Talk to psychologist Matthew Gerrie and you begin to wonder how many eyewitnesses to crime - and whose testimony leads to a conviction - can comfortably look themselves in the eye.
Memory is such an unreliable witness, he explains, vulnerable to manipulation and trickery. And - here is where memories can be downright deceitful - people with false memories tend towards absolute confidence that they are right.
The trouble in a courtroom is that judges and juries tend to place great store in eyewitness testimony - and the more confident they are, the more detail they flourish, the more the court is swayed.
"People really do think that they truly remember the person when they make a false identification," says Gerrie. This harks back to research on false memories and how memories are reconstructed.
"What we know is that, if you point to the wrong person, you can create a false memory that that person was indeed the offender when you can be entirely wrong."
Gerrie, a research fellow at Victoria University's school of psychology, is one of a handful of academics lending their skills to New Zealand's version of the Innocence Project, which investigates possible miscarriages of justice. And at the heart of many wrongful convictions is the testimony provided by eyewitnesses.
His research into the eye movements of eyewitnesses as they identify suspects this week won the science and society category of the Foundation for Research and Technology's MacDiarmid Awards for young scientists.
The Innocence Project, a joint venture involving Victoria and Otago University researchers, is meanwhile poised to lodge its first bid to overturn a conviction.
The courts rely heavily on eyewitness testimony in criminal convictions from street muggings and bank robberies to murders. But there's increasing awareness among lawyers, the police and judges of the potential to get it wrong, and the psychological factors which affect eyewitness recall.
It's a dilemma most of us who would struggle to describe our best friend to a police officer can relate to - how tall was the robber, what colour hair and eyes, was that jacket blue or grey? Was that a scar on his cheek or a six o'clock shadow? Then there's the role prejudice and life experience can play on our memories.
In the United States, eyewitness error is a factor in 75 per cent of wrongful convictions. In New Zealand, reliance on rape victims' identification of the perpetrator has resulted in two convictions which were subsequently overturned. Gerrie says New Zealand is no less susceptible to unsafe convictions than overseas countries. "What these cases show us is that these are systematic errors that can be prevented if we can just come up with the right techniques and the right way of selecting [reliable] evidence."
Three years ago, retired judge Sir Thomas Thorp released a report into potential miscarriages of justice which estimated as many as 20 people may be wrongly in New Zealand jails. He examined 53 applications to the Justice Ministry and found most raised issues deserving further investigation.
Gerrie, 31, became interested in memory distortion and crime while studying psychology. He says memories can be changed for all sorts of events - be it a crime they witnessed or a childhood event. He says the leading causes of wrongful conviction are all psychological ones - eyewitness misidentification; misinterpretation of lab tests which are then presented to juries; and false confession, either voluntary or through coercion. "So there's great potential for psychological science to have a large impact on the courtroom, just as forensic science does."
The classic TV image of an eyewitness behind one-way glass inspecting a line-up of dodgy characters is now a rarity in New Zealand - partly because many suspects refuse to parade and partly because of the difficulties police face in assembling a line-up. Photo montages are the norm, using mugshots stored in the police database when suspects are fingerprinted and photographed.
But there's similar potential for eyewitness error.
"We have known for a few decades that eyewitnesses tend to use a bunch of different decision processes when choosing a person from a line-up," says Gerrie. "Some of these involve simply relying on memory, which is what we want the witness to be doing. But there are [situations] where the witness uses a process of elimination to figure out which person they saw at the crime scene."
It's this process which is more likely to lead to misidentification. People do it without necessarily being aware of it, he says. The presentation of a line-up or montage, where all members can be seen at once, encourages the process. There's also a reluctance to reject the entire line-up.
Gerrie's current research is looking at the processes eyewitnesses use to identify suspects - research of undoubted potential to police to help ensure the right suspect is put behind bars. He and research assistant Tom Huthwaite began using an eyetracker, similar to the machines optometrists use for eye examinations, to distinguish the processes people use when making an identification. An infrared camera centred on the pupil of the eye detects exactly where the person is looking. In early studies, Gerrie found he could distinguish between when people were relying on their memory - the preferred decision-making process - and when they were using elimination. A computer records the number of comparisons the person makes.
"What we found was that these comparisons went up as they were using a process of elimination. And the more they compared among the faces the less accurate their identification tended to be."
The research is in its infancy but may lead to more reliable eyewitness decision-making.
Police are making efforts to improve their handling of eyewitness interviews and the identification process to reduce the potential for error. Line-up (or montage) selection is carefully considered.
