University Study: Eyewitness Testimony Not Reliable
Eyewitness accounts often inaccurate, study finds
Abram Katz , Register Science Editor 06/21/2004
Victims who get a good long look at violent criminals are unlikely to identify them accurately later, Yale and U.S. Navy researchers have found.
This caveat follows from a unique study of 509 Navy and Marine officers undergoing elite survival training at Fort Bragg, N.C.
Results suggest that police and juries may give eyewitness testimony too much credibility, said Dr. Charles A. Morgan III, a Yale psychiatrist and lead author of the study.
"Memory in healthy people is not inherently terribly accurate. There’s a substantial amount of error," Morgan said. "Maybe we should demand more evidence."
Authors wrote, "The present data have a number of implications for law enforcement personnel, mental health professionals, physicians, attorneys and judges."
Mario T. Gaboury, director of the CrimeVictimStudyCenter at the University of New Haven, said, "Eyewitness testimony is often inaccurate. I don’t think anyone understood the magnitude of the problem until the past few years."
Previous research has called the reliability of eyewitness accounts into question.
The current study, which was published in the International Journal of Law and Psychiatry, is unusual in that participants were educationally, physically and mentally similar and all underwent nearly identical stressful events, Morgan said.
Groups of top officers undergoing realistic training at FortBragg are placed in a mock prisoner of war camp and subjected to low- and high-stress interrogations by U.S. officers acting as the "enemy."
The 40-minute high-stress session includes the threat of physical violence and creates stress levels equal to landing on an aircraft carrier at night for the first time and actual combat.
Details of the training are classified, but the study implies that participants are also "man-handled."
Twenty-four hours after the grueling sessions, the officers were asked to identify "interrogators" and "guards." They viewed a lineup, a group of photos and a sequence of photos.
Morgan and colleagues found that in the live lineup 30 percent of the high-stress group made correct identifications versus 62 percent of the low-stress group.
Using sequential photos the high-stress accuracy rate was 49 percent, while the low-stress rate rose to 76 percent.
The photo-spread method, which is used by most police departments, yielded even more lopsided results.
About 32 percent of the identifications in the high-stress group were correct, while 68 percent were wrong.
Around 88 percent of the low-stress group picks were correct, with a 12 percent error rate.
This means that almost seven out of 10 high-stress officers made mistaken identifications.
Furthermore, there was no relationship between the confidence level and accuracy of the memory, Morgan said.
Officers who were absolutely positive that they had selected the right person were no more likely to be correct than officers who expressed some doubt.
"Unfortunately, that’s what people on juries listen to," Morgan said.
Morgan said high levels of stress hormones such as cortisol and adrenaline may degrade spatial memory.
Norepinephrine, also produced under stress, apparently interferes with the brain’s prefrontal cortex, where memories are integrated, Morgan said.
Morgan said he hopes to measure hormone levels in trainees under various degrees of stress.
John H. Mace, professor of psychology at the University of New Haven, said many studies have cast doubt on the accuracy of eyewitness memories.
Mace said the Yale study is important because it apparently corroborates many previous hypotheses and results.
It may be a long time before defense lawyers start to challenge eyewitness testimony on the basis of the Yale and other memory studies, Gaboury said.
Court rulings typically lag behind scientific consensus, he said.
"We must be cautious. We don’t want the pendulum to swing too far," Gaboury said.