If you were watching a video of people passing basketballs back and forth, and a person in a gorilla suit strode into the middle of the game, stopped, pounded his chest, then walked on, you'd remember that, right?
If you were busy counting basketball tosses, about half of us would miss the gorilla entirely and would swear he never appeared, according to the famous "invisible gorilla" experiment done by social scientists.
This is amusing, but faulty visual memories have a more insidious side when an eyewitness to a crime uses an inaccurate recollection to identify a suspect. Too often, the result is an innocent person is convicted.
The conundrum for the criminal justice system is that eyewitness testimony is notoriously faulty even though juries tend to take it as gospel. We all think our memories are better than they probably are, and we instinctively trust a crime victim who can point to a mug shot and say with confidence, "He did it." But thanks to the revolution in DNA evidence we know that even the most ardently confident eyewitness can be flat-out wrong.
Jennifer Thompson-Cannino is a prime example. When the then-22-year-old college student was raped in 1984, she made a point of studying her attacker's face. In two lineups she picked Ronald Cotton as her assailant, who then spent nearly 11 years in prison before DNA tests helped prove that he was innocent and another man — a serial rapist whom Thompson-Cannino rejected as the perpetrator — was guilty.
Thousands of studies have been published on the fallibility of eyewitness identification. They suggest that about a third of the tens of thousands of annual eyewitness IDs are erroneous. Mistaken identifications have been involved in 75 percent of the more than 270 convictions overturned based on DNA evidence.
What we know about visual memory is that it is not dependably photographic and can be easily contaminated by subtle cues and suggestion.
When a police detective administers a lineup, the officer may unconsciously communicate signals to the witness about the "right" choice. When a witness is shown photos simultaneously, she may "comparison shop" and choose the one that most closely resembles the perpetrator rather than an exact match.
The controversial case of Troy Davis, who was executed Wednesday for the 1989 killing of a police officer, was riddled with these types of issues. They included Savannah, Ga., police showing some eyewitnesses Davis' photo before the lineup. Such problematic eyewitness IDs may have led to a massive injustice.
But change is coming. In a huge step forward, the New Jersey Supreme Court in August directed judges to more thoroughly scrutinize the way police conduct eyewitness identifications and instruct the jury on potential flaws. And there's an eyewitness ID case before the U.S. Supreme Court in November — the first one since 1977 — although it's on a narrow issue of whether judges should look more closely at all questionably obtained eyewitness testimony or just when police did something wrong.
We do know what effectively reduces inaccurate IDs. Doing lineups that are "double-blind," where the administrator doesn't know who the suspect is, makes a substantial difference, as does notifying the witness that the perpetrator might not be there at all. Florida's Innocence Commission has recommended that all state and local law enforcement agencies adopt these practices.
A new study out of the American Judicature Society also should put to rest any doubt that lineups done sequentially, where the witness is shown one photo at a time, are more accurate than those where the witness sees a group of photos at once. The field-based research found that the sequential approach demonstrably reduced the error rate.
No one wins when the wrong person is convicted and a dangerous perpetrator escapes justice. Police could avoid this by changing their practices, and courts could do more to block iffy IDs and urge juries to be more skeptical. As the invisible gorilla crossing through the basketball game demonstrates, sometimes we don't really know the truth when we see it.