4 ms·

Eyewitness Memory Is a Lot More Reliable Than You Think

38 points by DamonHD 9y ago

ThomPete 9y ago
The problem still stands. You can't really rely on it to convince someone wrongly.
Theodores 9y ago
This gets complicated at public events where something goes wrong and the reported story is not consistent with the facts. Bystanders can witness the event with their own eyes and then go home to watch the news, to then believe the news-worthy version of the event. A spiral of silence sets in so all observers of the original event fall under the same spell and assume they are wrong and adjust their view to the world view rather than what they saw with their own eyes. At public events you also get a completely different view of what happened depending on whom you ask. Not everyone gets the complete story or picks up on all the clues. People bring their own expectations of what they expect to see and have cognitive dissonance to make it so they do see what they see and not necessarily what happened. The JFK incident was in this realm of unreliable eyewitness memory, those that thought they saw and heard the second shooter changed their view over time to align with the accepted story. This is how it works.
teddyh 9y ago
The article is good but the title is false. Eyewitness testimony is, as a rule, always filtered through layers of perception contaminated by various unchecked influences, which is largely the point of the article. The fact that properly isolated witness testimony is surprisingly reliable has no practical benefit for anyone, since, in practice, no witness testimonies are properly isolated.
- MarkMc 9y ago
  > no witness testimonies are properly isolated. Can't we make them properly isolated? The article says you need 3 things for reliable witness testimony: 1. Do not repeat the lineup, 2. Ensure the suspect doesn't stand out, 3. Record the confidence of the witness when identifying the suspect. If law enforcement follow these 3 rules and the witness identifies the suspect with a high degree of confidence, it seems the identification is very reliable. That would be a practical benefit.
  - microcolonel 9y ago
    The witness is also susceptible to suggestion. If they are reached over the telephone, or have a conversation with a family member, or the police accidentally (or intentionally) suggest something in an interview, then they are tainted. If the defendant is resourceful, then the chance of any of these happening blows up.
- abecedarius 9y ago
  In the last decade or two blinded lineups -- where the police officer with the witness doesn't know who is the suspect -- became the standard in a fair number of U.S. states. You can take this pessimistically: wtf, it took this long to introduce some basic epistemic hygiene? But it also shows progress does happen. "No practical benefit for anyone" is just giving up.
tzs 9y ago
A problem with lineups may be that memory is too good. There is an experiment that has been done several times that shows how surprisingly good our memories can be in some ways. The experiment is to show someone a very large number of images or normal things, showing them each image for a short time. The images might be things like 5 apples on a table, a man petting a dog, a couple of people talking, and that kind of thing...just normal things you might see every day. Then later, sometimes days later in some versions of the experiment, you show those people pairs of images, where one of the images was from the set they were shown earlier, and one is not. You then ask them which of the two images they have seen before. This has been done with sets as large as 10000 images, and people get 80% or more correct two days later [1]. There was a version where in the identification portion the two images shown were one from the set they had seen earlier and one that was similar. Examples might be a stack of $1 bills and a stack of $5 bills, or a bell with a narrow handle and a bell with a wide handle. Even with these similar images people did well. The one I've seen discussed used 2500 images and 2500 similar images, and people got 90% correct [2]. That particular one tested shortly after the people were shown the images. I wonder if this works when instead of the recall test being one image you have seen before against one that you have not, it is one you have seen against several you have not? And I wonder if it works when it is something live you are seeing instead of images? And how important is it that when shown the initial set of images people know they are being shown images and are going to later undergo some kind of test involving them? If the answers to these three questions are "it still works", "it still works", and "not important", then it suggests lineups may be inherently problematic. In the course of an ordinary day in city most people are exposed at least briefly to a large number of strangers. If a witness only got a brief look at a criminal, and then is shown a lineup that contains a suspect that is not the correct criminal but is someone that the witness happened to see a few days ago out and about in the city without thinking about it, is there a high chance that person will be picked in the lineup? It seems that with human memory, three things are true. (1) We actually remember a lot more than we think we do, (2) we are not very good at accurately consciously recalling it, and (3) when we recall it we may unconsciously update the memory to better match the inaccurate version we recalled. Unfortunately most use of memory as evidence depends on accurate conscious recall, and requires the witness to recall it multiple times between the events witnessed and their testimony at trial, giving many opportunities to mutate the memory to match the prosecutor's narrative. If I am ever on a jury I am going to greatly discount eyewitness testimony unless the side the testimony is for explains what steps they took to address these problems. [1] Lionel Standing (1973), "Learning 10,000 Pictures", Quarterly Journal of Experimental Psychology 25, 207–22. [2] Timothy F. Brady, Talia Konkle, et al. (2008), “Visual Long-Term Memory Has a Massive Storage Capacity for Object Details,” Proceedings of the National Academy of Sciences 105, no. 38, 14325–29.
mythrwy 9y ago
How does scientificamerican.com know how reliable I think eyewitness accounts are? Why do people and publishers use that line? It can't be accurate and I find it presumptuous and offensive. It's a form of emotional click bait. "You don't want to be wrong about something do you?? Click Here!" A nice calm "a lot more reliably than commonly believed" would be better.
olliej 9y ago
I few people have already commented on needing to properly isolate the memories, and how difficult it is, and a few others have said that’s better than nothing (and I agree replacing the officers who know which is the suspect in a line up with those who don’t is a basic but massive improvement). The problem is how easy /accidental/ compromise is, even if you’re trying to avoid it. Just imagine you reading back their testimony to verify it is correct and misspeaking, that is enough. That said the wonders of CSI, etc have gone a long way to over represent the fallibility of people and to over promise the capability of science/equipment/magic. #MakingGUIinVBtoBacktrackTheKillersIP
mannykannot 9y ago
Given the title, the article presents little evidence as to the achievable reliability of eyewitness memory. The closest it comes is the experimental result that identification by eyewitness is 97% accurate in those cases where the witness is very confident of the identification. I am skeptical as to whether witnesses to actual crimes would accurately estimate their confidence level in actual line-ups. Nevertheless, the suggested reforms are well-argued and seem to be sensible.

