This originally started as an investigation into the various ways that error can be introduced into eyewitness (as opposed to expert witness) testimony. With over two decades of human factors experience under my belt, I am well aware of how memory works, the challenges it faces, and how counterintuitive much of it is. But for this study, I wanted to limit the scope of my review to studies specifically about eyewitnesses. For Daubert hearings (where the judge decides on the scientific validity and relevance of the expert for the case), it is always helpful to be able to cite specific, focused research because judges don’t always grasp the generalizability of cognitive research.
Even with the cynicism that my experience has forced upon me, I was shocked by just how much evidence there is that eyewitness testimony is perhaps the least reliable kind of evidence available to the court (since hearsay and total bullshit are usually left out). I plan to write this up in more detail and complete with references for the Annual HFES Conference in the fall. But I wanted to get something more informal out now while the emotion is still fresh and I can give you all a laugh.
Let’s start with the context in which the event is witnessed. First of all, the witness is rarely looking directly at the location where the incident occured. Until their attention is refocused, if it is at all, then it is the notoriously inaccurate peripheral vision that starts the memory out. And even then, the contents of the witnesses’ existing vision, thoughts, beliefs, daydreams, stereotypes, and past memories are stronger contributors to what gets encoded. If the witness has been in similar situations before, these are all blended together into the memory of this event. This causes the witnesses’ memory to condense, smooth out, fill in gaps, omit inconsistent details, and reorder the sequence to fit their personal experience and expectations.
After the witness does notice that something interesting is going on and refocuses, their attention is captured by the most salient and central attributes rather than what is important to be an effective witness. We know this as the “weapon effect.” They remember the gun in incredible detail but couldn’t tell you how tall the perp was, the color of his hair, or about the big stain on his pant leg. This is hilariously captured in the famous Invisible Gorilla test by Daniel Simons.
If it is a stressful event for the witness, the fight or flight instinct starts the release of hormones that degrades memory (who cares what the tiger looks like, just get the hell out of there). Attention narrows to just the most salient attributes and anything more subtle is missed.
Then what happens after the event introduces even more inconsistency into the witnesses’ memory. We all have a tendency to exaggerate the event when we first tell the story, especially if this first telling is to ourselves, friends, or especially the media. By the time the on-scene investigator starts his or her interview it is way too late. And unless the investigator is both well trained and well intentioned, it is easy to implant whatever memories will get the case solved as fast and easy as possible. Consider the difference between these questions:
· What happened . . . . . ?
· Was there a weapon?
· Was there a gun?
· What color was the gun?
In this last case, witnesses who never saw a gun all of sudden remember one clearly. And then later on, when they are recounting the story, the gun becomes a permanent part of their story. By the time trial rolls around, the witness is absolutely sure there was a gun. The prosecuting attorney can make it even easier by asking the same question, “What color was the gun?” Just the natural desire for internal consistency would cause even the most well meaning witness to give the same answer as they did the first time.
What makes this even worse is that juries believe eyewitnesses more than they believe most other kinds of evidence. As Groucho Marx famously said, "Who are you going to believe, me or your lyin' eyes?"