See No Evil
Eyewitness testimony may be unreliable, but the Supreme Court doesn’t want to be the one to say so.
How reliable is eyewitness identification?
Illustration by Natalie Matthews-Ramo.
A few weeks ago, testifying before the Senate Judiciary Committee, Justice Antonin Scalia reminded us that the Supreme Court doesn’t reach out to decide issues—it merely decides cases. The unreliability of eyewitness identifications is an issue. Perry v. New Hampshire is a case. And at oral argument this morning, it is immediately clear that this case is not only the wrong vehicle for solving the problem of mistaken eyewitness identifications, but that the Supreme Court believes itself the wrong institution to fix it. As Justice Elena Kagan puts it, new research “should lead us all to wonder about the reliability of eyewitness testimony.” Just don’t expect the high court to do much more than wonder.
Anyone who followed the Troy Davis case is aware of the enormity of the problem. A man went to the death chamber based largely on the eyewitness testimony of nine witnesses, although seven later recanted. Our entire criminal justice system is constructed around the proposition that our eyes don’t lie and our memories are infinitely looping YouTube videos. As Justice William Brennan wrote in a 1981 dissent: “There is almost nothing more convincing than a live human being who takes the stand, points a finger at the defendant, and says, ‘That’s the one!’ ”
The problem, of course, is that you can be very convincing and also wrong. In his book Convicting the Innocent (excerpted in Slate), Brandon Garrett studied 250 DNA-based innocent exonerations, and concluded that 190 of them (a whomping 76 percent) were based on false eyewitness identifications. False identifications, then, aren’t so much a problem as a plague.
The high court used to worry a good deal about this. In 1977—the last time it examined the reliability of eyewitness identifications—it issued some fairly definitive, if upside-down, proclamations about what makes for good eyewitness evidence. In Manson v. Brathwaite the court laid out specific criteria for determining the scientific reliability of an eyewitness ID—including how much opportunity the witness had to view the perpetrator and how certain she was of her identification.
Social science now suggests most of those conclusions were wrong. As Adam Liptak recently explained: “There is no area in which social science research has done more to illuminate a legal issue. More than 2,000 studies on the topic have been published in professional journals in the past 30 years.” This morning’s case was meant to allow the law to catch up to the science. That probably isn’t going to happen.
The case involves an alleged car break-in and a witness who offered the cops a less-than-satisfying identification—it was a “tall black man”—then voluntarily pointed to the suspect who was standing outside her apartment window with the police. (Later, at the police station, the witness was unable to identify the defendant from a photo lineup). But the police did nothing wrong or suggestive, which arguably makes the case different from all those 1970s precedents which sought to deter police misconduct. In Perry the question is whether, absent police manipulation, the defendant has a constitutional right not to have unreliable eyewitness evidence introduced at his trial.
New Hampshire public defender Richard Guerriero, representing Barion Perry, argues that this case turns on the inherent awfulness of eyewitness IDs and not on police misconduct. Nobody seems to be buying. Scalia isn’t just refusing to buy—he’s selling short: “Why is unreliable eyewitness identification any different from unreliable anything else?” he asks. If you’re going to constitutionalize the introduction of bad evidence, why draw the line at bad eyewitness IDs?
Guerriero replies that “eyewitness identification evidence is unique” and that the Supreme Court has identified it in the past as “probably the leading cause of miscarriages of justice.” Scalia disagrees. “If we accept your argument for eyewitness we should similarly accept it for everything else. There is nothing special about eyewitnesses.”
Justice Anthony Kennedy suggests that if the police did nothing wrong in this case, the proposed rule just makes their jobs harder. “I don’t know what you want the police to do,” he snaps. Both Chief Justice John Roberts and Kennedy take us on brief reminiscences of the good old days. “I remember in law school, one of the things in criminal law, the professor says, ‘All right, everybody be quiet,’ ” Roberts says. “And then a certain amount of time goes by and then he starts asking people, ‘Well, how much time went by?’ And people—some people say four minutes, some people say one minute. And it turns out, if I'm remembering correctly, to be a lot shorter than most people think.” The point being that people are as lousy at estimating time as they are at identifying criminals, but we don’t constitutionalize bad time evidence.