38 points by DamonHD 9y ago

The problem still stands. You can't really rely on it to convince someone wrongly.

This gets complicated at public events where something goes wrong and the reported story is not consistent with the facts. Bystanders can witness the event with their own eyes and then go home to watch the news, to then believe the news-worthy version of the event. A spiral of silence sets in so all observers of the original event fall under the same spell and assume they are wrong and adjust their view to the world view rather than what they saw with their own eyes. At public events you also get a completely different view of what happened depending on whom you ask. Not everyone gets the complete story or picks up on all the clues. People bring their own expectations of what they expect to see and have cognitive dissonance to make it so they do see what they see and not necessarily what happened. The JFK incident was in this realm of unreliable eyewitness memory, those that thought they saw and heard the second shooter changed their view over time to align with the accepted story. This is how it works.

teddyh 9y ago

The article is good but the title is false. Eyewitness testimony is, as a rule, always filtered through layers of perception contaminated by various unchecked influences, which is largely the point of the article. The fact that properly isolated witness testimony is surprisingly reliable has no practical benefit for anyone, since, in practice, no witness testimonies are properly isolated.

MarkMc 9y ago
> no witness testimonies are properly isolated. Can't we make them properly isolated? The article says you need 3 things for reliable witness testimony: 1. Do not repeat the lineup, 2. Ensure the suspect doesn't stand out, 3. Record the confidence of the witness when identifying the suspect. If law enforcement follow these 3 rules and the witness identifies the suspect with a high degree of confidence, it seems the identification is very reliable. That would be a practical benefit.
- microcolonel 9y ago
  The witness is also susceptible to suggestion. If they are reached over the telephone, or have a conversation with a family member, or the police accidentally (or intentionally) suggest something in an interview, then they are tainted. If the defendant is resourceful, then the chance of any of these happening blows up.
abecedarius 9y ago
In the last decade or two blinded lineups -- where the police officer with the witness doesn't know who is the suspect -- became the standard in a fair number of U.S. states. You can take this pessimistically: wtf, it took this long to introduce some basic epistemic hygiene? But it also shows progress does happen. "No practical benefit for anyone" is just giving up.

