Can I Get a Witness

Seeing Is Believing?

People with limited working memory tend to see more statistical correlations than exist so they jump to the wrong conclusions.

- author unknown

I read an interesting article in a recent Nature about the reliability of eyewitness accounts. The gist of the article was that often people see what they expect to see. (You have to be a scientist to figure this out?) Scientists call this tendency the “base-rate error.” Consider the following:

A cab was involved in a hit and run accident at night. Two companies, Yellow Cab and City Cab, operate in the city. You are given the following data:

(a) 85% of the cabs in the city are Yellow cabs and 15% are City cabs.

(b) A witness identified the cab as a City cab. The witness is 80% reliable.

What is the probability that the cab involved in the accident was a City cab rather than a Yellow cab? Almost certainly less than the average jury member thinks.

The witness is wrong 20% of the time. Thus, it is more common for the witness to see a Yellow cab and mistakenly call it a City cab (.85 x .20 = 17%) than for the witness to see a City cab and label it correctly (.15 x .80 = 12%).

If the witness reports seeing a City cab, the probability that the cab actually WAS a City cab is .12/(.17 + .12) = 41%.

Most jurors ignore base rates (how many Yellow and City cabs are actually operating in the area) when deciding for themselves if a generally reliable witness is correct in what he says.

This is why it pays to get a good lawyer — preferably one familiar with statistics.)

Don't Rain on My Parade

The value of identity parades is questionable. For a start, the number of people in a normal line-up means that the chance of picking the suspect by chance is a lot higher than a statistician would consider safe for any scientific test. Then there is the question of whom it is meaningful to put in a parade besides the suspect. But most dangerous of all is the risk that a witness may be “led” into a choice by remarks, whether chance or deliberate, by somebody involved in conducting a parade.

This last risk, at least, is well recognised. But there is a widespread assumption that, an identification having been made from a parade, subsequent banter can do little harm. Not so. According to Gary Wells and his colleagues at Iowa State University, even remarks made two days after an identity parade might compromise a witness's eventual reliability in court.

Dr Wells's research, just published in the Journal of Experimental Psychology, took more than 250 subjects and showed them a video of what appeared to be a terrorist planting a bomb. At one point, the camera focused in so that the man's face was clearly visible. The subjects were then shown a photographic line-up of six people who looked similar to the “bomber”. In fact, none of these photos actually was of the bomber. Nevertheless, with a bit of encouragement to make some sort of choice, every one of the subjects identified one of the pictures as being of the bomber — itself an observation that raises questions.

Having made a choice, the subjects were assigned at random to one of a number of groups, each of which was treated differently. Dr Wells wanted to find out both the effects of feedback, and how they changed depending on when the feedback was given. The three sorts of feedback were: none at all; a confirmation that the subject had picked the correct individual; or an observation that the wrong individual had been picked. This feedback might be given either immediately, or at a second session which took place 48 hours later. The subjects were also asked (again, either immediately, or after 48 hours) a series of questions about how confident they were in their choice, how good a view they had had of the incident, how much attention they had paid, and so on.

Dr Wells expected, from previous results, that those who had immediate confirmation that they had picked the suspect would be more confident about their decision than those who had not. He was not disappointed in this expectation. The new and worrying result was that exactly the same thing happened even when the first “confirmation” that a subject had chosen correctly happened 48 hours after the identity parade. That, if it is confirmed by subsequent research, has great implications for the reliability of witnesses in trials, since it is easy to design codes of conduct that prevent immediate feedback, but harder to devise ways of stopping details slipping out later.

This matters because research has shown that, in America at least, it is not only the fact that a witness picked out a suspect in an identity parade that matters in court. The confidence with which the witness defends the choice under cross-examination sways verdicts. Identity parades have long looked unsatisfactory as material evidence. This result is another reason to question them.

Source: The Economist print edition 3 Apr 2003

I Re-Memory You

I read in the excellent book by Daniel Schacter, Searching for Memory, that, though memory is pervasive and influential, it is vulnerable to inaccuracy and distortion. (This is an important reason why contracts and secure public records are crucial.)

We reconstruct memory from fragments, each of which may be influenced by the cues available at the time of retrieval. Out of this, a new entity emerges: the recollective experience of the rememberer.

