One intruder had on a blue backpack. The other was wearing tan slacks. Or was it a green backpack and brown slacks?
It happened so quickly--less than a minute--that 42 witnesses could not agree on such vital details when asked about the robbery two days later through conventional police interview techniques. The witnesses had been assembled by UCLA psychologists ostensibly to view a slide show. But during the viewing, two participants in the experiment barged into the room, turned on the lights and made off with the projector.
Then the interviewers changed tactics, with remarkable results. They didn't just ask, "What happened?" Rather, they told the witnesses to first recall the physical setting and their own frame of mind at the time of the robbery. They told the witnesses to omit no detail, no matter how trivial. They told them to reconstruct the crime in a variety of orders, not just from start to finish. Finally, they told the witnesses to re-create the incident not only from their own perspective but also from that of the robbers.
This "cognitive interview" technique, pioneered by UCLA psychologist R. Edward Geiselman, elicited up to 35% more details about the crime--all of them accurate and many of them crucial, according to the researchers.
With little notice, psychologists have been developing new and relatively simple ways to jog the human memory, raising the potential of making eyewitness testimony far more reliable than previously thought possible.
These advances may also prove useful well beyond the world of criminal justice, they say. For instance, such memory-enhancing tools can aid public health officials in tracing partners of people with venereal diseases, said Ronald Fisher, a psychologist at Florida International University. The techniques can even help a person locate a missing wallet or keys, he said.
"We've known for a long time how to increase learning, but until now little has been known about how to enhance recall," Fisher said.
The cognitive interview, in particular, has been adopted by many law enforcement agencies, including the FBI, the U.S. Secret Service and the Metro-Dade Police Department in Florida.
"I really believe in this technique," said John S. Farrell, chief of detectives at the Metro-Dade Police Department. "This is a whole different ballgame."
Researchers in Florida and Minnesota have also found that false identifications at police lineups can be reduced by as much as half when witnesses view suspects one at a time instead of six simultaneously, which is the prevalent practice.
Scientists elsewhere have shown that memory can be significantly affected--for better and worse--simply by the way a witness is asked about crime details.
These emerging insights are timely because many scientists now believe that hypnosis is far less reliable than most people assume. And courts across the country, including in California, are placing severe restrictions on its use in criminal cases.
New ways to increase eyewitness reliability are especially important because juries tend to give great weight to eyewitness testimony--sometimes without realizing how unreliable such witnesses can be.
"The confidence of the eyewitness is the most powerful predictor of verdicts," said Brian L. Cutler, a Florida International University psychologist. Yet research shows that confidence is "only weakly correlated with identification accuracy," he said.
'Cause of Wrongful Conviction'
"Faulty eyewitness testimony, in my opinion, is the major cause of wrongful conviction in this country," added Elizabeth F. Loftus, a University of Washington psychologist. She estimated that as many as half of such wrongful convictions may be due to faulty eyewitness testimony.
Loftus has found in experiments that 18% of mock jurors voted to convict in a case involving only circumstantial evidence; but after a positive eyewitness identification is introduced, the conviction rate shot up to 72%. Interestingly, a third jury voted to convict by nearly the same proportion--68%--even after it was told that the eyewitness had "very poor vision" and was not wearing glasses at the time of the crime.
Any memory-improving techniques would be a welcome addition in the fight against crime, especially in places such as California, where many criminal cases drag on for years without adjudication.
With every court delay, prosecutors and defense lawyers alike are faced with the problem of fading memories on the part of key witnesses, making cases increasingly difficult to prosecute or defend.
For instance, in the current Night Stalker murder trial in Los Angeles Superior Court, numerous witnesses have been unable to recall potentially important details. One woman, who survived a shooting from point-blank range in a darkened garage nearly four years earlier, was unable on the witness stand to positively identify her assailant, as she had done during pretrial proceedings years earlier.
In another much-delayed Los Angeles case involving the 6 1/2-year prosecution of Virgil Byers for two fatal drive-by gang shootings, two key eyewitnesses now have such conflicting accounts of the killings that they seem to be describing entirely different events, Deputy Dist. Atty. Loren Naiman said.
How the human memory process operates is still much debated. Some believe that it operates much like a tape recording or video recorder that can be stopped or played back, perhaps aided by tools such as hypnosis. Others say memory is an incomplete abstraction that can be highly susceptible to distortion.
"Models of the mind in general, and theories of memory specifically, seem to follow the latest advances in gadgetry," such as the telephone switchboard or a computer, according to Loftus.
She said recollections not only fade with time but also become increasingly susceptible to alteration by subsequent information or subtle cues, including questions posed by investigators.
