HELEN Paterson has the work of an arsonist to thank for her fascination with the reliability of witness memories.

One night in the late 1990s, when she was a psychology student living near the University of NSW, Paterson was woken in the night by the wailing of a smoke alarm in the hallway outside her flat. Rubbing sleep from her eyes, she rushed to open the door and saw a line of flames running from the middle of the carpeted hallway directly towards a neighbour’s front door.

“The firefighters said the fire was started by something near the door,” recalls Paterson, now a lecturer in forensic psychology at the University of Sydney. “The old woman who lived in the flat couldn’t think what could have started it, and I said it must have been arson . . . There was a line of fire leading to the door.”

Arson experts confirmed the fire had been deliberately lit, so Paterson found herself being interviewed as a witness by police. At the time she was learning about interview techniques at university, so she paid close attention to their approach.

“I told them that when I walked out of my apartment, I looked to the left and saw a line of fire heading to her door,” she explains. But the policeman insisted that this wasn’t possible. “He said no, you’d have had to go around the corner to see that.”

Paterson stuck to her guns. “Because it was my own apartment building, I felt quite confident about the location,” she says. “But if I’d witnessed something in a dark alley that I was not familiar with, I would have allowed my memory to be swayed.”

Paterson moved out of the building, but her fascination with the ways memories can be contaminated by others remains the focus of her academic career.

It’s a topic that has considerable significance for the criminal justice system.

“Usually [in] conversations we never let the truth get in the way of a good story, and it’s OK if we exaggerate things,” Paterson says. “But within the legal setting accuracy is so important. They want to know what actually happened in the event and sometimes all you have is witness evidence.”

Even with the advent of DNA evidence and other hi-tech tools, human memory evidence remains persuasive, especially for juries, she says. “If someone says, ‘Look, I was there, I’m 100 per cent confident he’s the one who did it’, it’s very compelling.”

But, as Paterson and other experts have shown repeatedly, memory is incredibly flexible and subject to various influences.

In their latest study, published in Applied Cognitive Psychology, Paterson and colleagues investigate how two witnesses to the same event can influence each other’s memories by discussing the episode after it happens.

Using University of Sydney students as volunteers, they set up two computer screens and told pairs of students they would both be shown footage of the same crime, a robbery.

“What they [didn’t] realise is that we actually [had] two slightly different versions of the crime video and they each were shown just one version on their individual computer monitor,” Paterson explains. The experiment was designed to reflect the realistic situation in which two witnesses might see things from slightly different angles or remember matters slightly differently.

The researchers then asked the pairs of volunteers to discuss with each other what they saw in the video, before being interviewed about it.

Even though the videos of the crimes were subtly different — for example, in one the criminal might have been wearing a cap, while in another he was not — when interviewed, people consistently said they remembered specific details that weren’t in the version of the video they watched.

“The things they report were in the other version of the video, so the only way they could have got that information was through the co-witness discussion,” Paterson says, adding that even after they warned participants that their partner might have seen a slightly different version of the video and that they should report only what they remembered from their version, volunteers still incorporated details from the alternative form.

At such a point, separating the real memory from the false one can be difficult, if not impossible. Clearly, memories are malleable.

In another experiment, researchers obtained photographs of study participants as children from their family members and confirmed with the family members that they had never been in a hot air balloon. The researchers then doctored photographs so it looked as if the participants had been in a hot air balloon when they were young. After several interviews, half the participants recalled, partially or clearly, the fictitious hot air balloon ride.

In yet another experiment, Paterson and Richard Kemp, at the University of NSW, had a man stand at the lecture room door and interrupt the class to make an announcement. A week later, they went back and asked the students in the class to explain what happened. Unbeknown to the class, the team had planted a collaborator whose job was to claim he remembered there being two men standing at the door.

“It’s amazing how many people believed that, yes, there was a second person,” Paterson says.

“They will even attempt to pick this person from a line-up, [even though] he never existed. We even had one person say, ‘I remember the second guy was standing by the door scratching his armpit.’ He didn’t even exist, but they have this vivid memory of this guy . . . scratching his armpit.”

For the police and courts, ensuring witnesses don’t contaminate one another’s memory in this way is a tricky business.

In courtrooms, witnesses are prevented from hearing each other’s testimony in an attempt to avoid this problem, but the reality is that by then the horse has bolted. Case in point: a survey conducted by Paterson and her colleagues showed that when there was more than one witness to an event, in 86 per cent of cases, they would have discussed details of the event with a co-witness. They’ve also interviewed police officers who said co-witness discussion was very common, often taking place before police arrive on the scene.

“For the most part, police officers have received instructions to separate witnesses and instruct them not to discuss the event,” Paterson says.

“But even when they do this, it can be difficult, [if not] impossible, to prevent discussion. As soon as they leave the sight of the police officers, witnesses will talk among themselves about the event.”

University of New England law professor Eilis Magner argues that work on ways witnesses can influence each other’s memories is critical: “You need to be able to trust what witnesses say . . . If memory is malleable then we have a huge problem and judges need to warn the jury to be careful.”

According to Don Thomson, a barrister and memory researcher at Deakin University in Victoria, the greatest concern over the malleability of memory probably is in cases involving children.

“What [Paterson] has done with adults becomes accentuated with children,” he says. In cases of suspected sexual abuse, for example, Thomson is concerned that discussions within families about “what happened” can affect what a child thinks they remember.

“To have the fiction that this phenomenon doesn’t exist can lead to miscarriages of justice,” Thomson warns.

“I think it’s terribly important that the judge and jury are informed of precisely what has happened between the alleged event occurring and the evidence being given, so they can make judgments about whether memories might have been influenced.”

Beyond greater awareness, police forces and researchers are working on ways of trying to ensure witnesses to crimes can document their recollections without having them contaminated by others, Paterson says.

In Britain, for instance, she says trials are under way in which police give notebooks and pens to witnesses at the scenes of incidents so they can write down what they saw without discussing it with anyone.

“Police typically don’t have the resources to monitor people and make sure they are not discussing things,” Paterson says. “That’s why I think the self-administered interview is possibly a very good solution.”

As for her own encounter with police interviewing techniques back in the 90s, Paterson says she left the country after the arsonist struck but heard that he soon lit another fire. He did more damage that time. Or at least that’s what she remembers.