Skip navigation

Double Blind: Preventing Eyewitness Error

Police had little to go on. The only witness to Bullock’s murder – her 17-year-old friend, Tyeisha Powell – described their assailant as black, about 5-feet-10 or 5-feet-11, with dark brown eyes and a slight beard, according to press reports at the time.

Investigators used that bare-bones description to create a composite of the killer, and within weeks they had a tip. The sketch matched a 23-year old homeless drug user who had recently been released from county jail. His name was Malcolm Jabbar Bryant.

Detectives presented Bryant’s mugshot to Powell, together with images of five other men, in a photo array commonly known as a “six pack” lineup. She identified the suspect as the man who attacked them. Powell would positively identify Bryant once more – this time in court, during his trial for first-degree murder.

In spite of five alibi witnesses who insisted that he was with them at the time of the attack, Bryant was convicted and sentenced to life in prison based solely on Powell’s testimony. For the next 17 years Bryant maintained his innocence; and in 2016, DNA evidence would prove he had been telling the truth all along.

Powell had picked the wrong man.

Bryant was freed from confinement last May, and received a formal apology from Baltimore’s State’s Attorney. But the apology came too late. On March 8, 2017, less than a year after being released from prison, Bryant died of a stroke. He was 42 years old.

In April 2017, the Quattrone Center for the Fair Administration of Justice, at the University of Pennsylvania Law School, hosted a panel of leading experts in witness misidentification to discuss how procedural reforms could help prevent sending men like Malcolm Bryant to prison for crimes they didn’t commit.

The discussion was part of the Quattrone Center’s 2017 Spring Symposium, “Common Ground: Preventing Errors in Criminal Justice” – which brought together a variety of stakeholders to seek consensus on ways of improving the criminal justice system.

The panel dedicated to improving eyewitness identification was moderated by Amanda Bergold, a professor at Penn who studies the intersection of social psychology and the legal system.

Participants included Sgt. Paul Carroll – who spent three decades in the Chicago Police Department and now trains law enforcement on investigatory techniques; Karen Newirth, an attorney with the Innocence Project; Mark Larson, chief deputy in the Seattle District Attorney’s Office; and Gary Wells, a professor of Psychology at Iowa State University who has written extensively on the subject of witness misidentification.

Short of a confession, eyewitness testimony is recognized as one of the most powerful pieces of evidence that can be presented against a defendant accused of a crime. But as the panelists testified, witnesses often get it wrong.

Mistaken eyewitness identification is believed to be the largest single factor contributing to the conviction of innocent people. More than 70 percent of roughly 350 prisoners exonerated by DNA evidence were convicted based in part, or in whole, on the testimony of a witness.

But Wells says that’s barely scratching the surface.

“If you look, the vast majority of DNA exonerations involve sexual assault, but this is actually a very small number of eyewitness ID cases,” he said. “Most eyewitness ID cases are robberies. So the fact that these mistaken ID cases coming from DNA are in the hundreds, we’re missing every robbery case that pretty much ever existed. This is the tip of a very large iceberg.”

Thanks to advances in brain science we now know more about the mechanics of mistaken identification than ever before. Human perception is highly malleable, particularly during times of anxiety and stress. What people remember about a particular situation or event is easily influenced by the language others use to describe it, and scientists now say that the mere process of recalling an event can change one’s recollection of it.

Criminal investigators can turn these tendencies to their favor, knowingly or not, with something as simple as a strategically placed photograph or a disapproving facial gesture. But few detectives have received even rudimentary training in the science of witness memory recall.

“I’ve given over 400 lectures to police on investigation procedure, and when you go into a room and ask officers how many of them have had any training on eyewitness identification, you’re lucky if you get one hand go up,” said Carroll, who favors laws that require police to follow certain procedures.

In 1999, the year Malcolm Bryant went to prison, Carroll, Larson and Wells were part of a working group convened by the National Institute of Justice (NIJ) to explore eyewitness identification reforms.

They presented five recommendations for police when dealing with eyewitnesses to prevent misidentification. These included requiring law enforcement to be trained on the psychology of memory; providing standardized instructions to witnesses; employing “double blind” techniques in which both the witness and the administrator at a lineup are unaware who the suspect is; capturing identifications on video; and questioning witnesses on their confidence level immediately after an identification is made.

Research shows that jurors tend to equate confidence with credibility. And while there seems to be no direct relationship between confidence and accuracy, in one study a juror’s perception of eyewitness confidence accounted for 50 percent of the variance in a juror’s decision to believe a witness.

“When a confidence statement is asked, and we have that one record, we really know a lot more, we are preserving something that is forensically significant,” said Larson.

According to a 2013 survey of eyewitness identification procedures in police agencies across the U.S., a large number of agencies obtain confidence statements from witnesses. But the study – conducted by the Police Executive Research Forum (PERF) – found that most departments had not fully implemented the recommendations.

For instance, the vast majority lacked written policies for eyewitness identification procedures, and fewer than a quarter employed audio or video recording. Only a handful of departments used double-blind lineups.

“This is a staple idea in science,” said Wells. “You would reject manuscripts in scientific journals if they didn’t use double-blind procedures, but we’ve sentenced people to life in prison and worse without them.”

New Jersey became the first state to require double-blind administration of lineups in 2002. Since then more states have been taking it on themselves to pass laws reforming lineup and photo array procedures. Many are based on a set of five best practices published by the Innocence Project.

Last year Nebraska became the 18th state to pass a law establishing eyewitness identification standards. Legislatures in another seven states, including Pennsylvania, are currently considering eyewitness reform bills. And cities including Boston, Dallas, Philadelphia and San Diego have adopted eyewitness identification best practices.

“We need to address police practices on the front end but then we also need to ensure that judicial gate-keeping of challenged identification evidence is robust, and that fact finders are given appropriate context for weighing these factors,” said Newirth.

A big part of this involves courts delivering clear and concise jury instructions on the potential fallibility of memory.

“It’s highly important that jurors who weigh identification evidence do so with some context in how to weigh that evidence and how to evaluate it,” she said.

Newirth says courts also need to think more about what witnesses should be allowed to testify about, and how they are permitted to testify. The Innocence Project has been trying to get courts to take a more critical look at in-court identifications, which are most detrimental to a defendant.

“For decades jurists have acknowledged both the power of the in-court identification and also the theater of it,” she said. “They lack any probative value and can be extremely prejudicial.”

Courts in Massachusetts and Connecticut have begun to take steps to limit in-court witness identification.

Meanwhile, nearly 20 years after the first NIJ study on the issue, the federal government is once again taking an interest in eyewitness reform. In January 2017, the Justice Department issued a first-of-its-kind set of procedures on eyewitness identification.

The new protocols, which will apply to agents at the FBI, Drug Enforcement Administration (DEA), Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) and the U.S. Marshals Service, address the use of “photo arrays,” and are designed to ensure that law enforcement personnel do not consciously or unconsciously lead a witness. 


This article was originally published on April 26, 2017 by The Crime Report (; it is reprinted with permission, with minor edits.