers.
Petrea brings up an interesting point -> especially after a trauma, your cognitive processes are unreliable. If you take a victim of a violent crime and put them in front of a bunch of people who look somewhat like the perp, your brain is likely to gloss over the details and pick the one that looks most like the actual criminal.
"All ______ look alike (to me)"
There are reasonable studies that support this idiom. The more time we spend with people of one race, the better we become at differentiating them...identifying detailed differences that distinguish one individual from another. Conversely, our identification abilities diminish with regards to individuals of other races.
Moreover, our minds have a talent at 'filling in the blanks' as far as missing data goes. While this is handy in most instances, it can lead to false eyewitness recalls. It can also lead to songs getting stuck in our head.
My experience is that juries aren't always trivial or gullable. I was sequestered in a small room with no air conditioning in the summer for three entire days with unpleasant overnight stays at dumpy hotels, yet no one bent under pressure of getting back to their life. The whole debate was over whether the eyewitnesses to an armed robbery saw what they claimed, or if they'd "convinced themselves" that the person the police showed them was the attacker. We agreed they most likely weren't lying, but we couldn't convince one another that their subconscious hadn't spun their story.
The reality is that you're dealing with humans, who have issues. People lie. People have egos. People are racist. People have agendas. People have histories. The jury I sat on as a collective only believed that what anyone said fit into their perception of "the right thing to say on the stand." Real attorneys aren't like the ones on TV and some of them miss huge holes, so what needs to be said in a trial often goes unsaid.
As for New Mexico's treatment, I think they've got the right idea, but perhaps the wrong execution. I understand the attempt to thwart excessive bureaucracy, but sometimes you have to let the judge and jury decide instead of making hard-and-fast laws.
I look at this like a business process. We have a process (defense, prosecution, judge, jury, etc.). Collect some meaningful statistics on its effectiveness and if they show we're very wrong (like a 20% wrongful conviction rate), tweak. Never will the process be perfect, which is why there's so much human oversight. With some processes, if 75% of cases are "abnormal" and need to be handled manually, you've done a good job. We're so lucky to have a system that works for so many people in the USA.
Dr. Safer is one of the principal investigators of these studies. They have published on their findings with judges , and are working on the same kind of studies with lawyers, jurors, and police officers.
The unreliability of eyewitness testimony has been discussed in the mainstream US press before. I had read about it but my own memory played some tricks... Some websearch helped locate the articles I had seen:
Scott Turow alludes briefly to this in the context of a death penalty discussion on The New Yorker:
"We also proposed altering lineup procedures, since eyewitness testimony has proved to be far less trustworthy than I ever thought while I was a prosecutor."
Margaret Talbot, in The Atlantic Monthly also mentions this problem:
"When witnesses see six people at once, they make relative judgments, comparing the six and picking whoever looks most like the person they remember from the crime scene, rather than evaluating each individually. Conducting lineups sequentially seems like a minor change, but research by Wells and others has shown that it reduces the number of mistaken identifications—by as much as one half—without significantly reducing the number of correct ones."
This need to "pick one" is often accentuated by the eyewitness need to be helpful, and in the case of victims, of getting resolution. Then there's the helpful police officer guiding their choice.
PS Malcolm Gladwell makes for interesting reading, but he is often not very rigorous -- I think one should read him as a way to trigger interesting lines of inquiry, nothing more.
What gets me is the disconnect that exists in evidence weighting between witness tesimony and physical evidence.
Those of us who weren't hiding under a rock during the OJ Simpson trial (I tried...) may recall discussions of false-positive probabilities of DNA identification couched in terms of how many millions of people live in LA, implying that 1:100,000 or 1:1,000,000 FPP (depending on the test) meant that potentially hundreds of people might have committed the crime. As if the police had gotten the DNA, then pulled in all the matches from the LA area, instead of comparing it with the DNA of the guy who was picked up on the basis of totally separate evidence.
Meanwhile, juries did and do routinely accept perp IDs from witness line-ups containing as few as 6 people -- a 17% FPP -- as good enough to send the accused to the big house, or even to the chop.
If we were to hold witness testimony to the same standard of evidence to which we hold DNA testing, we would have witnesses pick the perpetrator out of a football stadium crowd, not out of a 6-person line-up.
Most responses here are recollections, so I don't believe them.
I agree that eyewitness testimony is unreliable at best and the system should be reformed. But won't performing lineups serially cause a whole new series of errors based on "first/last" thing you saw?
And of course in the "duke lacrosse/rape case" the right thing to do was pretty much the opposite of everything the DA did. Good thing for him he is an attorney and immune to responsibility for his actions or he would be in jail.
