Most people believe that forensic science is an exact science that can always find the right perpetrator. But is that actually true? Today’s expert explains the reality behind fingerprint analysis, DNA, and other forensic sciences.
Professor Brandon L. Garrett shares how a variety of cognitive biases and our basic psychology can cause crime labs to make devastating errors that destroy lives. Not only can innocent people be falsely imprisoned, but the guilty can go free and possibly commit future crimes. Listen Now!
Brandon L. Garrett is the L. Neil Williams Professor of Law at Duke University School of Law where he directs the Wilson Center for Science and Justice. His most recent book is “Autopsy of a Crime Lab: Exposing the Flaws in Forensics,” available from University of California Press. His previous books include “Convicting the Innocent: Where Criminal Prosecutions Go Wrong,” “Too Big to Jail: How Prosecutors Compromise with Corporations,” and “End of Its Rope: How Killing the Death Penalty Can Revive Criminal Justice.”
Gabe Howard is an award-winning writer and nationally recognized speaker who lives with bipolar disorder. He is the author of the popular book, Mental Illness is an Asshole , available from Amazon; signed copies are available directly from the author.
To learn more about Gabe, please visit his website, gabehoward.com.
Producer’s Note: Please be mindful that this transcript has been computer generated and therefore may contain inaccuracies and grammar errors. Thank you.
Announcer: You’re listening to Inside Mental Health: A Psych Central Podcast where experts share experiences and the latest thinking on mental health and psychology. Here’s your host, Gabe Howard.
Gabe Howard: Hello and welcome to this week’s episode of Inside Mental Health: A Psych Central Podcast. I’m your host, Gabe Howard, and I want to thank our sponsor, Better Help. You can get a week free by visiting BetterHelp.com/PsychCentral. Calling into the show today we have Brandon L. Garrett. Professor Garrett directs the Wilson Center for Science and Justice at the Duke University School of Law and is the author of Autopsy of a Crime Lab: Exposing the Flaws in Forensics. Professor Garrett, welcome to the show.
Brandon L. Garrett: Thank you. Thank you so much for having me.
Gabe Howard: I have this overwhelming desire to title this episode like CSI: Inside Mental Health, because forensic science is everywhere. Now on television, forensic science is perfect. It’s unwavering and it’s unbiased. But judging by the name of your book, I suspect that the reality is very different.
Brandon L. Garrett: Yeah, so, you know, CSI shows are fun to watch. Someone kinda runs some stuff from the crime scene in to the computer and like the screens all blare match and it’s a match. You know, in some ways, actually, there’s something realistic about the CSI show, something realistic about the fakery. That’s kind of what experts say in the courtroom. In criminal cases, they say it’s a perfect match. We don’t make errors. They don’t actually run it through computers. They don’t actually use a lot of technology. But that same kind of bravado, maybe without the leather gear and sunglasses and cool hair, that same bravado is responsible for a lot of problems in our system where people assume that these experts can make a perfect match because they’ve been talking about making perfect matches for decades and it just doesn’t work like that.
Gabe Howard: There’s a lot of interest in forensic science right now, and I’m going to say that it’s probably driven by popular culture and TV shows like Law & Order and CSI.
Brandon L. Garrett: It’s not just the media that’s making forensics stuff popular, there’s been a revolution in the technology that can be used in criminal cases. When I started my career as a young lawyer, I represented people who had been wrongly convicted and freed by DNA evidence and DNA evidence worked a real revolution in criminal justice. It really is incredible. Cold cases that are decades old can be overturned using this new technology. DNA came from the outside. It was developed by geneticists, mathematicians, hard scientists. Billions continue to be invested in genetic technology because it’s revolutionizing medicine. What DNA did, though, was it shed a harsh light on the more traditional forensics and, most criminal cases, you can’t do a DNA test. There just isn’t any stuff to use DNA on. The bulk of what crime labs do is still like fingerprinting or tool marks and firearms looking at patterns side by side, blown up images or under a microscope, looking at patterns and deciding whether you think they match. None of the statistics, none of the technology of DNA. So when DNA came along, everyone was like, oh, wait a minute, the stuff we thought was infallible, like there are no statistics, they’re no databases. What are we relying on here? Why do we still use this antiquated stuff? I did a lot of research on the cases of people freed by DNA. Many of those people were freed by DNA technology, but they were convicted in the first place based on faulty forensic evidence, whether it was a bite mark comparison or a hair comparison, in some cases fingerprint comparisons. Good forensic set them free, but really problematic, traditional forensics caused them to spend all those years in prison in the first place for crimes they didn’t commit.
