Published on 2026-07-25
Is 'I saw it with my own eyes' actually good evidence?
Eyewitness misidentification played a role in most DNA exonerations of wrongly convicted people. Seeing something happen is not the same as remembering it right.
There's a sentence people reach for whenever their certainty needs backup: 'I saw it with my own eyes.' It ends most arguments. It's meant to close the case. And it happens to be one of the weakest kinds of proof that psychology has spent the last fifty years quietly taking apart, not because eyes lie, but because seeing something and correctly remembering it afterward are two separate jobs, and the second one fails far more often than we assume.
Attention was never watching everything
In a 2015 study, psychology researchers Noelle Sammon and John Bogue had 126 participants watch a minute long video of a staged theft in which the actor playing the thief was secretly swapped for a different person partway through, a classic change blindness setup. Some participants watched with full attention. Others listened to music through headphones at the same time, splitting their focus the way a real bystander's attention often gets split by traffic, conversation or simple distraction. Twenty minutes later, everyone tried to pick the thief out of a lineup of twelve photos. Just under 27 percent of all participants correctly identified the actual perpetrator, and people whose attention had been divided during the video did noticeably worse at picking out the real culprit than those who'd watched undistracted. Nobody in that study was lying about what they'd seen. Their eyes had simply not been recording what they assumed they were recording.
What courts learned the hard way
This isn't just a curiosity from a psychology lab. According to the Innocence Project, a nonprofit that works to overturn wrongful convictions, more than 60 percent of its clients, people later proven innocent through DNA testing, were originally convicted at least partly on the strength of an eyewitness pointing at the wrong person and saying, in effect, I saw it with my own eyes. The organization separates the causes into two groups. Some sit inside the witness: stress during the event, distance from what happened, poor lighting, the well documented difficulty of recognizing a stranger's face across racial lines. Others sit inside the system: how a lineup gets shown, how a question gets phrased, how confident an officer sounds while asking it. Neither category involves the witness making anything up. Both categories involve a sincere, confident person turning out to be wrong.
A video doesn't fully rescue you either
It's tempting to think this problem disappears once there's footage: surely watching a video with your own eyes is different from watching a live event unfold. It helps, no question, a recording exists and can be reviewed. But the memory you build of what a video showed is still built the same reconstructive way any other memory is. People routinely swear a video showed a hand reaching for a weapon, or a crowd of a certain size, or an event happening in a certain order, when a careful rewatch shows something a little different from what they now believe they clearly saw. The footage is fixed. The account of the footage, once it's passed through someone's memory and been repeated a few times, is not.
Your eyes are not a camera, and even when a camera is running, your memory of what it showed is still yours to reconstruct.
None of this means witnesses are useless or that you should distrust every memory you form. Most everyday recollections are perfectly serviceable: what you ate, where you parked, whether you locked the door. The trouble concentrates in exactly the high stakes moments where certainty matters most, brief, stressful, high pressure scenes that are precisely the hardest conditions for accurate memory to form in the first place. The more a situation demands a confident, detailed account, the more that account deserves a second look rather than automatic trust, including from the person who lived through it.
It also helps to remember that confidence and accuracy grow from different soil. A witness can feel entirely certain while looking straight at the wrong face in a lineup, the same way you can feel entirely certain about a video you half watched while doing something else. Feeling sure is a signal about how strongly a memory has settled in your mind, not a measurement of how closely it matches what actually happened in front of you.
So before 'I saw it with my own eyes' settles an argument, it's worth asking what conditions that seeing actually happened under, and whether the certainty in your voice is measuring the event, or just measuring how many times you've told the story since.