The US Federal Bureau of Investigation admits its forensic experts gave flawed hair analysis evidence for decades. But it may not have been intentional. It may have been a case of "wishful thinking."
DW: Dr Itiel Dror, how reliable is hair analysis, compared with simple fingerprints at a crime scene, or DNA fingerprinting?
Dr Itel Dror:The issue that came out with the FBI
is not specific to hair analysis. All forensic domains, including fingerprinting and DNA, they all have a subjective element. And because they have a subject element of interpretation, they are not totally objective. The human is important in the judgment. And I think the problem comes up in other forensic domains, where experts overstate the evidence when they work in the adversarial legal system and they're trying to make a case against a suspect. And what happens is that they are affected by the context. Right now we are working to improve forensic judgment by making sure the experts focus on the scientific evidence - so they're not exposed to a lot of contextual information that biases their interpretation. And I think that was one of the problems with the hair analysis, but it also exists in DNA, fingerprinting, handwriting… where the examiners don't focus on the forensic science, but are affected and influenced by a lot of other irrelevant information which they know about the case, and that distorts their judgment, and then they come to court and overstate the evidence, rather than being focused on the specific evidence they can examine.
So let's unpack this: To what extent is hair analysis subjective?
How subjective depends on the specific case: In some cases the analysis is relatively clear cut. In those cases the subjectivity and interpretation is minimal. In other cases, which are more difficult, with more ambiguity, there subjectivity and interpretation are greater because the evidence is not clear cut. But when it is subjective, and subjectivity is almost always involved, then they are affected by context, then their interpretation can overstate the results because there's no objectivity instrumentation that makes it black or white.
Peter Neufeld of the Innocence Project says this overstating of the significance of similarities in hair samples would "[leave] juries with the false impression that hair recovered from the crime scene must have come from the defendant." He's suggesting that it was intentional on the part of the FBI. What's your take?
I would say most forensic examiners do not intentionally overstate the evidence. We talk about cognitive and brain processes. Because they are involved in the case, because they are exposed to a lot of contextual information, they actually see similarities beyond what exists. It affects the perception and the judgment. So I don't think it's a problem of forensic examiners intentionally doing that. If that happens, it's very rare. The problem is larger, and it's across the board, because it is done unintentionally. It affects every forensic examiner in every case when they are exposed to irrelevant contextual information. It's a kind of wishful thinking, or self-fulfilling prophesies, where they come to the case, many times before they see the forensic evidence, and already know what they expect to find, because they're exposed to contextual information, and that biases the mind when it's a subjective domain, and almost all forensic domains involve a subjective element.
So how can we remove the dimension of human error?
I don't know if we can totally, but we can minimize it, and in a relatively easy way. For example, we have a process called "linear sequential unmasking." [It] aims to do something very simple, which is to say to the forensic examiners "Do not be exposed to irrelevant information." There is no reason for you to know if eye witnesses identified the suspect, if other forensic evidence incriminates the suspect, or what the detective thinks - and a lot of other information which is irrelevant for the act of forensic examination. Whether you're doing hair analysis or DNA or fingerprints or handwriting, blinding examiners to irrelevant contextual information is very easy and will achieve a very big step in minimizing the kind of errors of contextual bias.
Dr Itiel Dror
is a senior cognitive neuroscience researcher at University College London. He is interested in how the brain and cognitive system perceives and interprets information, and was cited by the Scottish Fingerprint Public Inquiry Report and in the American National Academy of Science Report on Forensic Science. The original story surfaced in a Washington Post report.