Gabe Howard: It actually sounds like you’re saying that fingerprint matches are not dead on. I thought the fingerprint matches and DNA were both outside of the realm of bias or preference or psychology. I mean, it matches or it doesn’t.
Brandon L. Garrett: Yeah, I thought that, too, you know, early on in my practice, I figured, OK, we know that there’s some junkie forensics these days because we’ve heard some of the stories about like the blood spatter evidence, bite mark or like hair comparison or like the ripped duct tape comparison. There are some cheesy forensics out there which people are increasingly aware aren’t real science. But fingerprints? Come on. Early on, they called DNA testing DNA fingerprinting because fingerprints are like perfect and reliable and unique. And we now know that, well, no, if you look at your fingerprints, there’s a lot of detail there. There’s a lot of information. And if you gently roll from side to side and produce a fingerprint for employment or to get a passport or something, there’s a lot of information, hundreds of points, lots of detail. Problem is, criminals are not trying to neatly roll their print on a nice surface that preserves it well. So any prints from a crime scene may be really, really smudged. You can barely tell that it’s a fingerprint, much less, you know, is it a whirl or a loop or no details, really. You’re comparing something which is super distorted and smudged with a pristine print. And people assume that fingerprints are unique, it hasn’t been tested, but even assuming that pristine prints are unique, how many super smudged, tons of information missing, messy prints could coincidentally look like each other? We don’t know. It hasn’t been tested. We don’t know how rare or common it is to have any particular features in a fingerprint. And these patterns are examined side by side on a computer screen by people who look at them and try to make a call.
Brandon L. Garrett: But is there like a rule? Like what do you measure? How many points do you have to observe in order to say, oh, yeah, they came from each other? How many differences do you have to observe to say, oh, no, no, no, this could not have been that person’s print? There are no rules. It’s a very open ended process based on someone’s training and experience. In other contexts where we’re counting on someone’s training and experience to make momentous decisions, we test them to find out how good they are. People look at slides to decide whether samples are cancerous and they’re looking at visual information to make an important call which can have life or death consequences. They test those people to find out did you correctly call it? Were there cancerous cells on the slide or not? Unfortunately for a lot of these forensics that involve the same thing, looking at lots of visual stuff all day, you know, how good is the person at the job? They’re not actually tested under meaningful conditions. And so until recently, we knew very little about what the error rates were. What we have known, though, is that some really high profile mistakes where they’ve gotten it wrong, even really experienced fingerprint examiners. And that’s really, really shocking to people.
Gabe Howard: It’s interesting that you say made a mistake because mistakes are just part of life, everybody has made a mistake even in things that they excel at. We see that in professional sports, you know, top athletes, etc. But
Brandon L. Garrett: Right.
Gabe Howard: You suspect that it’s not so much a mistake as a cognitive bias. Obviously, the fingerprint folks, for example, work for the crime lab. The crime lab is part of the state. Police officers are part of the state. So you’re running this on behalf of your coworker. Your coworker drops it off and says, we know that these belong to Gabe or even just an accident. I’m sure that this is right. Compare it to this. I don’t want to create this idea of corruption, etc. It’s not that right. It’s just undue influence, accident, being pretty sure. It’s just being human. It’s accidental, yet devastating.
Brandon L. Garrett: Well, it would be pretty wild, right, if someone came up to you and said, look, we all try to do our jobs, well, no one’s perfect. Except me. I’m perfect actually. I do my job right. I never make mistakes. And that’s why you should rely on everything that I say. It’s a con, like it can’t be perfect at everything. Fingerprint examiners and a host of other forensic disciplines, they would come into court and say, I have a zero error rate. There is no possibility of error. I am an expert and I’m never wrong. That was a crazy thing to say, particularly when there were no studies supporting it and where error rates hadn’t actually been studied. These were bold claims that were totally unsupported and don’t make any sense because even using a ruler, even measuring something, there’s uncertainty. Right? You may not be lining up the ruler exactly right. There’s a limit to how precise your eyesight can be. There’s uncertainty doing anything, even when you’re using a really precise measurements. But when it’s not precise at all, when you’re just kind of making a call based on your judgment? Of course, there’s an error rate in many of these forensic disciplines. They claim to be perfect. They claim to have zero error rates. It was horribly prejudicial in court. Jurors were, wow, this is incredible.
