Alda was the unwitting guinea pig of Elizabeth Loftus, a UCI psychologist who has been obsessed with the subject of memory and its unreliability since Richard Nixon was sworn in as president. Early on in her research, she would invite people into her lab, show them simulated traffic accidents, feed them false information and leading questions, and find that they subsequently recalled details of the scene differently - a finding that has since been replicated hundreds of times.

More recently, she has come to believe that lab studies may underestimate people's suggestibility because, among other things, real life tends to be more emotionally arousing than simulations of it. So these days she takes her investigations outside the lab. In a study soon to be published, she and colleagues describe how a little misinformation led witnesses of a terrorist attack in Moscow in 1999 to recall seeing wounded animals nearby. Later, they were informed that there had been no animals. But before the debriefing, they even embellished the false memory with make-believe details, in one case testifying to seeing a bleeding cat lying in the dust.

"We can easily distort memories for the details of an event that you did experience," says Loftus. "And we can also go so far as to plant entirely false memories - we call them rich false memories because they are so detailed and so big."

She has persuaded people to adopt false but plausible memories - for instance, that at the age of five or six they had the distressing experience of being lost in a shopping mall - as well as implausible ones: memories of witnessing demonic possession, or an encounter with Bugs Bunny at Disneyland. Bugs Bunny is a Warner Brothers character, and as the Los Angeles Times put it earlier this year, "The wascally Warner Bros. Wabbit would be awwested on sight", at Disney.

Elizabeth Loftus' research has obvious implications for the reliability of eyewitness testimony. And it was as a result of her findings that in 1994 she co-wrote her book, The Myth of Repressed Memory, and took a strong stand in the recovered memory debate of the 90s, for which she was reviled by those who claimed to have uncovered repressed memories of abuse - alien, sexual or otherwise.

The American Psychological Association (APA) now takes the line that most people who were sexually abused as children remember all or part of what happened to them, and that it is rare (though not unheard of) that people forget such emotionally charged events and later recover them. But it states that, "Concerning the issue of a recovered versus a pseudomemory, like many questions in science, the final answer is yet to be known." And the debate simmers on. Several new lines of evidence suggest that the interaction between memory and emotion is more complex than was thought. Powerful emotions, it seems, can both reinforce and weaken real memories. We may be able to actively degrade painful memories. And false memories, once accepted, can themselves elicit strong emotions and thereby mimic real ones.

To try to tease apart these complex relationships, the psychologist Daniel Wright and his colleagues at the University of Sussex have been looking into what it is that makes some people more susceptible to false memories than others. On average, studies show that around a third of those subjected to the "misinformation effect" wholly or partially adopt a false memory, but it seems to depend on both the person and the memory. Alan Alda swallowed the hard-boiled egg story, to the extent that he declined to eat one at the UCI picnic, but he wasn't taken in by Bugs Bunny in Disneyland. In one study published last year, 50% of volunteers were persuaded they had taken a ride in a hot-air balloon when they had not. But when Kathy Pezdek of the Claremont Graduate University, California, tried to make people believe they had received a rectal enema, she met with almost universal resistance.

Amid all this variability, Wright's group did find one significant correla tion - though it was not dramatic: those who were more vulnerable to false memories also tended to suffer more frequent lapses in attention and memory. The trouble is, he says, "People who have been traumatised also tend to score higher on tests of lapses in memory." Their traumatic experiences may contribute to their forgetfulness, but their forgetfulness may lay them open to memory distortion - so true and false become harder to disentangle.

Among the symptoms suffered by victims of post-traumatic stress disorder (PTSD) are chilling flashbacks. But, says Michael Anderson of the University of Oregon, "People who suffer PTSD represent a very small fraction of the people who experience trauma. The great majority of people who experience trauma never develop PTSD and eventually are able to adapt in the face of these events." He argues that they do so by suppressing the memory, and that this suppression gradually erases it.

Two years ago, Anderson's group showed that people who deliberately try to keep a word out of their mind find it harder to recall later than if they had not suppressed it. Counter- intuitively, this form of forgetting seems more likely to occur when people are confronted by reminders of the very memory they want to avoid. Anderson says an extreme example of this might be a child who is forced to live with an abusing care-giver, and must put the memory of abuse to one side in order to interact with that care-giver. "If people continue to work at it, the amount of forgetting grows with repetition and time," he says.

