Are some wrongful convictions inevitable? Experts don't see police errors in Tillman case

By Alex Wood , Journal Inquirer

The most striking aspect of Journal Inquirer interviews with four eyewitness-identification experts about the case of James Calvin Tillman, the East Hartford resident who served more than 18 years of a 45-year prison sentence because of a mistaken eyewitness identification, is what they failed to produce:

* None of the experts identified a clear error in the procedures Hartford police used in investigating the case.

* Although several experts suggested that the jury in Tillman's 1989 trial would have benefited by hearing from an expert in eyewitness identification, none identified a compelling point an expert could have made to change the jurors' minds.

The victim of the crimes Tillman was wrongfully convicted of committing, a 26-year-old woman, told police that a man she didn't know forced his way into her car as she prepared to leave a downtown Hartford parking lot in the early morning hours of Jan. 22, 1988, a Friday.

She said the man punched her in the face at least twice during the crime and drove her to a more remote parking lot several blocks away, where he robbed her of money and jewelry and then raped her. Finally, she said, he drove her several more blocks before parking the car and running away with her pocketbook.

The woman reported the crimes immediately and was taken to Hartford Hospital for treatment.

Identified from 'mug book'

She went to the Hartford police station the following Monday, where she was interviewed in detail about the incident. Detective Stephen Kumnick showed her eight color Polaroid photos of black men, none of whom she identified as her attacker, according to summaries of the trial evidence by the prosecutor and defense lawyer who argued Tillman's appeal before the state Supreme Court. The court upheld Tillman's conviction in 1991.

Kumnick then asked the woman to look through a book of Hartford police color photographs of black men who had been arrested in the area of the city where she had been accosted. The initial attack occurred in a parking lot near the Arch Street Tavern, where she had been socializing with co-workers after completing a noon-to-10 p.m. shift at the Travelers Insurance Co.

Kumnick left the woman to look through the book with Janette Getz, a social worker with the Hartford Police Crisis Unit, according to the prosecutor's summary of the evidence. The victim looked at 11 photos in the book, then picked the 10th, a nearly 5-year-old photo of Tillman, according to Kumnick's affidavit seeking a warrant for Tillman's arrest.

The detective said the victim was positive of the identification but commented that the man looked younger in the picture. Kumnick said in the affidavit that he then obtained a "black-and-white contact photo print" of Tillman that was less than 8 months old, and the victim reported that it showed the current appearance of her attacker.

One of the major concerns expressed by eyewitness-identification experts about traditional police lineups or photo arrays is that investigators may unconsciously steer the witness toward selecting the suspect.

Gary L. Wells, a psychology professor at Iowa State University, has written that if the witness says the number of the photo of someone who is not a suspect, the investigator might reply, "Take your time. Make sure that you look at all the photos." Wells added that if the witness wavers between photos of the suspect and a nonsuspect, the investigator might focus the witness' attention on the suspect by asking, "What about No. 4 makes you remember him?"

But the known facts suggest that no such steering occurred in the Tillman case. Police evidently had no suspect, and Kumnick, the lead investigator, was out of the room when the victim made her initial selection.

Safeguards lacking

It is true that the "mug book" approach to identification used in the Tillman case lacks some of the safeguards built into a traditional photo lineup:

* Everybody in the mug book is a potential suspect, so there is no "wrong" answer for the witness to give, unless the person selected can be excluded by biological evidence or has an unassailable alibi, such as being in jail.

* The photos in a photo lineup are selected because they look similar to the suspect. But many of the photos in a mug book may look nothing like the perpetrator, leaving the witness with many fewer genuine choices. Evidently, the only things the people shown in the mug book in the Tillman case had in common were that they were black men who had been arrested in a given part of Hartford.

But to criticize mug-book identifications raises the difficult question of how police could solve crimes committed against strangers if they were deprived of this technique.

Steven D. Penrod, a psychology professor at the City University of New York's John Jay College of Criminal Justice, says Tillman's case provides "a lesson about the unreliability of eyewitness identification."

But he adds, "There is not a lesson about police practices in this case."

"I would not fault the police for doing what they did," agrees Robert W. Shomer, a psychologist in Encino, Calif., who has testified hundreds of times as an expert on eyewitness identification over 30 years.

The chief moral of the story thus may be a particularly sobering one: Some erroneous criminal convictions may be unavoidable.

Big element of luck

Tillman was freed in large part due to the work of two public defenders, Karen A. Goodrow of Vernon Superior Court and Brian S. Carlow of Middletown Superior Court, who together head the Innocence Project of the state public defender's office.

They tracked down the rape victim's clothing, stained with bodily fluids, in a box being held in storage at the Hartford Legal Aid Society, whose lawyers had obtained a DNA test on Tillman's behalf during a "habeas corpus" lawsuit in 1991.

But Tillman's exoneration is also attributable in large part to sheer luck: He was convicted of a rape in which the perpetrator apparently left bodily fluid at the scene, and the advanced DNA testing now available revealed that the fluid wasn't Tillman's. The 1991 DNA test had failed to produce results because the amount of DNA for testing was insufficient with the technology then available.

There are not means to correct all erroneous convictions. People frequently are convicted of crimes such as armed robbery on the basis of eyewitness identification by strangers when no DNA evidence is available to check the accuracy of the identification, says Otto H. MacLin, a psychology professor at the University of Northern Iowa.

Of the first 130 DNA exonerations in the United States, 101 involved mistaken identifications, most of which were eyewitness identifications, according to the Innocence Project at Yeshiva University's Benjamin N. Cardozo School of Law in New York.

Moreover, studies show that 20 to 25 percent of witnesses pick a nonsuspect out of photo arrays, according to Penrod. If these witnesses could be characterized as guessing, Penrod adds, other witnesses who are guessing must be picking the suspect. Based on data from the studies, he estimates that unreliable suspect selections may run as high as 5 percent.