Brandon L. Garrett: It’s really impressive to hear someone who’s really, really confident. We can all be biased by things around us. Any time we make a judgment, we rely on shortcuts. And sometimes those shortcuts really help us out to make our life more efficient. The stakes are really different in a criminal case, obviously, where erring on the side of calling a match can mean that an innocent person could spend years in prison. And if you work for law enforcement and you assume that when you get evidence submitted to you, it’s because they’ve arrested someone for a serious crime and maybe you even know what the crime was, maybe even getting calls from the detective saying, hurry this up. This is a seriously dangerous person. We need expedited results here.
Gabe Howard: One of the things that I’m thinking about while we’re talking about this is we’re really discussing this in the abstract. Do you have a specific case where psychological factors played a role in a forensic error?
Brandon L. Garrett: Sure, sure, lots of examples. One really well-known example that I walk through in the book is the case of Brandon Mayfield, and I think it’s a good one because it involves fingerprinting. We all have long assumed that fingerprints are unique and that patterns stay with you since birth and fingerprints are detailed. Highest profile fingerprint error in history occurred in the case of the Madrid terror bombing, and they found a plastic bag which had some undetonated explosives and there’s a latent fingerprint on the bag. FBI, eager to cooperate in this international investigation, ran the print through the interconnected system of fingerprint databases. And you know, one sort of CSI conception people have is that, well, maybe individual people are biased here and there, but if you’re using something computerized, computers only help, they make it more objective. Well, what these databases do, if they’re looking through tens and tens and it’s over, you know, hundreds of millions of prints in these larger interconnected databases, what is the job of the algorithm? The algorithm’s job is to pick prints that most closely resemble your crime scene. And so what they’re going to do is they’re going to make the job really, really challenging. Right? Because if it’s working well, it’s going to generate a list of 10 or 20 prints, which look a lot like the one from the crime scene. These algorithms, they can’t find a match, but all they do is select candidates.
Brandon L. Garrett: And these are candidates that look a lot like each other. And the bigger the database, the more you’re going to have prints that very, very closely resemble each other. And here they’re looking at a print from a crime scene which is missing a lot of information. You only have a little portion of the picture, but you’re comparing it to all these other different pictures. It’s a cognitively challenging thing. It’s also tiring. You’re going through one print after another print and all these details and your eyes are getting tired. Your mind is getting tired. Mayfield’s print was the fourth one that came up. So it wasn’t the one that the algorithm thought was maybe the best match. But again, these things aren’t perfect. And the examiners were doing something that it’s a different kind of cognitive bias, it’s context bias. So their practice at the time was to look at prints side by side. Now, if you look at things side by side, you may be more likely to find similarities and not pay a lot of attention to differences. The same way, you look at couples and the more you see them next to each other, the more you think that they look alike. You look at dogs and their owners and they look more and more alike over time or those kids’ challenges where there’s like two scenes of a playground and you try to find the similarities and the differences.
Brandon L. Garrett: And it’s really hard to find the differences because you look at them next to each other and both playgrounds look the same and you don’t see that, you know, in one tree there’s a squirrel and the other tree there is a pelican or whatever. There’s the circular reasoning process that took over in this high profile terror case where they were looking at Mayfield’s print and looking at the print from the database back forth, back forth. And they actually didn’t notice several differences as they were going back and forth. But these very experienced FBI examiners discounted them because of the circular reasoning phenomena. And unfortunately, there was more. They had other types of bias. So the first person that looked at the prints passed it on to a second person. A high profile situation, they actually had it reexamined by a third. But the second and third person knew what the first had found. So they weren’t independently going through all the candidates in the database and having no idea whether any of them would be deemed to be a match. They saw the work of the first person, so they weren’t really doing an independent reexamination and their reasoning may have reinforced each other. There was another source of bias. He was Muslim and he was a lawyer in Portland. He’d never even been to Spain. But the investigators start to suspect him because we have a potential hit on someone who is Muslim and maybe has a secret terrorist identity.