"If police include [only] people who look similar to their suspect, people tend to rely on a process of elimination. The way to encourage people to use their memory is to include people who match the description they gave of the person they saw at the crime scene, so that they have to go further and above the description they gave of the offender."
Awareness has grown of the need to avoid corrupting the eyewitness by suggesting "whodunnit", giving biased instructions or using positive reinforcement after identification has been made, which is likely to bolster a witness's confidence when giving evidence.
"Giving them a thumbs-up or telling them they did well can increase confidence and we know that juries tend to believe witnesses who are highly confident and give lots of details.
"And we know from false memory and false identity research that the people who tend to have false memories tend to be the most confident and give lots of different details."
Just why our memories play such tricks is a matter for further research. "What we think it comes down to is a simple principle known as discrepancy detection," Gerrie says. "The criminal justice system wants our brains to work like a computer, or a tape recorder. It wants us to be able to distinguish perfectly between information which we saw at the crime scene and information which was given to us from some other source.
"But that's not the way our memory works. When we see a crime and then we are given a suggestive piece of misleading information about it, we find it very difficult to distinguish between the two because our memory is reconstructive. Sometimes we're reconstructing it with true parts of the event and sometimes with the suggested details as well."
Should the public - and more importantly, judges and juries - be much more wary of eyewitness reports?
That's not what Gerrie wants. Eyewitness testimony is such a key part of investigations it would be dangerous to stop using it, he says. The task is to come up with better techniques for testing people's memories and preserving the evidence that comes from memories.
"The first step is to treat an eyewitness memory just as we would any physical evidence at a crime scene. We know not to trample through a blood splatter because that may hold information which is required to identify the perpetrator. We should do exactly the same with an eyewitness memory - we shouldn't trample through it, we shouldn't disparage it with all sorts of suggestive information or bad interviewing techniques that will distort it and make that memory unusable later.
"Instead what we need are good techniques that are reliable at not changing people's memories but maximising the amount of accurate information that is coming out of it."
Dr Matthew Gerrie will talk about his research at Auckland University on Wednesday, September 2 at 4pm in the Human Sciences Building, 10 Symonds St.
MEMORY WORK FOR NZ POLICE
The psychological factors which affect memory are acknowledged in new witness interviewing techniques being introduced by the New Zealand Police.
More than 6500 frontline staff are being trained in interviewing skills designed in part to ensure the right suspect is convicted.
Detective Inspector Ross Grantham, of Police National Headquarters, says the psychology of memory is at the heart of the training. The programme encourages interviewing methods that do not lead the witness or complainant, asking open-ended rather than closed questions and "free recall" conversation management. Interviewers ask witnesses to go back over initial statements to add to the picture and gain an understanding of the lead-up to the crime.
The programme is based on the Peace model developed in Britain - planning and preparation, engage and explain, account, closure and evaluation.
Some of the techniques are considered too complicated for all frontline officers to be able to deal with, but nearly 100 investigators have so far received training in enhanced cognitive interview techniques.
CONVICTIONS UNDER REVIEW
Since its launch in 2007, the Innocence Project has received dozens of inquiries from people wanting convictions reviewed. It hopes within a month to lodge its first application asking the Governor General to refer a case back to the Court of Appeal - that of Alan Hall, who spent eight years in prison for the 1985 murder of 52-year-old Arthur Easton during a robbery in Papakura.
Hall has been on life parole since his release from prison in 1994.
Project manager Matthew Gerrie says it is looking at a number of other cases, some of which relied on eyewitness identification.
Gerrie says New Zealand is fortunate in having two researchers of international standing working in this field - Innocence Project co-directors Maryanne Garry of Victoria University and Harlene Hayne of Otago University. Garry is an expert in adult false memories while Hayne's specialty is the way children's memories develop and how they should be forensically interviewed. She has urged reconsideration of Peter Ellis' 1993 conviction for sexual abuse at the Christchurch Civic Creche.
Associate director Rachel Zajac, a psychology lecturer at Otago, received a $170,000 grant from the Marsden Fund in 2007 to further her research aimed at reducing false identifications by placing wildcards in the line-up. The wildcard has been shown to increase witnesses' willingness to reject line-ups that do not include the perpetrator.
In New Zealand, the 1993 conviction of David Dougherty remains the most high-profile example of wrongful conviction based on mistaken identity. Dougherty was jailed for seven years after an 11-year-old rape victim identified him as her attacker.
He was freed in 1997 after new DNA evidence indicated he was not the rapist.
In 2005, sickness beneficiary Aaron Farmer was found guilty of the late-night rape of a 22-year-old woman in Christchurch after she picked him out from a photo montage. His conviction was overturned in 2007 after fresh testing of DNA found on the victim.