The Frontal Cortex

Factoid of the day:

Bad eyewitness identifications contributed to 75 percent of wrongful convictions in cases that were overturned by DNA evidence.

Given these dismal statistics, some states have tried to fix their procedures for eyewitnesses. New Jersey, for example, used to do police lineups the standard way: witnesses identified suspects from an in-person lineup as detectives stood next to them. The police officers would sometimes offer words of encouragement. New Jersey has now done away with the lineup, and instead presents people one after the other. This is supposed to prevent witnesses from making "relative judgments about which individual most looks like the perpetrator."

From the perspective of neuroscience, eyewitness testimony is an extremely unreliable type of evidence. The first reason is that our vision is vulnerable to all sorts of top-down influences, which alter (corrupt, some might say) the actual inputs witnessed by our eye. To paraphrase Paul Simon, "A man sees what he wants to see and disregards the rest." This doesn't mean that our senses aren't rooted in reality, but it does suggest that we shouldn't place too much trust in the details of perception, especially when the event in question happened fast, from a distance, and in bad lighting.

The second reason eyewitness testimony is unreliable has to do with the nature of memory. Neuroscience now knows that every time we remember a memory, that memory is "reconsolidated," slyly remade and reconfigured. The idea of reconsolidation should make us distrustful of our memories. They do not directly represent reality. Instead, they are an imperfect copy of what actually happened, a Xerox of a Xerox of a mimeograph of the original photograph. This directly contradicts an underlying assumption of eyewitness testimony, which assumes that our memories are immutable impressions, locked away in the brain's vast file cabinet. Recalling the memory shouldn't change the original memory.

But it does change the memory. The very act of remembering what you saw changes the neural substrate of the original memory trace. For more on this research, check out the work of Karim Nader.

Of course, our legal system will always need eyewitness testimony. The important thing is to make eyewitness testimony more reliable, to design legal situations that discourage eyewitness biases and mistakes. More states should follow the lead of New Jersey.

But I think this research also speaks to another controversy perpetually roiling the legal system: the death penalty. The awareness of our own epistemic limitations - and neuroscience is great at illuminating those limits - should make us skeptical of our verdicts. People make sensory and memory mistakes all the time. That's just the nature of the brain. But nobody should die because of a mistake.