In downtown Des Moines several years ago I witnessed a woman fatally injured by a car as she crossed the street. I was an ideal witness. I had an unobstructed view of the accident, which had happened directly in front of me. But I learned later that, when police questioned me, the only accurate things they learned from me were my name and address. To my embarrassment, my recall of the material facts – whether the victim was in the crosswalk when she was struck or the location of the car – was mistaken. My experience made me an instant convert to skepticism about the accuracy of eyewitness testimony.

The New Jersey Supreme Court recently added its considerable weight to skepticism when it cited the large and growing body of research pointing to the untrustworthiness of what people claim to have witnessed. I once attended a meeting of judges addressed by one of the top researchers on eyewitness fallibility. The judges were on guard because they were prepared for a demonstration of how they could be manipulated; even so, in a test, a large number of highly experienced judges were mistaken about what they believed they had witnessed.

Circumstantial evidence is widely derided as unreliable. By contrast, eyewitness testimony often is cited as the kind of direct testimony juries should have faith in. Therein lies the crux of the problem; eyewitness testimony appears to be credible, but the human eye and brain can, in fact, play tricks that make them dangerously untruthful.

Smart police departments are borrowing from the research to weed out unreliable evidence. For example, they are barring officers who investigated cases from conducting lineups in the arrests they make to prevent them from subtly influencing the identification choices witnesses make. But police often resist the idea that they may be responsible for misidentification. It doesn’t help that much of the research about wrongful identification has been done by academics easily dismissed as ivory-tower types without real-world experience.

The New York Times reported recently that only about 25 to 30 percent of police departments have revised procedures to improve lineups. The Times noted that, of the 75,000 witness identifications each year, about a third may be in error.

It’s inexcusable that many police knowingly use identification methods that can produce false accusations. Every police reporter in the country ought to bone up on the witness-identification research and to question whether their local police departments are familiar with the research and, if it is not being utilized, why not.