Murder, eyewitness identification and the limits of human vision

Geoffrey Loftus’ latest research reads more like a murder mystery than a scientific paper.

The University of Washington psychologist’s new study opens with a savage beating and murder on the streets of Fairbanks, Alaska. It features cameo appearances by Julia Roberts and other celebrities. It ends with the conviction of two men based on the eyewitness identification of the defendants from a distance of 450 feet. And, in a post-script, an appeals court orders a new trial based in part on “scientific trials” and conversations conducted by jurors outside the courthouse, without the judge’s knowledge. In between, the limits of the human visual system are explored.

Loftus, who testified as an expert witness in the case, examines why it is easier to identify someone close up rather than at a distance in an upcoming issue of the journal Psychonomic Bulletin & Review.

“When you see anything at a distance the human visual system starts to lose small details. The greater the distance the coarser the detail you lose, ” Loftus said.

“At 10 feet you might not be able to see individual eyelashes on a person’s face. At 200 feet you would not even be able to see a person’s eyes. At 500 feet you could see the person’s head but just as one big blur. There is equivalence between size and blurriness. By making something smaller you lose the fine detail.”

Co-author of the study is Erin Harley, who recently earned her doctorate at the UW and is now a post-doctoral researcher at the University of California at Los Angeles. The research was funded by the National Institute of Mental Health.

The researchers conducted a number of experiments to establish the relationship between blurriness and distance. First, they started with very small, unrecognizable images of famous people such as Roberts, Michael Jordan, Jennifer Lopez, Bill Gates and President George W. Bush. Next, the researchers gradually made the images larger until subjects could identify each celebrity. They recorded the size at which each celebrity was recognized and converted this to a corresponding distance. Loftus and Harley did similar tests using initially blurred images of celebrities and gradually clarified them until the test subjects were able to recognize the celebrities. This time they recorded the amount of blurring that made a face unrecognizable. All of the subjects in the experiments had at least uncorrected or corrected 20/20 vision.

“We determined that blurriness and distance are equivalent from the visual system’s perspective,” said Loftus. “When you make an image smaller you lose information in exactly the same way as happens when you keep the picture large but make it blurry. That is why when witnesses say they viewed something from 120 feet, for example, I can take a picture and know precisely how much to blur it to match that distance.”

Loftus’ formula for how much detail is lost is based on 20/20 vision and normal daytime light. It can be adjusted for nighttime or when a person has extremely good or poor eyesight.

In the Fairbanks case, a witness standing several blocks away viewed the 1997 beating of one man by four suspects. The suspects were tied to a separate fatal beating of second man, but that crime had no witnesses. The suspects, however, were tied to the murder by circumstantial evidence. The witness, who saw the non-fatal beating from a distance of about 450 feet, later identified two of the suspects from police photographic lineups. He also identified the two in a subsequent trial.

Loftus also testified as an expert witness, noting, among other things, that identifying a person from 450 feet was the equivalent of sitting in the centerfield bleachers at New York’s Yankee Stadium and being able to recognize someone in the box seats behind home plate.

The jury convicted the four men at end of a 1999 trial. But the story didn’t end there. Loftus began his research, and in 2003 found out that a newspaper reporter turned journalism professor had his students interview people involved in case. Four of the jurors admitted conducting an experiment during a break in the trial, thinking they had permission to do so. They had one or more jurors pace off given distances and see if the others could recognize them. One of the jurors said although he had bad eyesight and couldn’t recognize faces at those distances, he believed the other jurors because they could. The jurors said this gave credibility to the eyewitness identification. In late 2004, an appeals court granted a new trial.

“It is becoming more apparent that there are serious problems with eyewitness testimony,” said Loftus. “Misidentifications can occur, and the quality of memory is limited by the distance at which a witness sees a person. This research, which specifies mathematically the relation between memory quality and distance, results in our being able to present intuitive information to a jury, which can help it come to the best possible decision in a case.”

Loftus said the distance-blurriness effects are not unique to faces and they also have been demonstrated in laboratory experiments using images of vehicles.

Outside of the courtroom, he sees a number of other practical applications for the research. These include the design of sensing devices for spotting terrorists and the reliability of people identifying potential sites for weapons of mass destruction from aerial photographs.