Brandon L. Garrett: And that may have biased the investigation, too. And so you had this outside information. You had the context in the way that the prints were presented to the examiners. You had a crime scene print that was ambiguous. There’s a lot of information missing from it. And if you have something ambiguous, it’s more easy for sort of your own biases to fill in the dots and ignore problems. And then you had these three people reinforcing each other, kind of a groupthink where the number two and number three all look at what number one did. And it all combines to make them one hundred percent certain that his print was the one from the Madrid bombing. All three of them are wrong. A couple of weeks later, the Spanish authorities located an Algerian terrorist who really was in Spain and really was part of a terrorist cell. The FBI apologizes to Mayfield. He’s released from detention. He’s no longer a material witness detainee. He’s no longer facing the federal death penalty. He’s released, apologizes, but it went so terribly wrong with three experienced examiners, not at some hokey local police lab, but at the FBI in a case that could not have been higher profile. And that caused a lot of soul searching in forensics because it brought to light all the ways that biases can cause terrible, terrible errors even in fingerprinting.
Gabe Howard: And it’s exceptionally important to point out that while this particular case had a quote unquote, happy ending, detained for a few weeks, released, civil settlement, some people will hear that and say well, see the system worked. It took a little longer, but it worked. It’s very important to understand that that’s not always the case. People spend 20 years in prison. People die in prison and are exonerated after their death. This happens a lot more than we’re probably comfortable admitting. Is that correct?
Brandon L. Garrett: It is. I mean, we will never know how often it happens, you know, when the FBI did an audit after the Brandon Mayfield case did the FBI, go back and look at, OK, maybe the problem was that there was just a single latent print in this case. Maybe that leaves extra room for bias? Maybe we should go back and review all of our single print cases? That would be a lot of cases. They didn’t do it. That said, the FBI did go back when they found that there are some examples of their examiners testifying in a really unscientific, hyped up, exaggerated way about hair evidence. And everyone agreed, going back decades that you really can’t say that hairs match. Hairs look really different even on one person’s head. All you can say is they look kind of similar or it’s possible this hair could have come from that person. But, you have these FBI examiners coming into court saying it’s very, very likely, it’s microscopically indistinguishable. I’ve looked at hundreds of thousands of hairs. I’ve never seen hairs that didn’t come from the person I thought it did, that kind of thing. The FBI did an audit of thousands of cases and found that more than 95 percent of the time their examiners gave this unscientific, flawed testimony.
Gabe Howard: We’ll be right back after these messages.
Sponsor Message: Is there something interfering with your happiness or preventing you from achieving your goals? I know managing my mental health and a busy recording schedule seemed impossible until I found Better Help online therapy. They can match you with your own licensed professional therapist in under 48 hours. Just visit BetterHelp.com/PsychCentral to save 10 percent and get a week free. That’s BetterHelp.com/PsychCentral. Join the over one million people who have taken charge of their mental health.
Sponsor Message: Hey everyone, my name is Rachel Star Withers and I live with schizophrenia. I’m also the host of Inside Schizophrenia, a podcast that dives deep into all things schizophrenia. Featuring personal experiences and experts to help you better understand and navigate schizophrenia, Inside Schizophrenia is a Psych Central and Healthline Media podcast and we are available right now on your favorite podcast player. Check us out!
Gabe Howard: And we’re back with Professor Brandon L. Garrett, author of Autopsy of a Crime Lab: Exposing the Flaws in Forensics. Let’s transition over to how we can fix this. I think the research is clear that something needs to change. What is that? What can we do?
