If you were a prosecutor getting ready to go to trial, just about the best evidence you could hope to have, other than a full confession from the main suspect, would be an eyewitness account, from a victim of the crime or someone who saw it transpire, identifying the defendant as the actual culprit. What could be better? No complex, hard-to-follow forensic data or indirect circumstantial evidence — just a clear-eyed, confident witness willing and able to pin the crime on the defendant.

Well, perhaps you’ve heard of the Innocence Project, a long-running non-profit affiliated with Yeshiva University’s Cardozo School of Law; it pioneered using DNA testing to test convictions handed down before genetic testing became widely used. The non-profit’s efforts have exonerated well over 300 persons, 18 on Death Row at the time, who had been found guilty of crimes genetic evidence now proves they did not commit. Those exonerated had been in prison an average of 13 years before the Innocence Project secured their release.

You probably haven’t heard, however, of the leading factor behind those erroneous convictions. The Innocence Project says that in about 75% of cases where it has exonerated a wrongly convicted prisoner, prosecutors had presented eyewitness testimony identifying the defendant as the guilty party in gaining the eventually-overturned conviction.

How could so many cases with eyewitness identifications have been in error? It turns out, there are many ways eyewitness testimony can be untrustworthy. Though most jury members might not realize it, studies have shown the degree of confidence that eyewitnesses exhibit in presenting their accounts has little or no relation with their accuracy. Of course, some cases may involve honest mistakes by the eyewitnesses.

In fact, juries usually believe eyewitness identifications are more accurate than decades of extensive scientific studies have been shown them to be. Leaving aside physical factors capable of impeding accurate perception — amount of light, distance, time to observe an unexpected or chaotic event, or vision shortcomings — research has shown many other factors can affect perception or memory, producing unreliable recollections.

These include: trauma or stress caused at the time of the perceived threat or in recalling it; witnesses’ tendency, when a weapon is present, to focus on it rather than anything else; and even a documented “cross-race impairment,” which makes it more difficult for eyewitnesses to identify members of other races. Many of these studies were summarized in a National Academy of Sciences study issued last October on factors affecting reliability of eyewitness testimony.

That study also examined the ways eyewitnesses’ memories can be affected by the criminal investigation process. Police investigative techniques, like witness questioning and identification line-ups, can reshape witness perceptions, unless handled carefully. That’s why some states and cities have adopted research-validated best practices for them, to prevent influencing witness recollections.

To date, 22 states have adopted laws or rules requiring police to videotape portions of eyewitness interrogations in which an identification is made, and 10 have put in place measures to regulate line-up procedures for identification witnesses (federal courts have had such guidelines for identification line-ups since 1999).

Gov. Cuomo supported a measure in 2013 that would have mandated that police tape witness identifications in major crime investigations and would have set procedures for identification line-ups, but the state legislature did not send it to him.