Can you recognize someone from hundreds of feet away?

Innocence Project client Darrell Edwards was convicted of a New Jersey murder based, in part, on testimony from a witness who claimed that she saw him from 271 feet away on a dark night, when she was not wearing her prescription eyeglasses.  Scientifically, it is impossible to see someone well enough from that distance and make an accurate identification.

In 1999, Edwards was convicted of an execution-style shooting of a known heroin dealer in Newark, New Jersey. The Innocence Project is working to overturn Edwards’ conviction based on DNA testing of the murder weapon and a black-hooded sweatshirt that was worn by the perpetrator. The DNA analysis excluded Edwards. Central to the state’s conviction of Edwards was testimony from an eyewitness who saw the perpetrator from 271 feet away at night. In addition to the DNA evidence, the Innocence Project submitted to the court a new scientific study showing that she could not have identified him from that far away, even in broad daylight.

On an August evening in 1995, Errich Thomas was shot with a single bullet to the head as he was closing the Newark sandwich shop he owned. Two men—described later as a tall man wearing a black hooded sweatshirt, and a shorter man—were seen by several people on the street as they walked out of the shop after the murder, turned down a side street, and disposed of the sweatshirt and the gun in a trash can. The black hooded sweatshirt and the murder weapon were recovered by police and preserved as evidence.

Darrell Edwards had lived in the neighborhood where the crime occurred all his life. After he was arrested for the crime, it would take three years for his case to go to trial and four trials to convict him. The first two ended in mistrials, and the third resulted in a hung jury and the acquittal of a co-defendant. Although two witnesses who knew Edwards and saw the perpetrators close-up maintained with certainty that he was not one of the perpetrators, neither witness testified at the final trial.

The state’s case hinged in part on the eyewitness testimony of a heroin user who knew Edwards casually from the neighborhood. This witness said she was sitting on her front porch at around 9 p.m. and saw Edwards dispose of the gun and sweatshirt in the trash can. It was dark, she was not wearing her prescription eyeglasses, and the distance from her porch to the trash can was 271 feet—nearly the length of a football field.  In 2007 the witness recanted her identification of Edwards in a sworn affidavit saying that on the night of the crime she was high on heroin and had been drinking, and that she had been “just guessing” when she identified Edwards.

In its motion seeking a new trial, the Innocence Project cited a scientific study by Geoffrey Loftus of the University of Washington and Erin M. Harley of the University of California, Los Angeles, two national experts in the field of visual cognition and eyewitness memory. Their 2004 study “Why is it easier to identify someone close than far away?” shows how the human visual system becomes more and more unable to perceive and identify facial details as the face they are looking at moves further away – even if they know the person they are trying to identify.

The study consisted of experiments designed to test the hypothesis that the visual system is unable to “perceive and encode progressively coarser-grained facial details as the face moves further away.” The study used a mathematical model to ensure that the same special proportions in a close image were replicated at distances.  The study found that after 25 feet, face perception diminishes. At about 150 feet, accurate face identification for people with normal vision drops to zero.

The study used well-known celebrities in experiments, which helped determine whether knowing the subject aides visual identification at these distances. The study concluded that being familiar with a subject does not impact people’s ability to make an identification from such distances. 

A 2008 article in the journal

Law and Human Behavior

by Gary Wells and Deah Quinliven at Iowa State University describes experiments into how people suffer from “hindsight illusion” that leads them to believe they could see a face better than they actually did, once they are told it was someone they know. (This may have been a factor in the Edwards case, since the witness said an officer pointed to Edwards in the photo lineup and said he was the same man another witness had picked out.)  The authors of that paper concluded: “When people were told ahead of time the identity of the person observed from a distance, they estimated that the face was clear and recognizable at distances that simulated several hundred feet. When actually tested under conditions in which they were not told the identity ahead of time, however, performance reached zero at approximately 150 feet.”

More on Darrell Edwards’ case

Read the Loftus-Harley study, “Why is it easier to identify someone close than far away?”

See an abstract of the Wells-Quinlan article, “Suggestive Eyewitness Identification Procedures and the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later”