Her own research has shown that witnesses to a complex event, such as a crime or accident, often make serious mistakes, including misidentifying the color of a car, remembering a yield sign where a stop sign had been and even placing non-existent items at the scenes, something as conspicuous as a barn.
Such distortions, she added, can cause "ripple" effects that lead to wholesale contamination of the original memory. And once such alterations occur, it is almost impossible for a witness to retrieve the original memory, Loftus said.
She and other researchers, including Steve Penrod, now at the University of Minnesota, have found that the accuracy of identifications is also significantly affected by the value of a stolen object.
Similarly, research subjects who learn that an offender may be treated severely are more likely to make positive identifications than those who learn that the offender may be treated mildly.
Other scientists have confirmed in recent experiments that cross-racial identifications generally are more difficult than same-race identifications, a little-understood phenomenon, according to Cutler.
Another major factor that influences recall is the stress associated with witnessing a violent event, which alone can lead to poor retention of details. Loftus and others demonstrated this phenomenon in a series of experiments involving the viewing of two short films depicting a shooting, one graphically violent, the other far less so.
In the violent version, which portrayed a boy being shot in the face, only 5% of the viewers could recall the number on the boy's jersey. But almost 28% of those who saw the less-graphic version could recall it.
All these findings come at a time when hypnosis is generally falling into disrepute as a courtroom tool.
"Hypnosis does not lead to accurate recall," said Dr. Martin T. Orne of the University of Pennsylvania. "Indeed, the memories obtained prior to hypnosis are considerably more reliable than those obtained during hypnosis."
Hypnosis can induce "pseudo-memories" that can become "extremely dangerous in the courtroom" because they can make "honest liars" out of well-meaning witnesses, he said.
But Orne agreed with hypnosis proponents that hypnosis can still be a valid investigative tool under certain, limited circumstances, such as when it might lead to evidence that can be independently corroborated.
One promising alternative to hypnosis involves research on how police lineups can be more effective. Cutler and Penrod, for instance, have found that sequential lineups, in which one suspect is viewed at a time, can reduce false identifications by 50%.
In their study, published in the Journal of Applied Psychology last May, they had hundreds of college students view videotapes of staged robberies. Afterward, when the lineup included six suspects appearing simultaneously, 39% identified an innocent person. But the error rate dropped by half when the same suspects were viewed one at a time.
As a way to enhance accurate identifications, Cutler and Penrod suggest that suspects in a lineup be asked to move about and to speak.
These methods may become even more effective if enhanced by the cognitive interviews, especially in cases in which suspects wore disguises, the scientists found.
For now, the most promising memory-enhancing tool appears to be the cognitive interview, a technique developed by UCLA's Geiselman. It already has been incorporated into the training programs of the U.S. Secret Service and the FBI.
In addition, Geiselman has given training sessions on the technique in Chicago and throughout California, including to the Los Angeles County Sheriff's Department.
Other hallmarks of the cognitive interview include asking whether a suspect reminds the witness of anyone he or she knows and, if so, why. If a name had been mentioned during a crime, a witness is asked to think of the first letter by going through the alphabet, and then repeat the process with syllables.
If numbers come into play, a witness is asked whether the number had been high or low, and how many digits. Witnesses are also asked about a suspect's speech characteristics.
Geiselman and his colleagues have demonstrated the efficacy of the technique in numerous experiments, many of them using students at UCLA as well as people from all walks of life.
Often, the technique is more effective than taking a witness back to a crime scene because such scenes rarely remain unchanged, Geiselman said.
More recently, the interview technique, with further refinements, was field-tested by Miami detectives investigating more than 70 actual armed robberies in 1988. It produced 50% more details than conventional techniques, researchers reported.
Some of Geiselman's latest research also suggests that the technique can help witnesses recall license plate numbers, when used in conjunction with a license-plate simulation device.
In one study, Geiselman found that witnesses on average were able to recall 3.85 (out of seven) characters on license plates. Without these tools, study subjects could recall only an average of two characters.
Such improvement would dramatically reduce the number of possibilities from 676,000 license plates to 260, he said.
In other studies related to the cognitive interview, psychologist Martin Safer at Catholic University of America in Washington has shown that eyewitnesses are able to recall significantly more details when questioned several times within the first 48 hours of a crime rather than in just one sitting. He said the findings challenge the assumption that witnesses are able to recall all the first time around.
The best thing about the cognitive interview, Geiselman said, is that it is "a viable memory-enhancement technique that is effective, efficient and legally acceptable." And it is easy to teach; his workshops typically take four to six hours.
Cutler and Penrod believe that the emerging memory-enhancing experiments provide sound justification for courts to allow experts to testify more frequently on the unreliability of eyewitness identifications.
"In order to minimize these errors, courts and lawyers must familiarize themselves with the developing knowledge base provided by the social sciences," Loftus said.