Brandon L. Garrett: A lot of people need to step up. There is much more awareness in the forensic sciences that there is a psychology problem, that there are these cognitive biases. It’s not like a forbidden subject anymore. That said, you can’t just tell someone, don’t be biased in your work, right? You need procedures in place to ward off errors. One of the best models is what we do in clinical laboratories. People get tested routinely when they don’t know that it’s a test. The Houston crime lab is doing that now. Five percent of their work are these blind cases which look just like real cases. If you emailed the detective to ask questions, the detective answers them, but they are, in fact, a test and the supervisors know what the answer is. They can catch mistakes as part of just routine quality control. The idea is that error happens, but you catch it. You build in systems to improve your quality. It’s part of basic quality assurance. And you also don’t give the examiners all this extra information about is this a murder case? Did the guy confess? Does the guy have a criminal record? You don’t need to know any of that. Your job is to look at the prints or look at the tool marks. Labs can build in these protections. They just haven’t ever been really required to like real scientific labs have long been required to. Part of it is that sort of forensics grew up as part of law enforcement, often involving police station basements and police officers on desk duty who start looking at prints rather than walking the beat.
Brandon L. Garrett: And this whole sort of orientation towards law enforcement is really a legacy of the sort of growing pains of forensics in this country. It’s a scientific function, it should have scientific oversight, and you should have scientific procedures. And that has not quite happened yet. And part of it is it’s the funding is coming from law enforcement. We need to rethink the whole model. Beginning in 2009, when the National Academy of Sciences issued this landmark report calling for a change in forensics, the scientific community has been saying this needs to be independent, there needs to be federal regulation. There should be a federal agency, just like the National Institute of Health, to oversee standards for forensics. Hasn’t happened yet, but there has been really good progress, particularly with some labs getting scientific oversight boards, some labs starting to routinely test for proficiency, more understanding that cognitive bias matters. Things are, you know, modestly getting better, but that’s still pretty slow progress. And when you have people getting wrongly convicted, spending decades in prison for crimes they didn’t commit, when you have scandals and lab after lab in state after state, it’s sort of like how bad does the quality control problem have to be before a more aggressive action is taken?
Gabe Howard: Professor Garrett, I hope that people are understanding that there’s like a systemic flaw here, it’s not an example of people being bad at their jobs or evil or disliking people. It’s not that. It’s just a general misunderstanding of how the science works and how we apply it to law enforcement. And there just doesn’t seem to be this desire to correct those flaws. But the people themselves are, I’m trying not to say good or bad, but this isn’t a case of corruption. I guess I just want to be clear on that, because I fear that anytime this gets brought up. . .
Brandon L. Garrett: No, no, yeah, no. I think it’s really important. And actually, whenever I talk about this book, I need to make sure I think about that question because I don’t want to demonize people who work at crime labs. They really do have really hard jobs. It’s like if they’re not trained any different and they’ve been told that oh, you can just eyeball the stuff and call a match, and then you’re right. That’s the norm in their field. They’re given a problematic process and no research has ever been done to shed light on how reliable it is. And so they assume it’s great. And what are they supposed to do? That’s their job. Certainly, there have been some bad apple cases, right? There have been outrageous cases like the two drug examiners in Massachusetts that were taking the drugs themselves and falsifying their tests. And tens of thousands of cases had to be reopened. It is true that when you have bad quality control and bad regulation, the bad apples can do a lot of damage.
Brandon L. Garrett: The vast majority of forensics professionals are professionals and are trying to do their job well and are talented and they feel like they get an unfair, bad rap. Then again, they don’t have the systems in place to catch the bad apples so the bad apples do way more damage than they would. Like in a hospital. If someone didn’t do the blood test, the patient would still feel sick the next day. Whereas in forensics, if you do a false lab report, the person will be convicted and no one will know the difference. They need to have their own quality controls in place, and if they don’t, then the bad apples do horrible, horrible damage. You know, that’s one of the also the insidious things about cognitive bias that people who are well intentioned, trying to do the right thing and under the burdens that we all have in our jobs during the day where we have time pressure and is only so much work that we can do, only so much time, you can spend it on one task. People tend to assume that they’re doing the right thing until they’re told different. If we had supervisors that didn’t really look at our work carefully and told us, good job, good job, good job every day, we would be really proud of ourselves. And, you know, if you were a doctor and you were told that every patient you worked on was cured, you would feel like[l1] a master of the universe. If you got meaningful, precise criticism about your work, you wouldn’t quit. You would just be better informed about the limitations of your job and you’d approach your work with more humility. And that’s what we want. We want to empower forensics professionals to do their work more carefully and with more appropriate humility so that they’re not coming into court and saying they have a zero error rate, that they’ve never made a mistake in their life and that their discipline is perfect. You want them to be giving accurate information to jurors.
