It's a standard of crime dramas: the witness' picking the suspect from a police lineup.
It's the foundation of the prosecution's case against two men on trial for their lives in the 2011 slaying of three people in a robbery at Lorena's Grocery, a corner store in Southwest Philadelphia.
On Thursday, lawyers for Nalik Shariff Scott and Ibrahim Muhammed called on a psychologist and expert in cognitive bias in the justice system to explain to the Philadelphia Common Pleas Court jury why witness identifications - however compelling - can be wrong.
Suzanne Mannes, an assistant professor of psychology at Widener University, testified that her review of about 30 studies of witness identifications shows that the identifications are accurate only about a third of the time.
Scott and Muhammed, both 35, were identified in court by sisters Jessica and Laura Nunez, who were 19 and 17 at the time of the Sept. 6, 2011, crime.
The sisters described how two armed men entered their family bodega and shot and killed their parents, Porfirio Nunez, 50; his wife, Carmen "Juana," 44; and his sister Lina Sánchez, 48.
Lawyers for the two men maintain that regardless of whom the sisters believe they saw, the prosecution is a case of mistaken identity. They have cited differences in the sisters' original descriptions of the gunmen, and suggested that investigators convinced them that Muhammed and Scott were the robbers.
Muhammed confessed to the robbery and killings after his arrest, although defense attorney Lawrence Krasner has argued that he has a history of schizophrenia and hallucinations, and was not medicated at the time.
The validity of suspect identifications by crime victims has come under increasing scrutiny over the last decade, and Mannes, questioned by Krasner's cocounsel Anthony J. Voci and Scott's lawyer, Jack McMahon, described scientific studies on the subject dating to the 1970s.
Mannes did not comment on the validity of the Nunez sisters' identifications. Instead, she described factors that she said studies have shown produce faulty identifications.
Moments of high stress and being menaced with a gun, for example, often undercut the accuracy of crime and suspect descriptions, Mannes said. Both were factors confronting the Nunez sisters.
So does the length of time between the crime and the identification, testified Mannes. The Nunez sisters were not asked to identify Muhammed until he was arrested five months after the killings, when he was identified as the suspect in another bodega robbery.
Mannes also cited studies showing that witnesses are more than 1.5 times more likely to mistakenly identify a suspect of a different race. The Nunez family was from the Dominican Republic; Muhammed and Scott are African American.
Mannes' opinions were strongly challenged by Assistant District Attorney Kirk Handrich, who is prosecuting the case with Assistant District Attorney Carlos Vega.
Handrich cited the bright store lighting at the time of the killings as well as the close view the sisters had of the gunmen. He noted that the sisters also grew up in a racially diverse community in the Dominican Republic.
Handrich also challenged the validity of the studies on which Mannes based her opinions.
Handrich read to Mannes from a 2014 report published by the National Academy of Sciences, "Identifying the Culprit: Assessing Eyewitness Identification."
The report cites several of the studies used by Mannes and reads, "None of the reviews met all current standards for conducting and reporting systematic reviews, and few met even a majority of these standards, making assessment of the credibility of their findings problematic."
Mannes, however, replied that the fact that the older studies do not meet current NAS standards doesn't mean the conclusions are invalid.
"Of course I believe more research is needed," Mannes said.