D&M Ch 8 - Law and Eyewitness Accuracy

Issue: eyewitnesses are supposed to share what they saw, everything they saw, and nothing else. But we've seen how inaccurate memory can be and how difficult it is to not fill in gaps -- the witnesses are essentially swearing to something impossible. But people tend to believe it is possible, and juries in particular have a preference for strong certainty. Witnesses can be wrong. People fail to detect lies. Also, hindsight bias -- the "I knew it all along" effect -- inadmissible testimony still taints their judgement.

Ex: show video of a crash, people are likely to say the speed was faster if asked how fast the cars 'smashed' into each other than if asked how fast they 'hit' each other. Repeating false information once or even multiple times can cause people to remember incorrectly. This is what makes lying politicians and news accounts so disturbing. Or people respond to fabrications with extreme hindsight bias, saying, that they know something's not true, but it's something that person would say, so it doesn't matter (like that Twitter fabrication described here)

Ex: implanted memory of being lost in a shopping mall. So easy to get people to recollect things in detail that did not happen, just by telling them it did with certainty.

Confidence usually means truth, but we know from the lab that people are frequently overconfident. How do people have such high confidence about something that didn't happen? How does confidence normally work as a reliability index? How can confidence be manipulated? Can confidence ever be infallible?

Many examples of high-confidence memory being incorrect.

Where does confidence come from?

• Frequency of exposure increases confidence.
• High confidence errors lead to a particular sort of surprise.
• Confidence increases people's level of belief in the statement.
• Usually confidence is a good marker, but confidence can be manipulated until it does not discriminate between true and false memories.
• Asking for memory source increases ability to identify actual memories from suggested/manipulated memories.
• Actively engaging metamemory processes improves memory identification. Living mindfully improves reflection

The cognitive interview: A good witness interview guides you through memory-enhancing procedures -- think about the context, including surroundings and your reactions; reporting everything that comes to mind, even if fragmentary or inconsequential; remembering events in different orders; recalling events from different perspectives. Allow to proceed interrupted. Do not suggest content.

Ex: faux patient bio, given to clinicians. Asked to make inferences based on background, given more background over time. Their confidence increased, but the accuracy was no better than chance.

Is Confidence Ever Infallible?

What if the events were traumatic? Extreme trauma and stress seems to produce less reliable memory.

Witness Confidence in a Trial

Experimental trial - witness confidence was the most powerful predictor of conviction. However, if caught in a mistake, a more-confident witness might appear less credible than a less-confident one.

What about lying?

deception cues -- signs of lying
leakage cues -- signs of the truth despite the lie

Lying is revealed more often in microexpressions -- usually only visible on slow-motion replay of a video. Sometimes lying is more cognitively demanding, so results in more pausing, errors, brow-furrowing, speech errors. But basic fidgeting and gaze averting is not a good sign. Extensive studies show that people are not good at detecting lies, even with training and expertise, with small exceptions.

Repeating a lie makes people more likely to believe it.

Hindsight Bias

AKA the 'knew it all along' effect
Given an unknown, then told the answer, people do a bad job of estimating retrospectively or to act as if they did not know the outcome.

OJ Simpson case -- similar -- people made stronger judgments looking back, saying that they knew that's how it would go.

Why can't people look back with accuracy on their views about an unknown?

• option: personal needs view. People like to be right -- need varies in how strong it is. Some people have low tolerance for ambiguity (used the 'Rokeach Dogmatism Scale'). Children and older people may have lower tolerance for ambiguity. But this isn't sufficient to explain the prevalence, some people who score fairly low on the personal needs ranking still display hindsight bias.
• option: memory view. Purely cognitive. Structure of hindsight bias is similar to mis-remembering things. Actual memory blends with what is suggested or learned later. Hard to retrieve the earlier information.
• option: anchoring view. Outcome pulls your perception toward it -- this overall effect has been shown widely.

I myself have found that when I have a 'knew it all along' response to a situation, it seems like a result of a feeling of "fit" -- the result is consistent with my thinking up to that point, and it all plugs together in a familiar/recognized way. For example, with the Trayvon Martin case, I found myself thinking both ways -- that the evidence as it was reported to me surely pointed toward conviction, and that the prosecutorial conduct of the case pointed toward acquittal. Once the verdict came out, the story came together. I also know that I was not in the jury room, and I am sure that there are many cases which look clear in the media but which are conducted in a way that the jury decides that the case has not been made effectively. This is a subtlety of the system that can be overlooked, but a careful juror is trying to make a decision based not on "what really happened" but rather "what the evidence proves". You hope that the evidence tells you what happened, but people lie, evidence gets overlooked or is simply lacking. So a vote to convict and a belief about reality are really different things. It does seem that people's ability to draw these types of distinction is susceptible to bias -- evidence may be weighed differently depending on what people suspect to be true based on race, gender, age, etc. A tighter and more vigorous case may be needed in situations where people are biased toward empathy for the accused.

Inadmissible evidence is very difficult for juries to ignore, for example a coerced confession when coupled with circumstantial evidence. In studies, even obviously coerced confessions led to higher conviction rates. Malpractice results look similar -- hindsight is 20/20 (the "retrospectoscope"!). Looking back, it seems obvious that everyone should've known there were issues. And yet in the moment, people may not.

Effect is pervasive -- a brain-teaser is obvious after the fact. A new discovery is something people 'knew all along'.