Discuss as:

Witness error: How mind tricks can put the innocent behind bars

Jon Adrian Velazquez, currently serving 25 years to life at Sing Sing prison for the murder of a retired cop, started writing letters to a Dateline producer in 2002. He claimed he was wrongfully convicted and challenged Dateline to find any evidence of his guilt. A 10-year investigation begins.  Luke Russert reports.

Updated 12:41 p.m. ET – Dorothy Canady said she would never forget the man who shot a retired New York City police officer, but at trial she identified Juror No. 6 as the assailant. Another witness to the crime said the attacker was a black man with braids, yet he picked an Hispanic man with short hair out of a photo lineup.

Though the jury laughed when Canady fingered one of their own, and despite other discrepancies among the accounts of other witnesses to the fatal shooting in a Harlem numbers (illegal gambling) parlor in 1998, Jon-Adrian Velazquez was convicted of second-degree murder and sentenced to 25 years to life in prison. Today, he is fighting to clear his name from a cell in New York’s infamous “Sing Sing.”

“The eyewitness misidentification is the central and critical reason for his wrongful conviction,” said Velazquez’s attorney Robert Gottlieb, formerly an assistant district attorney in Manhattan. “There is no other evidence in this case that could possibly be the basis for a guilty verdict other than the eyewitness identifications that were false. Unfortunately this story is not … so unique.”

Watch Part 1 of ‘Conviction,’  the story of Jon-Adrian Velazquez’s murder case


While hundreds of convicts have been freed from prison after being exonerated by DNA evidence in recent decades, many others who proclaim their innocence from behind bars don’t have that recourse because no such evidence exists. In many instances, their efforts to gain freedom boil down to their words against those of witnesses to the crimes they allegedly committed.

But in part due to the DNA exonerations, there is increasing concern about the reliability of witness identification in criminal cases. That, in turn, is forcing courts, state legislatures, police and district attorneys across the country to review convictions and make changes in the ways they collect what can be make-or-break testimony from witnesses to a crime.

Dateline NBC

Jon-Adrian Velazquez is serving a 25-year to life sentence in Sing Sing for killing a former New York City police officer. He maintains he is innocent and has been fighting to clear his name from his prison cell.

“We have come a long way since the days when people accepted without question an individual who would take the witness stand, point to the defendant and say, ‘I’ll never forget his face, that’s the person I saw,’” said Gottlieb. “It really was not until the advent of DNA analysis that we have been able to show that eyewitness identifications are one of the weakest forms of evidence.”

Conviction: Reporter's 10-year quest for answers in little-known murder case

The Innocence Project, a nonprofit group dedicated to freeing the wrongfully convicted through DNA testing and to criminal justice system reform, has helped win freedom for nearly 300 prisoners in 35 states -- including 17 who spent time on death rows -- in its 20 years of existence.

In 75 percent of those cases, the leading factor in their convictions was witness identification; in 36 percent of those cases, convictions were in part based on an identification made by more than one person, said Karen Newirth, eyewitness identification litigation fellow at the Innocence Project.

“Our DNA exonerations represent really just the tip of the iceberg … because they are the only cases where DNA exists in the first place,” she said.

Those cases include rapes, murders, robberies and other crimes. The Innocence Project said experts estimate that DNA testing was possible in only 5 percent to 10 percent of all criminal cases.

“Many of the kinds of crimes where eyewitness misidentification come up are excluded because they are not the kinds of cases where DNA would ever have existed,” Newirth said. “We can only say for certain how prevalent it is in our cases and then assume that that’s really just the tip of the iceberg.”

Dateline NBC

Velazquez writes in his prison cell in 2011. Three eyewitnesses who identified him have either recanted or said they're no longer sure he fatally shot a retired police officer in Harlem.

Memory is ‘nothing like a videotape’

Researchers have long studied how people create memories, how that works in witness identification and then how that information is used by prosecutors at trial.

