By LOU MARANO

Thursday, 5 July 2001 18:05 (ET)

WASHINGTON, July 5 (UPI) — Defense attorneys have long known that eyewitness identifications are notoriously unreliable in criminal proceedings. A new study sheds light on why.

In recent years, advances in DNA testing have exonerated many convicted “felons.” Disproportionately, these have been men sentenced to long prison terms after having been positively identified by rape victims.

In the Jan. 8 issue of the New Yorker magazine, Atul Gawande cited a study of 63 DNA exonerations of wrongfully convicted people. Of these, 53 involved witnesses who had made a mistaken identification. Almost invariably, those witnesses had viewed a lineup in which the actual perpetrator was not present.

In the July issue of the Journal of Experimental Psychology, professors Sharon L. Hannigan, of Bard College in Annandale-on-Hudson, N.Y., and Mark Tippens Reinitz, of the University of Puget Sound in Tacoma, Wash., have published a study indicating that illusions in memory might result from the basic human need to make sense out of events.

Through a series of experiments, they found that when people see an effect without seeing its cause, they automatically “fill in the blank” with a probable cause.

In everyday life, Hannigan and Reinitz wrote, these inferences are generally useful, adaptive — and correct. Even so, they are based on false recollection. And in criminal cases, the costs of false recollection are high.

Hannigan and Reinitz showed subjects a slide of an effect, such as oranges scattered on a supermarket floor. Then the investigators showed subjects pictures of the most probable cause of that effect, such as someone reaching for an orange from the bottom of a stack, and asked whether they had seen that picture before. A statistically significant number said they did.

The errors increased with the passage of time.

Hannigan and Reinitz found that these causal-inference errors were common in retrospect but not in looking to the future. In other words, exposure to the “effect” slides caused illusory memories of having seen “cause” slides but not the reverse.

In fast-paced and emotionally charged scenes, eyewitness identification errors are common even in the absence of inferential mistakes.

Gawande began his New Yorker article by recounting a staged altercation in a law school class at the University of Berlin in 1901. The professor asked students, as eyewitness, to describe exactly what they had seen. The most accurate witness got 26 percent of the significant details wrong. Some students present got up to 80 percent wrong.

“Words were put in people’s mouths,” Gawande wrote. “Actions were described that had never taken place. Events that had taken place disappeared from memory.”

Since 1901, the experiment has been replicated “thousands of times” with similar results.

This is important, Gawande wrote, because more than 75,000 people become criminal suspects on the basis of eyewitness identification in the United States alone, and the DNA exonerations show that the most common cause of false accusation is eyewitness error.

Gawande, a surgeon, wrote that in medicine this kind of systematic misdiagnosis would get intense scientific scrutiny. The law, however, “has balked at submitting its methods to scientific inquiry.” Ohio State University psychologist Gary Wells has tried to come up with practical solutions.

As reported by Gawande, Wells’ research shows that witnesses who picked the wrong person out of a police lineup were just as confident about their choices as those who identified the right person. Further, volunteer jurors believed inaccurate witnesses just as often as they did accurate ones.

Wells found that having multiple witnesses didn’t solve the problem. A crime might be witnessed by dozens of people, yet they would often finger the same wrong suspect.

In a “crime” staged to give 100 witnesses several good looks at a “thief,” 54 picked the perpetrator correctly, 21 said they did not think his picture was among the six they were shown, and the others spread their selection among the five others in the lineup.

A second group was given the same pictures minus the perpetrator. This time 32 people picked no one, but most of the rest chose the same wrong person.

Gawande wrote that researchers at the State University of New York at Plattsburgh discovered that witnesses who are not explicitly warned that a lineup may not include the perpetrator are substantially more likely to make a false identification under the misapprehension that they’ve got to pick someone.

But Wells and Rod Lindsay of Queens University in Kingston, Ontario, found that if witnesses were shown only one person at a time and made to decide whether he was the culprit or not before moving on, this reduced false identification by more than 50 percent without sacrificing correct identifications.

Of course, one would expect prosecutors and police departments to rush to adopt this method. One would be wrong.

In the 15 years since Wells and Lindsay published their findings, only a “scattered handful” of police departments, mainly in Canada, have implemented the reform. And prosecutors have been far more resistant than police.

Gawande quotes Wells as saying that people in the criminal justice system only want to know whose side you’re on — the prosecutor’s or the defendant’s. For them, science is just another form of spin.