Gabe Howard: Especially when the detriment or the flaw, as we keep talking about, is actually that an innocent person goes to prison, loses their life, or even in the best case scenario, spends tens of thousands of dollars or hundred thousands of dollars on legal fees to prove their own innocence. Both of those things are devastating to somebody’s life. I know many people who were found not guilty that their marriages ended. You know, they started to have drug and alcohol problems. They went bankrupt. And this is the system, quote unquote, working. They were found not guilty. So everybody’s like, oh, well, it’s fair. Yeah, it’s not fair. Their life is over.
Brandon L. Garrett: Sometimes there isn’t even an innocent person who is wrongly convicted, but someone looks at the evidence and misses something and they fail to connect evidence to a guilty person. The case of Keith Harward, who was wrongly convicted based on six different dentists all saying the bite was his bite. Well, the actual culprit was and they looked at his teeth, too, and they said it didn’t match. And he went on to commit additional crimes. That happens in case after case after case where you have guilty people who go free because of forensic mistakes.
Gabe Howard: Professor Garrett, thank you so much for that. It is important to realize that obviously, if the wrong person is convicted for the crime, that means that the person who actually committed the crime is still out there available to, well, commit more crimes.
Brandon L. Garrett: Couldn’t have said it better. The human costs of getting it wrong are horrible and, you know, you talk to these exonerees and hear what it’s like to spend 10 years, 20 years, 30 years in prison. You get out and it’s like a time machine, like what is GPS? What is a smartphone? You’re used to being confined and having no choices. You know, you can’t give someone those years back. You just can’t. It is often the best formative years of one’s life. When you were first married or go to college or get your first job or first car, all those milestones gone, those young adult middle aged years gone, you can never give that time back. If it goes well in the courts, maybe the person will get some compensation from the state. You can’t give someone money to give them back their life. 20 years later, 30 years later, the people who made the mistakes, whether it’s the forensic examiners or the detectives, they may be gone, they may be retired, they may have passed away. There’s often no inquiry, no audit, no you know, no autopsy of the crime lab and no apology. Our legal system is designed to put people away. It’s not designed to open its doors and it’s not designed to ask questions later. That’s fundamentally unscientific, too. But that’s problematic psychology. If when you make mistakes, you don’t ask questions and do some soul searching. I hope that the stories of forensics go wrong cause people to do some of that soul searching and it has had an impact. People used to think that mistakes didn’t happen. Now we know that they do because the victims are walking among us, having survived terrible ordeals.
Gabe Howard: It’s fascinating, this thing that is just part of our society, it’s everywhere. Everybody knows about it, but there’s just so much that we don’t know that we need to consider and that as a society, we just need to expect better from our crime labs so that we are safe and justice is served.
Brandon L. Garrett: I agree.
Gabe Howard: Professor Garrett, thank you so much for being here. I highly recommend that all of my listeners go out and grab Autopsy of a Crime Lab: Exposing the Flaws in Forensics by Brandon L. Garrett. Professor Garrett, do you have a website that they can go to as well?
Brandon L. Garrett: I do. I have an author website, which is just BrandonLGarrett.com, and I’ll have some videos and more information about the book. There we’ll have a guide posted up there for book groups. We’ll have some videos from some of the people who have been exonerated and cleared in cases where there were errors in forensics. And so there’ll be some short video clips that you can watch, all that will be available.
Gabe Howard: Very cool, thank you so very much. And a big shout out and thank you to all of our listeners. Wherever you listen to this podcast, click the follow button. It’s absolutely free. Also, give the show a review, use your words and tell other people why they should be listening. And my name is Gabe Howard and I am the author of Mental Illness Is an Asshole and Other Observations, as well as a nationally recognized public speaker who would absolutely love to be at your next event. You can grab a signed copy of my book with free swag or learn more about me over at gabehoward.com. I will see everybody next Thursday on Inside Mental Health.
Announcer: You’ve been listening to Inside Mental Health: A Psych Central Podcast from Healthline Media. Have a topic or guest suggestion? E-mail us at show@PsychCentral.com. Previous episodes can be found at PsychCentral.com/Show or on your favorite podcast player. Thank you for listening.