“People often think that memory works like a videotape and in fact it’s nothing like a videotape,” said Newirth. “That sort of highlights both the problem of how memories are made, but also how people become convinced of the correctness of their memories.”

A witness’ perception of a crime can be affected by lighting, distance, stress and the race of the alleged perpetrator – especially if it is different than their own, Newirth said. They also can be influenced by the suggestions of law enforcement and other witnesses as they try to fill in the gaps in their memory of that event, she said.

Suggestions from law enforcement – whether consciously or not -- could include something as simple as a nod by an officer or a positive reaction even when a witness says someone in the lineup “looks a lot like the guy,” said Jennifer Dysart, an associate professor of psychology at John Jay College of Criminal Justice.

Other times, witnesses may make a relative judgment: “So relative to all the others, who is the most likely … to be the police suspect here?” Dysart added.

To protect against guessing or suggestion, some jurisdictions have opted for “blind administration” of a lineup – where the officer conducting it has nothing to do with the case and does not know who the suspect is. The officer also is instructed to tell witnesses that the suspect may not even be in the lineup and  record statements on how confident the witness is about their selection.

A black man with braids or a Hispanic man with short hair?

In the Velazquez case, the witnesses said the man who killed retired police officer Albert Ward was a black man with a light complexion and braids, a description that led police sketch artists to create this "wanted" poster:

Courtesy of Robert Gottlieb

A wanted poster featuring a police sketch of the shooter in the 1998 murder of Albert Ward.

Tips poured in. Three people told police they either recognized the man in the sketch as a man named Mustafa or had heard that he was the shooter. Others provided information on who they understood had ordered the hit and a suspected accomplice -- neither of which were Velazquez. But police honed in on Velazquez, now 36, after one witness, Augustus Brown, identified him after sifting through 1,800 mugshots.

“Every eyewitness … said the shooter was a male black with braids, so how in heaven’s name is the shooter ultimately said to be a male Hispanic with short cropped hair?” Gottlieb, Velazquez’s attorney, asked rhetorically. “Something went wrong in the process.”

The police station lineup that resulted in Velazquez’s identification as the suspected killer was equally “absurd,” he said.

“Everyone looks either Hispanic or Caucasian; there isn’t one male black with braids in that lineup, so you start off with a rotten lineup,” Gottlieb said. “And then you have eyewitnesses who identify an Hispanic male and each of those eyewitnesses were vulnerable individuals -- drug dealers, drug users, down and out individuals -- who were very vulnerable to police manipulation, police suggestiveness.”

Velazquez, who said he was at home speaking on the phone with his mother during the shooting, voluntarily went to the police station and was put in the lineup.

Three of the six witnesses who viewed that lineup chose Velazquez and three did not, including the woman who later picked out juror No. 6 at the trial. However, days later she returned to the police station and said she had decided it was the man in position two -- Velazquez.

The police lineup in the 1998 murder of Albert Ward. Three witness picked out Jon-Adrian Velazquez (#2 in the lineup) as the shooter.

The NYPD did not respond to an email seeking comment on their procedures in the Velazquez case.

With many witnesses, ‘rich possibility’ of contamination

Dysart, the psychology professor, said wrongful identification of a criminal suspect can snowball if police don’t prevent witnesses from feeding off one another’s memories, a phenomenon known as “co-witness contamination.”

“When you have a chance to speak with someone who also saw the same crime, it’s possible that you are going to incorporate something that they said into your own memory of the event,” Dysart said. “The more opportunities that witnesses have to discuss things together … the more likely the memory report will become very similar. … Memories are so fragile really and we can be influenced by so many things.”

A more recent case that highlighted the growing controversy over witness identification was that of Troy Davis, who was executed in September 2011 for killing a Savannah, Ga., police officer. He maintained his innocence until his death, and seven of the nine witnesses who said he was the shooter later recanted all or parts of their testimony.