At the annual meeting of the US Society for Neuroscience in New Orleans last month, Anderson's group presented new data on how this "motivated forgetting" might arise in the brain. When people tried to suppress memories for certain words while having their brains scanned in a magnetic resonance imaging machine, not only did the researchers see a dampening of activity in the hippocampus, a structure known to be critical for memory formation, but the frontal cortex was highly active. Since the frontal cortex is important for conscious control, they believe that neurons here may be suppressing the representation of the unwanted word in the hippocampus, and in the process impairing its memory.

However, Anderson admits that his experiments ignore the effect of a memory's emotional intensity on a person's ability to suppress it. And there is plenty of evidence that memory for emotionally charged events can be enhanced - albeit at a cost. Also last month, Bryan Strange of the Wellcome department of imaging neuroscience at University College London and colleagues showed that people were more likely to remember a word if it was emotionally arousing - "murder" or "scream", say - than if it was neutral. And the words most likely to be forgotten were neutral ones presented just before emotionally arousing ones. The effect was more pronounced in women than in men, and both the enhanced memory for the emotional word and the forgettability of the preceding neutral one could be reversed by dosing the volunteers in advance with the drug propranolol.

Propranolol, a commonly prescribed beta-blocker, interferes with the neurochemical pathway thought to be responsible for making emotionally arousing events more memorable - the beta-adrenergic system - and it has already been used experimentally in the treatment of patients with PTSD. In one study, published in October, Guillaume Vaiva of the University of Lille and colleagues offered prop- ranolol to victims of assault or motor accidents shortly after their traumatic experience, and then invited them back for psychological testing two months later. On their return, almost all the patients exhibited some symptoms associated with PTSD, but they were twice as severe among those who had not taken the drug.

The finding that propranolol can be effective at blocking memory when given after an event as well as before is important because, as Loftus explains, "In the real world you can't be there to exert your manipulations right at the time an event is happening, but you can get on the scene later." It has been proposed that propranolol should be offered to victims of rape as a standard measure to prevent them developing PTSD. But could it also be used to erase false memories - for instance, "recovered" memories of alien abduction - that nevertheless elicit all the physiological responses associated with harrowing, real memories?

"If the formation of false memories depends on beta-adrenergic activation, then it would seem very possible that propranolol administration could affect them," says the UCI neuro- biologist Larry Cahill, who has also investigated the effects of the drug in PTSD patients. But Ray Dolan of UCL, a co-author with Bryan Strange of the study on memory for emotional words, points out that not all false memories have a common basis. If they are interpolations into gaps in memory, such as the gap that opened up before the presentation of an emotionally arousing word, or possibly the gap into which Alan Alda inserted a memory of having over-indulged in eggs, then it is conceivable the drug would work. But, says Dolan, "Other classes of false memory, for example, where the memories are fantasies or out-and-out fabrications, would be immune to propranolol."

The idea of doctors having the power to wipe the memory clean sends shivers down many people's spines. False memories could safely be erased, perhaps, assuming there was a reliable way of differentiating them from true ones. Although brain-imaging techniques highlight some differences in patterns of brain activation when a person recalls a true as opposed to a false memory, these are statistical differences only. "We are so far away from being able to use these techniques to reliably classify a single memory as being real or not real," says Loftus, "Yet that is what the courts have to do."

True memories, too, can get out of control and become destructive, leading to PTSD and other anxiety disorders. But they start out as an important self-defence mechanism - teaching you, for instance, that too many hard-boiled eggs are bad for you. Erasing them completely could be dangerous.

In the end, says Loftus, it will come down to personal choice. "What would you rather be in the world, sadder but wiser, all too well remembering the horrors of your past and feeling depressed, or perhaps not remembering them very much and being a little happier?"

Further reading

The Myth of Repressed Memory by Dr Elizabeth Loftus and Katherine Ketcham, 1996 paperback (St Martin's Press, New York). ISBN 0312141238

American Psychological Association website with links to questions and answers about memories of childhood abuse: www.apa.org/pubinfo/

Suppressing unwanted memories by executive control by Michael C Anderson and Collinn Green, Department of Psychology, University of Oregon, 2001 (Nature, 410 [6826], 366-9)