Memory: Accuracy of Eyewitness Testimony

This should be a fun topic for any (future) lawyers, private investigators, members of the police force, criminologists…  Yeah, I think it’s just a pretty fun topic.

Obviously, when someone commits a crime, we have to find evidence to convict them, or to find the right person to convict.  Sometimes, the physical evidence isn’t present, so we have to use eyewitness testimony, which is sort of what it sounds like.  It’s when people who’ve seen an incident taking place tell an investigator – or a courtroom – what they saw.  It’s the sort of thing you see in every episode of Law and Order, and probably about every fifth episode of Coronation Street, if memory serves.  (My best wishes and thoughts to David Platt, the true symbol of British pride).

However, eyewitness testimony is not always accurate, and this can be for a number of reasons.  The first is misleading information, and the second is anxiety.  Both of these are quite heavily study-based, but it’s not like Social Influence, where you’re given far too much information about a single study at once.  Accuracy of Eyewitness testimony is quite doable, if you ask me.  And, considering you’re reading this blog, I’d wager that you are sort of asking me, a bit.

Let’s start with misleading information.  Within misleading information, you have leading questions and post-event discussion.  A leading question is when someone is asked a question which causes them to alter their perception of the way an event took place, subconsciously or otherwise.  Post-event discussion is when witnesses talk to each other after an event, and their memory of the event becomes contaminated.

Loftus and Palmer carried out a study on leading questions.  This is one of those studies where they write ‘key study’ in the title, so if you’re taking notes, you should probably note this one down, just to be safe.  The first study they carried out, they got 45 participants and asked them to watch a video of seven different traffic accidents.  There was a critical question: “how fast were the cars going when they hit each other?”.  In each condition, the verb was different, with the other four conditions using the words smashed, collided, bumped or contacted.  These impacted speed estimates, with ‘smashed’ causing participants to estimate a speed of 40.8 miles per hour, whilst ‘contacted’ caused the lowest mean estimate of 31.8 miles per hour.

Loftus’ research led her to suggest that eyewitness testimony was generally unreliable, and should not be used as evidence in court (she’s quite passionate about this, if you wanted to look up one of her Ted Talks to see her talking about it in action).  Other researchers point out that a lab experiment does not arouse a witness’s emotional state in the same way as actually witnessing an event.  Yuille and Cutshall found that eyewitnesses to a real armed robbery in Canada generally gave very accurate information about the event, further reinforcing the idea that emotional state impacts accuracy.  We’ll cover this further a little bit further down, in the effects of anxiety on eyewitness testimony – with another armed robbery study, no less.

In fairness to Loftus, there is evidence to suggest that inaccurate information from eyewitnesses is one of the number one reasons behind people being falsely convicted.  DNA exoneration cases have proven these concerns hold some weight.

In the second experiment, the same participants were asked if they had seen any broken glass.  In spite of the fact that there was no broken glass on the scene, those in the ‘smashed’ condition answered ‘yes’ 16 out of 50 times, compared to just 7 out of 50 times in the ‘hit’ condition and 6 out of 50 times in the control condition.  This suggests that a leading question can change a participant’s actual memory of an event.

In post-event discussion, there are a couple of reasons why information can become contaminated.  One is the conformity effect and the other is repeat interviewing.  The conformity effect suggests that witnesses can come to a consensus view on the details of an incident.  This was the case in a study by Gabbert, wherein 71% of witnesses who had discussed an event went on to mistakenly report false details when questioned.  Repeat Interviewing considers the fact that comments from an interviewer can become ingrained in a witness’s memory.

Loftus studied a group of college students, all of whom had visited Disneyland, and asked them to evaluate the advertising material, which involved either Bugs Bunny or Ariel, in spite of the fact that neither of these characters could have been at Disneyland at the time.  In spite of this, participants in the ‘Bugs’ or ‘Ariel’ conditions were likely to report having shaken hands with the characters, suggesting that misleading information does have an impact on the accuracy of eyewitness testimony.

The elderly are thought to be more easily misled than younger witnesses, as they tend to struggle with remembering the source of the information (though the memories themselves are not impaired).  This suggests the importance of individual differences in eyewitness testimony.