“In Troy Davis’ case, you had a lot of marginalized people who were at the crime scene who allege that they were coerced into offering testimony that they later said was not true, that they did not see happen,” said Laura Moye, death penalty abolition campaign director of the human rights group Amnesty International.

‘The guy on trial I had never seen before’

There were recantations in the Velazquez case, too: Phillip Jones, who was at the numbers parlor with his brother, Robert, when the shooting occurred, said in an affidavit:

“I told the police this was the guy and I was sure, but this was not the truth. I felt pressured because the police were threatening to arrest me and my brother Robert for stealing money that Albert dropped on the floor after being shot. I was arrested some time after Albert Ward was killed and two detectives came to visit me upstate in Groveland Prison. The detectives told me they got the right guy and would help me get parole. I decided to testify at the trial because I felt pressured by the police. When I saw the deft. (defendant) in court, I looked in his eyes and knew I had picked out the wrong  guy, and the guy on trial  I had never seen before.”

Courtesy of Maria Velazquez

Velazquez with his girlfriend and two sons a month before he was arrested in 1998.

Despite this and other admissions, making the case to free Velazquez – a case in which there is no physical evidence that could prove his innocence -- will not be so easy, Newirth said.

“In cases like this where there’s no DNA, it’s incredibly difficult. I mean, I think that that’s a principle that the Troy Davis case really stands for is … courts necessarily treat recantations with ... great suspicion,” she said.

Moye of Amnesty said the bar for a reversal would be high.

At this stage, “You no longer have innocent until proven guilty, you’re pretty much guilty until proven innocent,” she said. “… How do you prove beyond … a shadow of doubt, you know, what people said they saw or didn’t see? It’s all so subjective.”

‘They didn’t commit the crime’

Due to the problems with wrongful convictions, states, courts, district attorneys and police have worked to implement changes. At least 13 states have enacted some form of legislation implementing many of the Innocent Project lineup reforms.

A key New Jersey Supreme Court decision in August 2011, State v. Henderson, which followed a report the high court commissioned to evaluate scientific and other evidence about witness identifications, concluded “that the current standard for assessing eyewitness identification evidence does not fully meet its goals.”

“Study after study revealed a troubling lack of reliability in eyewitness identifications. From social science research to the review of actual police lineups, from laboratory experiments to DNA exonerations, the record proves that the possibility of mistaken identification is real,” the court said. “Indeed, it is now widely known that eyewitness misidentification is the leading cause of wrongful convictions across the country.”

The court called for two remedies: Allowing defendants who can show evidence of suggestiveness a pre-trial hearing to explore those concerns; and developing enhanced jury instructions about witness identification – relevant factors and their effect on reliability --- for trial judges to use.

“I think the major kind of theme to take out of the Henderson decision in New Jersey is that courts are really beginning to re-assess … how they assess the reliability and accuracy of eyewitness evidence,” said Rebecca Brown, senior policy advocate for state affairs at the Innocence Project. “And I think that the Henderson decision really signals a shift in direction by the courts and … we’ve begun to see other courts kind of pick up on that.”

Not all courts agree: The U.S. Supreme Court ruled in mid-January that the due process clause did not require an inquiry into the reliability of a witness identification when it was not obtained under “unnecessarily suggestive circumstances by law enforcement.”

The lone dissenter in Barion Perry v. New Hampshire, Justice Sonia Sotomayor, wrote: “The Court's holding enshrines a murky distinction between suggestive confrontations intentionally orchestrated by the police and, as here, those inadvertently caused by police actions that will sow confusion. … And it recasts the driving force of our decisions as an interest in police deterrence, rather than reliability.”

“The majority … adopts an artificially narrow conception of the dangers of suggestive identifications at a time when our concerns should have deepened,” she added.

District attorneys in Chicago, Santa Clara county, Calif., Dallas and Houston have also created units to review questionable cases.

