Why can't we trust what we see?

The human memory can be impressive, but it is equally prone to letting us down. Now groundbreaking research has revealed the extent of just how fragile it can be - and how to use it better.

You're in the pub and trouble starts. There is shouting, someone is stabbed, they die. It happened right in front of your eyes and the police want to speak to you. But what exactly did you see?
It's long been accepted that eyewitness testimony may not always be as reliable as it seems. The problem is people simply don't remember exactly what happened, say psychologists. The mind does not work like a video camera, nowhere in the brain is the perfect memory of everything that has been seen, in the order it happened.

Can you pick out the criminal?

But however fallible human memory is, it's often the only thing police have to go on and eyewitnesses have been responsible for sending people to prison ever since the justice system began - both rightly and wrongly.
Now research has gone further than ever before to understand the fragile nature of our powers of recall.
The project - involving the Open University, the BBC and Greater Manchester Police (GMP) - is groundbreaking in several ways. The technology it used is cutting edge, including eye trackers - devices for measuring eye positions and eye movement. But just as unparalleled was the realism involved. It's always a big issue with research - how do you accurately test someone's everyday reactions when they know they are part of an experiment and in controlled conditions? Won't they try harder to remember details if they know they are doing a memory test?
'Appalling'
In this case the important action took place when they were least expecting it. The 10 volunteers were put through days of memory tests in a studio and assumed this was the research. In fact, two intricately planned and elaborate mock crimes - a fatal stabbing and an armed robbery - were really what mattered.
On one day the participants went for lunch in a local pub, which was really filled with actors, stuntmen and 10 hidden cameras. A fight broke out and someone appeared to be stabbed and killed. The whole scenario unfolded over 20 minutes.

Eyewitness Simon Woodthorpe The whole experience really surprised me. I'd never had a problem with my memory before but had never been in a situation where I'd been asked to recall things in that much detail. I was good at describing the overall situation. But when it came to the real detail of what had happened, I was actually creating a lot of what I thought I'd remembered. I'm not sure I would be confident about saying I'd seen anything after this. I'd be worried about incriminating someone.

"It was incredibly realistic," says Simon Woodthorpe, 44, a photographers' agent who was one of the volunteers. "We weren't expecting it at all and only started to get suspicious when the police turned up really quickly. By then it didn't matter, we'd not suspected it was staged so we hadn't consciously thought about paying extra attention to all the details.
"I always thought I had a good memory, but I was yards away from the incident, saw it all unfold and still got the murderer wrong. I said it was the wrong man."
What was also unprecedented about the project was the access to interviewing techniques used by GMP. Detectives treated each mock crime as if it were real, interviewing the volunteers, but unlike a real case, what they said could be checked against what exactly went on.
During the drama the eye trackers - still being worn by some volunteers during the mock robbery - were able to pinpoint exactly what people were looking at and compared to what they reported. The differences, say those involved, were in some cases staggering.
'Lifeblood'
"One person thought they hadn't seen the crime being committed, they were adamant about it," says Dr Graham Pike, a memory expert involved in the project. "When we reviewed the eye tracker we found they'd actually spent almost the entire time looking at it unfold. It was quite amazing."

FIND OUT MORE Episode one of the three-part Eyewitness is broadcast on BBC Two beginning on Sunday 18 April at 2250 BST Or listen here later

What makes the memory this vulnerable is the fact it is malleable and not fixed, says Dr Pike.
"It's not like inputting data into a computer, the mind does not store facts absolutely the way they are and it does not recall them absolutely accurately either."
There are three stages in memory, according to modern pyschology. The first is perception, which is what we see - also what we hear, taste, touch and smell. This in itself is a selective process. From the start we can fail to encode detail or simply not notice something, so the information going in isn't accurate. Secondly there's storage. We know we forget things over time, but we also revise our memories and re-write them to fit in with new ideas.
Finally there's the retrieval stage, where the brain searches out information. When you remember something, lots of different parts of the brain work together and from that emerges the mental representation that is going to be your experience of a memory. Every time you recall something, you reinterpret it all over again. And in every reconstruction process there are many opportunities for error.
'Empty the head'
In a crime situation memory is influenced by many factors such as stress, the presence of a weapon and even just the desire to help police solve the crime.
"Police know how fallible the memory can be," says Steve Retford, a former head of the investigative skills unit at GMP and now specialist interviewing adviser with the force.
"They also know this is usually not through mischievousness on the part of the witnesses, but through stress and shock."
Take the case of Jean Charles de Menezes, shot at Stockwell Tube station in 2005 by police who mistook him for a suicide bomber. Eyewitnesses said he had vaulted a ticket barrier when running away from the police. In fact it was later shown by CCTV that Mr Menezes had walked through the barriers, having picked up a free newspaper, and only ran when he saw his train arriving.

Memories are reconstructed unconsciously

In some cases erroneous eyewitness testimony has led to false imprisonment. In the 1970s, the overturning of several eyewitnesses cases resulted in the Devlin Committee's investigation of identification evidence. It found many witnesses overstated their ability to single out the right person.
What this latest research has proved is the extent of how fallible the memory can be, which is "massively important", say those involved.
"That the memory is vulnerable is not new," says Dr Pike. "But it is so important that we know how fallible it is and in what ways. By understanding this better we can design police techniques that make the most of memory."
For the police, the findings of the project are essential because eyewitnesses are still at the heart of most investigations - even with the growth of CCTV.
"Eyewitnesses are our lifeblood and without them you are usually stuffed," says Mr Retford. "Creating the right environment and using the correct psychological tools to get accurate evidence is vital. You have to be on top of your game and really empty the head of all the detail you need."