But two more-recent field studies have shown lower selection rates for nonsuspects.

Race difference a problem

Tillman's conviction is consistent with one phenomenon that has support in research: The rape victim is white, Tillman is black, and studies have shown that people are less accurate in identifying strangers of other races than they are in making such identifications among members of their own race.

In a 2001 article analyzing the results of 39 studies, Florida State University psychologists Christian A. Meissner and John C. Brigham concluded that people are 1.56 times as likely to misidentify a person of another race as they are to make such a mistake with someone of their own race.

Similarly, the psychologists concluded, people are 1.4 times as likely to make an accurate identification of a member of their own race as they are to identify correctly a member of another race.

Bigots aren't the only ones who make such mistakes.

Meissner and Brigham found no evidence that racial attitudes were a factor in the accuracy of identification. Interracial contact is a factor but only a small one, accounting for some 2 percent of the variation in accuracy, they added.

"It's not a stereotyping thing. It's a brain thing," MacLin says of the problems with cross-racial identification. "Everybody's brain has problems when looking at people of another race."

One theme sounded repeatedly by the experts interviewed by the Journal Inquirer was that expert witnesses should be permitted to tell juries about the results of these and other studies showing problems with eyewitness identification.

Experts rarely allowed

In Connecticut, as in a number of other states, judges traditionally have been reluctant to allow such testimony.
Expert testimony on eyewitness identification is "just beginning to be allowed," says public defender Christopher Cosgrove, who represented Tillman in his 1989 trial in Hartford Superior Court and now heads the public defender's office in Litchfield. "It was routinely ... not allowed back then."

One common misconception among jurors is that the stress of being victimized in a violent crime "stamps in" the memory of the perpetrator's face, Shomer says. In fact, he says, stress interferes with the psychological processes that produce reliable identification.

But even when an expert provides such information to a jury, the defense inevitably faces an uphill battle in an eyewitness-identification case.

"They see an honest, well-meaning victim pointing at a guy and saying, 'I'll never forget that face,'" Shomer says.

Experts also can expose juries to a more sophisticated understanding of the processes involved in memory than they might have when they arrive in the courtroom.

"Our memory isn't a photo album," MacLin says. "Memory is a mosaic, a blur."

One of the major issues that eyewitness-identification experts focus on is how the memory of a face can be affected by subsequent events.

For example, studies have shown that witnesses become more confident in their identifications when told that they have identified the police suspect and less confident if told that they haven't identified the suspect. Juries, in turn, give considerable weight to witness confidence in deciding whether to rely on the identification.

But Solomon M. Fulero, a psychology professor at Sinclair College in Dayton, Ohio, says studies have shown that the confidence of the witness is "poorly related to accuracy."

Emotion isn't proof

When the victim first saw Tillman's photo in the mug book, she began to shake and cry, according to the prosecutor's summary of trial testimony.

"That's gonna be used as evidence - that she shook," MacLin says. "That will be used as a validation of her memory."
But he questions such reasoning, in part on the basis of a personal experience. He says a man once attacked his girlfriend, who took off her stiletto heel and smashed him in the face with it.

There was a similar-looking man living in the area, MacLin recalls, and when he would see that man, the hair on the back of his neck would stand up. But he could tell the man wasn't the attacker, who had a scar from where he had been hit in the face.

"The hair on the back of my neck stood up for the wrong guy," MacLin says.

That, he says, is an example of a phenomenon psychologists call "stimulus generalization." Another example of the phenomenon, he says, is that people will stop for red lights that look slightly different from the ones they are familiar with.

MacLin stresses the importance of richly detailed information in analyzing an eyewitness identification. For example, available public records don't indicate the total number of photographs in the mug book from which the rape victim picked Tillman's photo. If there were only 12, MacLin suggests, she may have felt pressured to make a selection when she reached the 11th.

"Everything makes a difference," MacLin says.

Search an option?

When Hartford Superior Court Judge Thomas P. Miano dismissed the charges against Tillman, he suggested that one lesson of the case is that prosecutors shouldn't rely solely on eyewitness identifications by strangers.

Prosecutor Edward Narus, who tried the case, says he had more than that. Tillman gave police an alibi that didn't check out when police spoke to the alibi witnesses, Narus says, though one of those witnesses later supported Tillman's alibi in trial testimony.

In addition, Kumnick testified that Tillman had cuts and bruises on his knuckles when he was arrested, according to the prosecution appeal brief. When asked how he had gotten the cuts, Tillman first said he wasn't sure, then said he had gotten them working at a car wash, the brief adds. The hand injuries would have been considered consistent with his having punched the victim.

There also was physical evidence that failed to rule Tillman out as the attacker. One of the bodily fluid stains found on the victim's clothing was from a "non-secretor," someone who doesn't secrete the antigens used for blood typing into other bodily fluids in detectable quantities.

Tillman is a non-secretor.

But that evidence wasn't strong. It made Tillman one of 20 percent of the population who could have committed the crime.

There may have been at least one additional avenue police could have pursued to corroborate the victim's identification of Tillman: They could have sought to search for the more than $2,500 worth of distinctive jewelry stolen from the victim.

But failure to find the jewelry would have been unlikely to bring about Tillman's acquittal at his trial in 1989.

The perpetrator could have sold the jewelry or discarded it in the three days between the crime and the victim's identification of Tillman as her assailant. So failure to find the stolen jewelry during a search wouldn't have been evidence of Tillman's innocence, Fulero says. But discovery of the jewelry in Tillman's living quarters "sure would be incriminating," the professor adds.

Eyewitness ID
Truth in Justice