Just because you think you remember something accurately, that doesn’t prove that you do. You really mean you haven’t changed it much from the last time you remembered it. I know that from rereading a hard copy of my journal every couple of years. I wouldn’t dare read it on the computer now — I’d be too tempted to change it to match the way I currently remember events.) I can see where my creeping mental corrections fill in low spots and sand away edges.

Is it okay to edit the past? Or is thinking you really can nothing more than a delusion believed because it facilitates your being kind to yourself (like having a brandy in front of the fire)?

An important cause of memory inaccuracies is the proneness to source amnesia — knowing that we’ve been exposed to some bit of information, but not remembering the degree of accuracy we assessed at the time, in part because we usually assign a general veracity level to the source. Furthermore, imagined visual events are generated by some of the same brain systems that underlie and support visual systems.

The truth is that we accumulate inferences, generalisations and abstractions, not true “memories.” We don’t literally retain individual events that lead to and support our general knowledge.

Our mental records are constantly being condensed. And the things we think of most often are the memories that get condensed the most. Those condensations become concepts. (Of this we are formed.)

Memories Are Made of This

New York - From a lifetime of experience, how do specific moments make their way into the mind's album of memories? The answer, according to researchers, may lie in brain receptors that work to keep the past alive. These receptors, called NMDA (for N-methyl-D-aspartate), appear to play a key role in turning those short-term recollections into long-term memories - a process known as memory consolidation - according to results of studies in mice. The NMDA receptor is required for certain types of learning, but its role in memory consolidation - which takes days to weeks to develop - has not been studied in detail before now, according to Eiji Shimizu and associates from Princeton University in New Jersey.

To sort out the role of NMDA in memory formation, the investigators used genetic techniques to breed rats that contained NMDA receptors that could be turned on or off depending on what the rats were fed. In a variety of tests, learning was impaired in rats while their NMDA receptors were turned off, though they learned just as well as normal rats when their receptors were turned on, the authors report. Rats that engaged in learning tasks while their NMDA receptors were turned on later remembered those tasks if their receptors were kept on, the researchers note, but if their receptors were turned off after the task was learned, the rats forgot what they had learned.

Complex memories - remembering how to react in a particular situation, for example - were especially impaired by turning off the NMDA receptors, according to the report in the November 10th issue of Science. Based on their results, Shimizu and colleagues conclude that several rounds of memory reinforcement - each round requiring the NMDA receptor - are needed to be certain that what is learned is retained as long-term memory.

Taken from Science 2000;290:1170 - 1174 Headlines

Sources: HealthSCOUT/AP Reuters Friday 10 November 2000

Some Memories May Rest on Good Sleep, Studies Suggest

by Bruce Bower

When practicing a gymnastics move, a musical piece or any other activity that depends on effortless, virtually automatic execution, here's some memory-enhancing advice: If you snooze, you cruise. That, at least, is the implication of two new studies in which people who practiced a task that demands quick visual processing performed it better on ensuing trials if they were first allowed to get some sleep.

Moreover, one investigation suggests that the initial night of sleep after learning so-called procedural skills proves crucial for memory. The other findings indicate that sleep early in the night, which includes mainly slow-wave electrical activity in the brain, aids procedural recall. Rapid-eye-movement (REM) sleep later in the night strengthens only memories already bolstered by slow-wave sleep, the researchers report.

Both studies appear in the December issue of Nature Neuroscience.

"It's becoming increasingly clear that sleep is critical for consolidating procedural memory into a usable form," says neuroscientist Robert Stickgold of Harvard Medical School in Boston. Stickgold and his colleagues conducted one of the new studies. In an afternoon session, they trained 133 volunteers to perform a visual-discrimination task. Participants viewed, for a fraction of a second, an image including a small set of diagonal bars against a background of horizontal bars. The screen then went blank for a variable period, followed by a distracting pattern for a fraction of a second. Then, the participants reported the orientation of the diagonal bars.

The researchers determined the minimum amount of blank-screen time needed by each volunteer to overcome the distraction and discern the orientation. Participants who slept on the night after training performed the task better - just as accurately, using less blank-screen time - the following day. They showed further gains over the next three days. In contrast, people deprived of sleep on the night after training didn't perform the task better the next day. Even when allowed to catch up on their sleep over the next two nights, they exhibited little improvement.