tzs 9y ago

A problem with lineups may be that memory is too good. There is an experiment that has been done several times that shows how surprisingly good our memories can be in some ways. The experiment is to show someone a very large number of images or normal things, showing them each image for a short time. The images might be things like 5 apples on a table, a man petting a dog, a couple of people talking, and that kind of thing...just normal things you might see every day. Then later, sometimes days later in some versions of the experiment, you show those people pairs of images, where one of the images was from the set they were shown earlier, and one is not. You then ask them which of the two images they have seen before. This has been done with sets as large as 10000 images, and people get 80% or more correct two days later [1]. There was a version where in the identification portion the two images shown were one from the set they had seen earlier and one that was similar. Examples might be a stack of $1 bills and a stack of $5 bills, or a bell with a narrow handle and a bell with a wide handle. Even with these similar images people did well. The one I've seen discussed used 2500 images and 2500 similar images, and people got 90% correct [2]. That particular one tested shortly after the people were shown the images. I wonder if this works when instead of the recall test being one image you have seen before against one that you have not, it is one you have seen against several you have not? And I wonder if it works when it is something live you are seeing instead of images? And how important is it that when shown the initial set of images people know they are being shown images and are going to later undergo some kind of test involving them? If the answers to these three questions are "it still works", "it still works", and "not important", then it suggests lineups may be inherently problematic. In the course of an ordinary day in city most people are exposed at least briefly to a large number of strangers. If a witness only got a brief look at a criminal, and then is shown a lineup that contains a suspect that is not the correct criminal but is someone that the witness happened to see a few days ago out and about in the city without thinking about it, is there a high chance that person will be picked in the lineup? It seems that with human memory, three things are true. (1) We actually remember a lot more than we think we do, (2) we are not very good at accurately consciously recalling it, and (3) when we recall it we may unconsciously update the memory to better match the inaccurate version we recalled. Unfortunately most use of memory as evidence depends on accurate conscious recall, and requires the witness to recall it multiple times between the events witnessed and their testimony at trial, giving many opportunities to mutate the memory to match the prosecutor's narrative. If I am ever on a jury I am going to greatly discount eyewitness testimony unless the side the testimony is for explains what steps they took to address these problems. [1] Lionel Standing (1973), "Learning 10,000 Pictures", Quarterly Journal of Experimental Psychology 25, 207–22. [2] Timothy F. Brady, Talia Konkle, et al. (2008), “Visual Long-Term Memory Has a Massive Storage Capacity for Object Details,” Proceedings of the National Academy of Sciences 105, no. 38, 14325–29.

mythrwy 9y ago

How does scientificamerican.com know how reliable I think eyewitness accounts are? Why do people and publishers use that line? It can't be accurate and I find it presumptuous and offensive. It's a form of emotional click bait. "You don't want to be wrong about something do you?? Click Here!" A nice calm "a lot more reliably than commonly believed" would be better.

olliej 9y ago

I few people have already commented on needing to properly isolate the memories, and how difficult it is, and a few others have said that’s better than nothing (and I agree replacing the officers who know which is the suspect in a line up with those who don’t is a basic but massive improvement). The problem is how easy /accidental/ compromise is, even if you’re trying to avoid it. Just imagine you reading back their testimony to verify it is correct and misspeaking, that is enough. That said the wonders of CSI, etc have gone a long way to over represent the fallibility of people and to over promise the capability of science/equipment/magic. #MakingGUIinVBtoBacktrackTheKillersIP

mannykannot 9y ago

Given the title, the article presents little evidence as to the achievable reliability of eyewitness memory. The closest it comes is the experimental result that identification by eyewitness is 97% accurate in those cases where the witness is very confident of the identification. I am skeptical as to whether witnesses to actual crimes would accurately estimate their confidence level in actual line-ups. Nevertheless, the suggested reforms are well-argued and seem to be sensible.