Eyes on an Execution

The Troy Davis case shows how wrong eyewitness evidence can be.

Rev. Raphael Warnock speaks at a press conference in support of Troy Davis

The Troy Davis case was staged—pure theater. I do not mean “staged” because the case has attracted worldwide attention and high-profile supporters. Nor do I refer here to the drama surrounding the Georgia Board of Pardons, which at the 11th hour denied clemency again this morning, so that Davis faces execution tomorrow—despite powerful evidence of his innocence. By “staged” I mean that the eyewitness evidence at the core of his original criminal trial was, quite literally, staged by the police.

The federal court that finally reviewed evidence of Davis’ innocence agreed “this case centers on eyewitness testimony.” Yet that court put to one side the fact that seven of the nine witnesses at the trial have now recanted, and new witnesses have implicated another man. The court did so while failing to carefully examine how eyewitnesses ultimately came to identify Davis as the man who shot a police officer intervening in a fight at a Burger King parking lot. The Troy Davis case—which raises a wide array of flaws in our death penalty system, our post-conviction system, and the politics of criminal justice—is thus also a case about malleability of eyewitness memory and police misconduct.

We all know how hard it is to remember the faces of strangers we might encounter on the street. (Does she look familiar?) We might think that in a high-stakes criminal case, police would take precautions to carefully test the memory of an eyewitness who saw a stranger commit a crime. Indeed, police know they must do that; they conduct lineups as one kind of test. But we have also long known that serious mistakes can happen if the police suggest to the eyewitnesses whom to pick out. This can happen unintentionally, if the officer running the lineup knows who the suspect is and gives inadvertent cues. It is also police misconduct if the officers openly suggest to the eyewitness whom to pick out. It is considered suggestive if the police go around showing eyewitnesses single photos of just the suspect they have in mind. Or if they merely make suggestive comments.

Yet somehow the police did all of the above and more in Troy Davis’ case—a perfect storm of botched eyewitness-identification procedures. Police did show photo arrays to most of the eyewitnesses—eventually. Although police made up a five-photo array with Davis’ picture in it, they waited five to 10 days before using it to test the memories of any eyewitnesses. Why wait? Eyewitness memory decays rapidly. But in the meantime, police plastered wanted photos with Troy Davis’ image—the same photo they put in the photo array—all around the neighborhood, and it ran widely on all of the local media outlets. Witnesses did not miss those wanted postings. Witnesses also described feeling pressure to identify Troy Davis. For example, one testified at trial about being told that “if I don’t cooperate with them, that I’m gonna be in prison for ten to twelve years.”

And then, finally, there was the staging—a full reenactment that the police conducted. They brought three key witnesses back to the Burger King parking lot where the shooting took place. Police had one of them—the man who Davis’ team maintains is the real perpetrator—play a different role—the role of an innocent bystander. Police tried to get the eyewitnesses to agree on who was standing where. Frustrated, they then located a fourth key eyewitness, and asked her “if she could place everyone in the positions that she saw them in.” This was an effort to get all of their inconsistent stories straight. As a result, their stories started to merge. Only later were the witnesses shown photo arrays (which were not conducted properly in a double-blind fashion where the officer does not know which is the suspect).

One of those witnesses recalled at trial that he told police when he saw the photo array, that he was only “like sixty percent sure that was the guy,” because “I was nervous, I was watching the gun.” That was after seeing the wanted posters for ten days at the Burger King where he worked. But at trial—where Troy Davis was sitting at the defense table—he pointed him out to the jury. He has since recanted. Another witness pointed out Troy Davis at trial, telling the jury “you don’t forget someone that stands over and shoots someone.” This claim must have surprised the prosecution as much as the defense, since the police didn’t even bother to show him a photo array because shortly after the incident he told police, “I wouldn’t recognize him again except for the clothes.” He said he did see Davis’ photo in the paper the day before he testified, however. Yet another witness, who was severely injured when hit in the altercation that led up to the shooting, identified another man initially as the assailant; later at the police station he concluded he no longer recalled who it was. He agreed at trial that police “kind of reassured” him that he had it wrong. One summary of all of the various pieces of conflicting testimony is in the Appendix to U.S. Supreme Court briefs filed.

I do not know if Troy Davis is innocent or not—no one will ever know. Police have so utterly contaminated the evidence, and so much has since come to light, that there is not much of the case against him left standing—one could not imagine a jury convicting him today, much less sentencing him to death. Reading through his trial materials, I was reminded time and time again of what I saw when I read trials of the first 250 people exonerated by DNA tests, when researching my book Convicting the Innocent. (There is no physical evidence to test in Davis’ case.) Of the 190 DNA exonerees who had been convicted based on eyewitness testimony, nearly all the cases had eyewitnesses who were positive at trial that the defendant was the person they saw. Yet 57 percent of them had not been certain when they first identified the defendant. They became certain only later.

One might want to think that a whole series of eyewitnesses would be much more likely to get it right than just one. Yet in the wrongful conviction cases I examined, 36 percent were misidentified by multiple eyewitnesses, some by as many as three or four or five. Perhaps the best known of those cases is that of Kirk Bloodsworth, the first person exonerated by DNA tests from death row. He was misidentified by five eyewitnesses, and as a result, Maryland now bars death sentences in cases with such weak evidence. We know that eyewitnesses can be influenced by one another—and in several of the innocent exoneration cases I studied, eyewitnesses all viewed photo arrays as a group.

While suggestive procedures were used in 78 percent of the trials that I studied, I did not see re-enactments nearly as suggestive as the one staged in Troy Davis’ case. (Although in one Virginia case, they placed a stocking over Willie Davidson’s head and pulled it on and off, asking “is this it, is that it,” since the victim was attacked by a man in the dark with his face covered by a stocking mask.)

Far more subtle police suggestion at lineups has caused eyewitnesses to miss the actual culprit and instead identify an innocent man. Most remarkable is another well-known case hailing from Georgia, where the victim identified John Jerome White but DNA showed years later that not only was White innocent, but the guilty man was coincidentally placed right next to him as a “filler” in the same lineup. Maybe the Georgia Board of Pardons does not know the case.

In the past few weeks there has been a drumbeat of news on the fallibility of eyewitness misidentifications. There is a massive body of social science evidence on the problem of eyewitness memory. Just yesterday, still more was added to the corpus—a report with results of a national eyewitness-identification field study was announced, with findings confirming the importance of adopting best practices for lineups. The U.S. Supreme Court is hearing a case this fall about eyewitness procedure and state action. The New Jersey Supreme Court just issued its remarkable Henderson ruling establishing a comprehensive social-science framework of rules for eyewitness testimony in the courtroom.

The eyewitness identifications in the Troy Davis case do not remotely pass muster based on what we know today about eyewitness memory—they were not double-blind, they were not conducted properly, and the staged re-enactment was blatantly suggestive. Nor is there any other indication that these witnesses were reliable. (They were not initially sure of what they saw under stress, in the dark, perhaps focusing on a weapon, and their initial descriptions were erratic.)

The Troy Davis case is a case about the death penalty and also about how much risk of error we as a society can tolerate. It is a case about judicial reluctance to meaningfully examine new evidence of innocence, including witness recantations. It is a case about jailhouse informants. But the Troy Davis case is finally a case about the fragility and malleability of eyewitness memory. And the eyes of the world are now on Georgia.