Dallas County was the first to do so in 2007, when the newly elected district attorney formed a four-person unit to review 500 cases where defendants claimed their innocence. Until then, authorities had routinely rejected all such claims, said District Attorney Craig Watkins.

“Basically, 90 percent of the first cases that we looked at turned out to be cases where the defendant was telling the truth, they didn’t commit the crime,” he said. “… We started looking at all cases. We started taking requests from defense attorneys. We started actually reading the letters from defendants.”

Witness identification was the No. 1 problem investigators came across in the initial cases, which had DNA evidence that could be tested, Watkins said.

“Everyone that had been exonerated had been wrongfully ID'd,” he said. “There was somewhat of a suggestive nature to pick a certain person and so we advocated first with the police departments to change their procedures and they did.”

Under Watkins’ tenure, 15 people have been exonerated as a result of the unit’s reviews. Another 12 were exonerated before then due to the passage of a 2001 state law that allowed convicted inmates to request post-conviction DNA-testing.

‘Our system’s broken’

Watkins said the reviews met with a public backlash from people who didn’t think that was his job and that he should focus on getting the bad guys.

Law enforcement had a similar reaction.

“But as we went along, I mean, they couldn’t deny it,” he said. “It was ... proven that our system’s broken and it’s broken in more than one way. This is not just from a prosecutorial standpoint, it’s from how we investigated cases, it’s from police officer misconduct, prosecutorial misconduct. It’s just from the culture of law enforcement.”

They’ve now started to work on cases where there is no DNA to test. Five of their 15 exonerations have been in non-DNA cases.

“Those are somewhat more difficult to do because it’s subjective, it’s not scientific,” but they’ve had the real perpetrators step forward or there was "overwhelming evidence of prosecutorial or police misconduct in the cases and so it’s been pretty cut and dry,” he said.

Changes ‘wholly unnecessary’

But Scott Burns, executive director of the National District Attorneys Association, said it would not be feasible for many of the country’s nearly 3,000 DA offices to have such units since about 85 percent of them have five or fewer lawyers.

He also argued that the witness identification issue has been “blown out of proportion when you look at the total volume of cases that are prosecuted nationwide -- hundreds of thousands.”

“It’s always easy to come up with … a case here or a case there and spotlight it, and then jump to the premise that eyewitness identification is inherently bad or unreliable, and I think I can speak for DAs in saying that just isn’t true,” he said, “… Eyewitness identification, while it’s always been an aspect of a case that has to be dealt with in a special fashion, it’s still an important tool in prosecuting cases.

“Do eyewitnesses get it wrong some times? Absolutely. And, I think DAs recognize that and in those cases it’s … not only our obligation, it’s our sworn duty to rectify that.”

Burns said his organization has not taken a position on procedural changes, such as those recommended by the Innocence Project.

“Some DAs think they’re great and others think they’re wholly unnecessary and costly and inefficient,” he said.

‘There are times that those people are in fact innocent’

New York County District Attorney’s Cyrus R. Vance Jr. took up his post in 2010 with a pledge to create a Conviction Integrity Program that “re-affirmed the traditions” of the office.

Since the program began, the unit has vacated one conviction in a case that’s now awaiting re-trial, rejected two others after investigations showed that the original convictions were sound, and is reviewing others.

Gottlieb asked the unit in October to review Velazquez’s case, but he said prior to publication that he has not yet received an official decision though the two sides have been meeting. The DA’s office said that the case is under review.

“This unit is a great innovation as long as it is implemented not for show but for real results. … it now really depends on the commitment of people who have the power to undertake this work objectively, no matter where it leads,” Gottlieb said.

“Every day of the week, people are arrested, people are locked up … and it turns out that there are times that those people are in fact innocent,” he added. “For the public that would be fearful that, ‘Jesus, we start opening the jails for people all these years later who say that they’re innocent … that’s a frightening thought,’ except if they realize that that person being locked up meant that the real killer was not … that should be much more frightening for people to accept than what would happen if we released Jon-Adrian Velazquez.”