"These results suggest that the first night of sleep is crucial for acquisition of the new visual skill," says neuroscientist Pierre Maquet of University College London in a comment published in the same journal. Maquet has linked brain changes during REM sleep to procedural learning.

The slow-wave portion of sleep may nevertheless prove essential for procedural-memory formation, according to the second study, directed by Steffen Gais of the Medical University of Lubeck in Germany. Gais and his co-workers trained 21 volunteers on a visual task much like that used by Stickgold. Performance improved markedly for those who trained for one hour in the afternoon and then were tested after sleeping for the first three hours of the night, during which the brain shows mainly slow waves. No improvement occurred for those who trained after awakening from the first three hours of sleep and were tested after sleeping through the rest of the night. Volunteers who slept through the entire night after training exhibited a much larger performance boost than those tested after the first three hours of sleep. These findings suggest that slow-wave sleep triggers procedural memory formation, whereas REM sleep amplifies that process, Maquet remarks.

However, neither of the new studies establishes sleep as essential to memory consolidation, says Robert Vertes of Florida Atlantic University in Boca Raton. Stickgold's data also show memory improvement with the mere passage of time, Vertes points out. Researchers know even less about the effects of sleep deprivation on memories that require conscious effort, Stickgold notes. For instance, staying up all night to cram for a test may work for some students, although they probably forget what they have learned in a few days, he says.

Source: NandoTimes 8 December 2000 © Nando Media and Los Angeles Times Syndicate

Enzyme May Help Brain Clean the Slates

by Cart T Hall

A molecular "eraser" has been found in the brain, scientists report today, suggesting that without it our memory tracks would function about as well as a computer without a delete key. Reporting in the journal Nature, scientists describe the intricate workings of an enzyme of forgetting called PP1, shorthand for protein phosphatase 1. And if you find that name hardly worth recording and soon forget it, it might be because you have a healthy amount of the stuff.

A series of experiments on genetically altered laboratory mice suggests that the newfound mental eraser is part of a healthy tension between learning and unlearning that keeps synaptic circuits from becoming saturated early in life with needless information. The same memory constraint, perhaps running in overdrive, may play a role in some forms of age-related memory decline, suggesting a potential target for memory-enhancing therapies if further studies can determine just how PP1 might figure into memory-related syndromes suffered by humans.

Alcino Silva, a neuroscience professor at UCLA who co-wrote a commentary that also appears in Nature, said the experiments moved the scientific understanding of learning a critical step forward, although much of the mystery remains. "A great deal is known about the molecules of learning and memory," he said, "but this is the first piece of evidence I know of about the mechanisms of forgetting."

Conditions such as Alzheimer's disease, strokes and traumatic brain injuries also can wipe out memories, but that occurs typically through the death of millions of brain cells. The neurochemical system described in today's study is much different and far more subtle. Although chronic forgetfulness can be a symptom of disease, it also can be an ordinary part of healthy mental functioning. One thing the study makes clear, Silva said, is that "forgetting is not just chaos. It's an organised process."

Conclusions are tentative because no human experiments have been done. In fact, it's difficult even to conceive of any practical way to test how the activity of particular brain chemicals might explain what goes in our heads when we, say, memorise a new home phone number and forget our old one. But scientists said there is every reason to suspect that rodent neurobiology is not all that different from our own. "There are active processes to get rid of information," said Isabelle Mansuy, a neurobiologist at the Swiss Federal Institute of Technology in Zurich and senior author of today's study.

The findings underscore one of the most fundamental insights arising from a recent flood of brain research: healthy mental function, like that of virtually all body systems, is a matter of keeping opposing forces in a state of healthy tension, a condition known as homeostasis. In this case, the brain seems to be genetically hardwired to perform a complex balancing act between recording information and wiping the slates clean. Although it might seem preferable to minimise the latter business, an inability to forget trivial matters can be a serious disability - as popularised in the 1988 hit film Rain Man, in which Dustin Hoffman depicted an individual dealing with autism and savant syndrome. Dr Darold Treffert, a Wisconsin physician who served as a consultant on the film, said he always suspected that some savants with seeming total recall "may not have so much a fantastic ability to remember, but an inability to forget."