Just a theory to throw out there:
Although I never pursued it, I flirted with law enforcement career back around college time...had one job I worked with many who were actively pursuing it.
On many Police exams, one of the cognitive skills they test for is "recall" -- being able to remember details, license plates, faces, etc.
Then (again at good Police Academies) they'll get formal training in recalling details.
So perhaps there tends to be a systemic bias in law enforcement that people must have good memories? After all, they have good memories...
Attornies may have a similiar bias -- to pass their exams, they're having to recall accurately quite a bit of information. They too may be a group that is self-selected for, on average, very good memories for details.
"* How do you defend yourself?"
I accuse the police of being the monopoly protection racket that it is - and then countersue them under the RICO laws. Has never been done - should be interesting.
> Has never been done - should be interesting.
Only as comedy material on late night talk shows (if you were lucky enough to tickle a host's funnybone). Or do you seriously believe a judge would even blink once before tossing that out?
"Or do you seriously believe a judge would even blink once before tossing that out?"
What happend to the jury?
If the judge raises a stink then I ask why he disagreed with me. He of course will no logical or rational explanation because I'm merely stating the definition of what the state is.
And if he objects, I merely point out that he violates the utmost basic juridical principle of not being judge in his own case. And if he doesn't get it, then I say he is unqualified to be a judge.
Of course I'm joking and I would never say that - precisely because I know how ignorant the judges are of the law.
After all they got their degrees at accredited institutions which are required not to teach the law.
There was a serial killer a few years ago. Police made a "profile" that matched just about every white male in the state. They then went out and grabbed a few hundred "suspects" for DNA swabs. Turns out the serial killer, once caught and convicted, was black. Now there are ongoing lawsuits trying to get the wrongful, and what should be illegal, swabs off the record.
Tax dollars at work.
All this and not a single mention of Rashomon?
Memory works by association - you try to retrieve a memory by sending in what you think is part of the memory you're looking for. Get a close enough match, and everything associated with that will pop out.
The trouble is, while you're searching, you are also writing the search patterns somewhere in memory. It's RAM, and there's no such thing as a read-only mode.
hahaha...i did a text search before i posted. must have typed rach...oops
My father-in-law is a defense attorney who works on a lot of police-brutality and wrongful-arrest cases. He observes that many in the law-enforcement community resist changes which actually help them do their jobs.
One, as mentioned above by @crispm, is that ID accuracy goes up significantly with sequential ``lineups'', as opposed to the actual put-'em-all-in-a-line method.
Another is that conviction rates go /up/ when the police install cameras which record the entire questioning of a suspect---no more ``he said/she said'' or claims of the third degree. I can think of at least two reasons to explain the correlation, but leave that as an exercise for the reader.
@quincunx: Re: support by the criminally inclined.
You make a mistake all too common these days: thinking all this ``nonsense'' is about ``coddling the criminals''. Rather, it's about protecting the innocent, of whom there are plenty being accused. Would you care to bet what percentage of people held at Guantanamo are actually guilty of terrorism? The administration, under the pretense of keeping terrorists off the street, is fighting tooth and nail to prevent us finding who's held falsely.
The same sort of tension goes on 24 hours a day in every police station in the world. The civilized nation-states at least /try/ to sort out the innocent.
once (in the early '80's), I took an armed citizen self-defense course from an American of Middle-Eastern descent, who writes columns for NRA mags and teaches many, many courses in the law-enforcement and self-defense fields...excellent pragmatic courses....Masaad Ayoob. http://www.ayoob.com/
Excellent, excellent course....explained and showed in depth why you really, really, want to leave the use of firearms and the confrontation of criminals to the professionals....while nonetheless showing you the elements of armed self defense if you have no choice. One thing he drummed into us during the week long course was what was called witness management, which boils down to this: If you are approached with deadly force and are yourself carrying force, you want to (if there is any time or space whatsoever):
1) Shout out something like, 'STOP, don't come any closer, I see you have a gun/knife, I don't want to shoot', preferably while you have your gun either still hidden or at least pointed down
2) Back up while continuing to shout this out.
What you are doing is to draw attention to the confrontation and to the fact that it is being forced upon you. Otherwise, bystanders never notice and turn around until they hear the shot and the testimony can easily be something like, 'I saw the whole thing, the kid never had a chance, the guy (you) just shot him down'.
He said to carry some cash to just be able to throw at a mugger in the hopes that it will avert confrontation and violence and give you a chance to escape.....and try to give you a feel for what it would be like to do fatal damage to another person....and think about what your kids will remember if they are with you. And much much else, that TV viewers have no clue about..........
Schneier.com is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc.