We all rely on experts to make correct and consistent decisions: from medical professionals to fingerprint experts. We’d like to think these decisions are unaffected by their often highly charged emotional contexts, but is that really the case?
On the 11 March 2004 thirteen bombs were loaded onto commuter trains at Alcala de Henares station, 40km outside Madrid. Not long after, ten of them exploded killing 191 people and wounding more than 1,700. Using a fingerprint found on a plastic bag filled with detonators, the FBI, with automated computer matching and experienced experts, obtained and confirmed a positive identification with a US citizen, Brandon Mayfield. Mayfield, a recent convert to Islam, appeared to fit the bill perfectly and he was immediately arrested and imprisoned.
Nevertheless, two weeks later, Mayfield was released after Spanish police announced they had caught the real owner of the fingerprint. Mayfield was completely innocent, having simply been the victim of two misfortunes: firstly to have a fingerprint similar to the bomber and secondly to already be in the US fingerprint database (currently totalling around 10 million entries). So, is this an isolated case, or are there systematic problems in the way fingerprint experts work? Dr Itiel Dror, provides challenging evidence of systematic problems.
Initially, Dror, Peron, Hind & Charlton (2005) examined contextual effects in a student population. It was found that participants cued with emotionally charged contextual information, such as gruesome crime scene photos, were more likely to indicate a match between ambiguous fingerprints than uncued control participants.
But, more impressively, using a within-subjects design, Dror & Charlton (2006) actually re-presented experienced fingerprint experts with cases in which, five years previously, they had confirmed positive matches. This was carried out covertly so the experts themselves were in their normal work environment, and were unaware they were examining prints they had previously analysed. Contextual information was then provided to the experts in order to attempt to replicate the findings of the earlier study in a naturalistic setting. In an admittedly small sample size (n=6), two-thirds of the experts made decisions inconsistent with their own previous judgments.
So, what’s going on here, is fingerprinting fundamentally flawed? According to Dr Dror: No. In this BBC Newsnight interview, Dr Dror points out that while fingerprint evidence can sometimes be criticised, it is significantly more reliable than eyewitness testimony, which can be notoriously shaky. It is certainly not the case that fingerprinting is a fundamentally flawed process. Instead fingerprint agencies around the world should take into account new findings from cognitive science in the recruitment, training and procedures of experts.
At present many agencies give fingerprint experts contextual information about the crime, which may, clearly, bias their decisions in marginal cases. One simple recommendation would be to just provide experts with the fingerprints and no other contextual information.
Perhaps if the FBI fingerprint experts had been unaware the print they were attempting to match was allegedly one of the Madrid bombers, they wouldn’t have pronounced Brandon Mayfield a perfect match. As a result, major embarrassment and law suits would have been avoided.
Of course, as a result of this case and others like it blame has been apportioned onto individuals, where in fact it lies diffuse throughout the system. A system that, if these findings are further replicated, clearly needs a helping hand from cognitive psychology.
Acknowledgement: This post is based on a presentation given by Dr Itiel Dror on fingerprint identification at a UCL Departmental seminar. My thanks to Dr Dror for an enlightening talk.
Dror, I.E., Peron, A., Hind, S., & Charlton, D. (2005). When emotions get the better of us: The effect of contextual top-down processing on matching fingerprints. Applied Cognitive Psychology, 19(6), 799-809. [Abstract | PDF]