The other point is that justices like telling stories, and now it’s Kennedy’s turn. He tells of “a case I had where a prosecution witness was very, very certain, all too certain, and I said, ‘Do you ever take your wife out to dinner?’ And he said, ‘Oh, yes.’ I said, ‘Has it ever happened to you that midway in the meal you say, “Is that our waiter?” ’ And the waiter has brought you the menu, he has taken your order, he has brought your food, and you were under no stress at the time. … And there was good light?’ ” Kennedy then explains: “So you teach the jury this way. And you're just—you're just usurping the province of the jury, it seems.”
Justice Stephen Breyer jumps in to add that the federal rules of evidence already preclude judges from allowing in evidence that is more prejudicial than it is probative. “What is the difference between what you're asking for and what already exists in the law?” Kagan worries that other classes of testimony are as unreliable as eyewitness evidence. “Let's say that it turned out study after study after study [showed] that jailhouse informants lie,” she says. “And so the testimony of jailhouse informants is likely to be just completely unreliable, double as much as eyewitness testimony. Same rule for that?”
Justice Ruth Bader Ginsburg asks why all the myriad safeguards against admitting bad evidence—jury instructions, evidentiary rules, and cross examination—are not enough to keep it from being used at a trail. When Guerriero sits down, he looks a bit like he’s just been the victim of a mugging. And he can’t quite identify his assailant.
New Hampshire’s attorney general, Michael A. Delaney, spends his time explaining that unless the police manipulated the eyewitness testimony, there can be no due process violation. Justice Sonia Sotomayor says she’s not inclined to create a test that looks at the police officer’s intentions. Explains Delaney: “The standard is not reliability. The standard for due process is the use of orchestrated police suggestion.” In his view, without the latter, there can be no constitutional wrong.
Then Nicole A. Saharsky has 10 minutes to argue for the Justice Department, which sides with New Hampshire. Her delivery is a strangely effective mix of high-speed assertions, but her point is that without police wrongdoing, there is no claim: “The State can't create a false document and introduce it at trial,” she explains, and “it can't manipulate someone's memory and then use that evidence to prove guilt at trial.”
Kagan asks about a hypothetical case in which an identification “has been produced by torture, but the torture has been through a nonstate actor.” Does the introduction of such evidence violate the Constitution? Saharsky replies that prosecutors wouldn’t ever introduce such evidence and systemic checks could keep it out, but suggests that the Constitution wouldn’t prevent its introduction. She explains that “there are numerous trial protections outside of the constitutional limits” that bar bad eyewitness identification testimony, including special jury instructions. But, as she explains: “The Constitution has enshrined the jury as the fundamental guarantee—the fundamental protector of liberty,” and taking reliability questions away from the jury would be improper. She concludes that a constitutional rule about the admissibility of unreliable evidence would mean that “defendants throughout the United States (will be) making arguments about all different kinds of evidence not involving the police being unreliable,” opening the floodgates to claims that all evidence is as tainted as eyewitness testimony.
In his rebuttal, Guerriero tries to explain again that the reason you want to take fallible eyewitness identifications away from the jury is precisely because eyewitness testimony is both powerful and wrong: “The witness's sincerity has a powerful effect on the jury,” he explains. But it’s clear that this court will either dismiss or slide right past the old precedents that suggest that eyewitness evidence is uniquely dangerous. Oddly enough, the fact that other compelling evidence may prove equally untrustworthy seems to have immunized all the bad eyewitness evidence.
Meanwhile, police forces are already dealing with the issue raised by the Perry case, as are some state courts. Just recently, a special master appointed by the New Jersey Supreme Court to examine eyewitness evidence concluded that such memory should be treated “as a form of trace evidence: a fragment collected at the scene of a crime, like a fingerprint or blood smear, whose integrity and reliability need to be monitored and assessed from the point of its recovery to its ultimate presentation at trial.”
So maybe one day the worst procedures that produce bad eyewitness IDs will finally be eradicated. If that happens, it will be because of the efforts of virtually every institution in the U.S. criminal justice system—except the Supreme Court, which will still be wondering.
Dahlia Lithwick writes about the courts and the law for Slate.