Whether PP1 deficiencies might help explain the problem remains to be seen. But scientists said there is little doubt that a healthy brain needs the ability to perceive what is worth remembering and forget the rest. "There needs to be a balance in the brain between positive processes and negative processes," Mansuy said. "It's like any natural function. Forgetting is part of the homeostasis of the brain."

In the new experiments, Mansuy and her colleagues set out to find the molecular underpinnings behind the fact that memories usually lose fidelity and eventually vanish with time. Another question they sought to answer is why "cramming" is not a terribly effective way to learn, as compared with dividing study sessions into short periods with breaks in between. Earlier evidence had suggested that a gene encoding the PP1 enzyme was involved, acting in combination with some other key molecules. So researchers bred a line of mice that produced a neurochemical inhibitor of PP1, effectively blocking the action of the enzyme. The PP1-inhibited mice and normal control animals were then put through a series of learning tests - finding their way through cloudy water to a submerged platform, for instance - to determine how the active molecule functioned. Results showed that PP1 "determines the efficacy of learning and memory by limiting acquisition and favoring memory decline," the scientists found.

Inhibiting PP1 tended to prolong memory after learning, and the genetically modified mice proved to retain information long into their old age, while normal aged mice soon forgot what they had been taught. The findings "emphasise the physiological importance of PP1 as a suppressor of learning and memory, and as a potential mediator of cognitive decline during aging," the scientists concluded.

E-mail Carl Hall at chall@sfchronicle.com. Carl is the Chronicle science writer

Source: sfgate.com from the San Francisco Chronicle Page A - 6 Thursday 29 August 2002

Mind Makes Memories Fonder - but False

by Dan Vergano

Ah, memories - like the time you saw that entire car accident, or that spill in the grocery store, or Bugs Bunny at Disneyland. Only you didn't.

A novel set of experiments, released Sunday in the Journal of Experimental Psychology, suggests people unconsciously tamper with their own memories, inventing causes for the events they see around them to help make sense of things.

"Memory isn't a record. It's an interpretation, to a large extent," says psychologist Mark Reinitz of the University of Puget Sound in Tacoma, Washington.

In the just-released study, he and Boston University researcher Sharon Hannigan showed slides of scenes from settings such as a grocery store or restaurant to 144 student body participants. The slides depicted events such as dishes spilling as a meal is ordered, without showing the causes. Questioning the students one or two days after they had viewed the scenes, the researchers found (as have earlier studies) that people "fill in the blanks" in their memories, claiming to have seen slides not originally presented to them that fit a scene's narrative, Reinitz says.

Going beyond these findings, the researchers found that people claimed confident memories of viewing unseen slides that explained events in the scenes 68% of the time. For example, to explain a spill seen in the grocery store scene, many claimed to have seen a picture of someone pulling an orange from a pile of fruit.

"Surprisingly often they don't remember the slide as an inference, instead, they think their inference is a real memory," Reinitz says.

"They've taken this further (by) showing that people invent memories of causes," says psychologist Elizabeth Loftus of the University of Washington-Seattle. "It certainly does have practical relevance" to court testimony of eyewitnesses of crimes or accidents, she says. In her own research, Loftus has shown people advertisements depicting Bugs Bunny, a non-Disney character, welcoming them to Disneyland.

In an upcoming paper in Psychology and Marketing, she and Harvard's Kathryn Braun found about 35% claim to have shaken hands with Bugs on a past visit to the Magic Kingdom.

"Memory is malleable for a reason," Loftus says. "It helps us to remember ourselves in a more positive light."

Source: USA Today Monday 2 July 2001

Personally, I see dreams in much the same way. I don't feel dreams are coherent in any way. I see them as an emotional "storm" our brain's housekeeping causes. However, on awakening, we try to make sense of the way we are left feeling. For the brief time we can "remember" the feelings, we try to explain the impressions to ourselves or to a companion. Only then - after we awaken - do we construct a "story" to explain the emotional upheaval we just experienced. Thereafter we are convinced that what we said is what we actually "dreamed."

I think, for most people, waking life is a lot more "dream-like" than they would ever imagine - we are constantly concocting reasons for why our emotions leave us feeling the way we do.

For articles on bacteria, centrioles, chairs, nebulae, asteroids, robots, memory, chirality, pain, fractals, DNA, geology, strange facts, extra dimensions, spare parts, discoveries, ageing and more click the "Up" button below to take you to the Index for this Science section.