There’s one more evaluation point, on response bias, but it’s not laid out very clearly, and I’d rather not try to explain something I don’t understand.  Four evaluation points for misleading information is more than enough, though.  You’ll do fine – and, hopefully – so will I, knock on wood.

We’ve done misleading information, so let’s move on to Anxiety.  In anxiety, we have anxiety itself and whether it improves or worsens memory, for which we have a study called the Weapon Focus Effect.  I like the Weapon Focus Effect, because the whole thing inspires the memory of a ridiculous newspaper comic, where everyone has a very round head and over-exaggerated facial expressions.

Johnson and Scott carried out the weapon focus effect.  There were two conditions, and in both, participants sat in a quiet waiting room.  In the first condition, a confederate ran through the room holding a pen covered with grease (low anxiety).  In the second condition, the confederate was holding a knife covered in blood (high anxiety).  They were then asked to identify the person holding the instrument.  In the ‘pen’ condition, mean accuracy of identification was 49%.  In the ‘knife’ condition, it was 33%.  Loftus suggests that anxiety pulls witnesses’ focus to the details of the crime, which was supported when the focus of people’s vision was found to be on the weapon.

This being said, the Weapon Focus Effect might not be caused by anxiety, but by surprise.  To test this, Pickel carried out a study with high and low threat items and high and low surprise items in a hair salon (most crucially scissors and a whole raw chicken were used).  She found that identification of people was least accurate in the high surprise conditions, supporting the hypothesis that surprise is the cause of the weapon focus effect.

However, anxiety can also produce a positive effect on the accuracy of eyewitness testimony.  Christianson and Hubinette found enhanced recall when they interviewed 58 eyewitnesses to bank robberies in Sweden.  They conducted interviews on bank tellers and bystanders four to fifteen months after the event had taken place.  All showed recall higher than 75%, and bank tellers, who were in the highest anxiety condition, had the best memory of all.  This suggests that anxiety actually has a positive effect on the accuracy of eyewitness testimony.

This is a helpful study because it uses real-life studies, which improves its ecological validity.  However, Deffenbacher’s meta-analysis of 34 studies suggests that whilst lab experiments show that anxiety reduces the accuracy of eyewitness testimony, real-life events show that it reduces accuracy even more.  This puts Christianson and Hubinette’s research at odds with other existing research.

As such, it should perhaps be noted that there are no simple conclusions that we can draw about the effects of accuracy on eyewitness testimony.  Halford and Milne found that witnesses to violent crimes tended to have more accurate recall than witnesses to non-violent crimes.  This could account for the contradictory research by Christianson and Hubinette.

Yet, the research by Christianson and Hubinette doesn’t remove the research by Johnson and Scott – so we have a contradiction to resolve.   Deffenbacher studied 21 studies on the effect of anxiety on eyewitness testimony.  Of these, 10 showed anxiety as having a positive effect on eyewitness testimony, and 11 showed it as having a negative effect.  Deffenbacher’s explanation for this was something called the Yerkes-Dodson effect, which suggests that a moderate amount of anxiety strengthens accuracy, whilst an extreme amount lessens it.  It might be helpful to think of this as being like elastic, in that you can only stretch it up to a certain point before it snaps or loses its elasticity (my thoughts go out to all the hairbands I’ve lost like this).

A pair of researchers called Fazey and Hardy redeveloped the Yerkes-Dodson model, with their newer model being favoured over the Yerkes-Dodson model by Deffenbacher.  They call it the catastrophe theory.  They suggest that when there is an increase in cognitive anxiety, not just physiological arousal, it can cause a sharp decline in accuracy of recall.  Individuals with higher self-confidence – or a more “stable”, as opposed to “neurotic” personality – are more likely to exhibit the inverted ‘U’ characteristics of the Yerkes-Dodson Model.

That’s this all finished up.  We’re onto our last topic in Memory now, and then it’s onto Attachment!  Pray for me, if you’re of the praying inclination, to finish writing these posts before Wednesday, which is when my exam is.  I’m very glad I did those broad overview posts beforehand, or I would be – in crude terms – “up shit creek”.  Or not – because, believe it or not, this isn’t the only psychology revision I’m doing.