Slipping on a pair of dark glasses, Agent Kay raises a mysterious handheld device before a crowd of shocked New Yorkers. Suddenly, the unit emits a brilliant flash of light that vaporizes all memory of a violent attack by space aliens from the minds of panicked earthlings who had just witnessed the horror. That little flashy thing, as Will Smith's character calls it in this scene from the movie Men in Black, is not entirely science fiction; neuroscientists know how to erase memories of the recent past, while leaving well-established memories intact. And new research suggests that even long-term memories could be deleted.

Erasing bad memories could be extremely therapeutic. Many people are haunted by painful experiences that cause lasting psychological problems. Forty-nine percent of rape victims suffer post-traumatic stress disorder (PTSD), as do 17 percent of people who survive serious vehicular accidents and 14 percent of those who unexpectedly lose a family member, according to the Posttraumatic Stress Disorder Alliance. Uncontrollable feelings of fear and horror can overwhelm sufferers. Devastating social and psychiatric complications can result, including depression, alcohol and drug abuse, and suicide. Persistent fatigue, digestive disorders and unexplained chronic pain are also common. Sleep may offer no solace, as the distressing events return in vivid, recurring nightmares.

There is heightened interest in how to treat PTSD in the aftermath of the World Trade Center attacks, the Gulf War, Hurricane Katrina and other traumatic events. Psychiatrists expect the extended Iraq War to produce thousands more soldiers with recollections of terrible events, too. Psychotherapy and sedatives can help relieve symptoms, but the treatments have never been widely effective. The best antidote would be to uproot the cause--erase the horrible imagery. In principle, this is not a fantastic notion. We forget things all the time, and memory loss often follows accidents that involve head injury. Scientists are focusing on this line of therapy as they have come to understand exactly how the brain records as well as forgets events. And some questions are already being raised about whether erasing bad memories can be achieved without also disrupting good or necessary ones.

What to Keep

The shortcut to school crosses the old Dugan place, overgrown with weeds and littered with junk cars. Just as you set foot on the property, Old Man Dugan throws open his screen door and two pit bulls charge, snarling with teeth bared. You run for your life, narrowly escaping. The next morning and thereafter you take the long route instead. Returning years later to that place, your heart still races, even though Dugan is long gone. And in the years since the incident, you have developed a lifelong phobia for dogs.

Sometimes we don't need to be told twice. It takes repetition to learn that six times seven is 42, but a single experience can burn a fear of dogs and Dugan's place into a person's brain. Why? Because from a biological or evolutionary perspective, memory is about the future. There is no survival value in having a cerebral recording system that accurately retains every event and sensory experience. (Anyone struggling to manage his e-mail knows the solution is not a bigger inbox--it is to delete files that will not be needed.) The trick for the brain is to somehow assess our minute-to-minute experiences and pick out instantly which ones should be retained for reference and which should be discarded.

The survival and reproductive value of certain events are immediately apparent, and memories of them are socked away permanently; after the Dugan episode you will never fail to recognize the snarl of an onrushing dog. Any experience that sparks fear or passion, any situation that is truly novel, anything you put in your mouth that tastes foul or delicious--each has a high probability of being retained as an event important for the future.

Knowing how a memory is encoded offers clues to how we could possibly erase it. Memories are not held inside neurons--the brain's cells. Both short- and long-term recollections are set in the connections between neurons called synapses--tiny gaps where the signal-emitting finger of one neuron (an axon) sends a message across to a signal-receiving finger of another neuron (a dendrite). A memory is created when a network of synapses is strengthened--temporarily for a short-term memory and permanently for a long-term one. Over time the network of connections can be strengthened further, weakened or broken.

The challenge of manipulating a memory can become bewildering, however. The dendrite of one neuron can be surrounded by 10 to 100,000 axons, and the human brain contains more than 10 billion neurons.

One way neuroscientists attempt to understand memory networks is to retrieve thin slices of rat brain and artificially keep them alive in a lab dish. They then send electrical pulses into the neurons, which causes certain signals to fire across various synapses. Electrodes pick up the pattern of firings and display them on a computer screen.

The firing is actually a response to a neurotransmitter, a chemical messenger molecule released by an axon that crosses the synapse and binds to a dendrite's protein channels on the other side [see box above]. The channels allow a small flow of ions (charged molecules) to drain some of the voltage from the recipient neuron. When the voltage drops enough--because many synapses around it are firing together--the neuron sends an impulse down its own axon to relay the signal on to the next neuron in the network.

In 1973 Tim Bliss and Terje Lmo of the University of Oslo discovered that if they delivered a brief burst of pulses at the right frequency, around 100 hertz, the magnitude of the synaptic signal grew stronger and remained that way when it was tested minutes later. They named this phenomenon long-term potentiation, or LTP. A greater synaptic signal meant a stronger functional connection had formed between the two neurons--a piece of a memory.

Interestingly, the synapse remains stronger for several hours after a short series of shocks, but then the voltage slowly subsides to its original level. If three consecutive shocks are delivered at about 10-minute intervals, however, the synapse becomes permanently strengthened. As we all know from trying to remember people's names when we first meet them, repetition is necessary to move the data from short-term into long-term storage. Repeating the person's name three times right after you are introduced will not help as much as repeating the name to yourself every 10 minutes. In evolutionary terms, a stimulus encountered repeatedly is more likely to be important.

Making Memories Stick

There is a complication, however. The molecules that establish current flow around synapses are proteins, and all proteins in the body degrade and are replaced constantly over a period of hours or days. To strengthen a neural connection for a lifetime, some other process must take place to bolster the physical structure of a synapse or form additional synapses between the neurons involved.

The transition from temporary to permanent memory is called consolidation. Many experiments have determined that consolidation requires many hours, and it can be enhanced or blocked in various ways. As a young rock climber I used to marvel at how veteran climbers in Yosemite Valley in California could recount in vivid detail every inch of a climb that might have been several thousand feet, relating exactly where to find the hidden holds in sequence and just how to contort the body to make the next move. Later I learned that I could do the same thing. If the next 10 seconds of your life could be your last, you will remember them, even if the interval is followed by another 10 dramatic seconds, and so on, for the hours or days it might take to reach the summit. The heightened state of attention, stress and novelty stimulates the consolidation phase of memory.

Neuroscientists have discovered how this consolidation happens. An epinephrine (a.k.a. adrenaline) rush releases a flood of stress hormones and neurotransmitters that activate the amygdala, the brain region that processes fear and emotion. The amygdala connects to many other regions where different kinds of memories are stored, and it boosts incoming data that have emotional impact. Consolidation, therefore, possibly can be aided by increasing levels of these neurotransmitters or hormones. This idea is the basis for memory-enhancing drugs, such as the illicit use of the attention-deficit medication Ritalin, or the mild and temporarily cognitive-enhancing effects of caffeine or nicotine. Clinical trials are now under way to improve memory consolidation in Alzheimer's patients using nicotine patches and more powerful medications. What if drugs were given to inhibit these neural circuits instead? The persistence of memories could be weakened.

Recent experiments on rats by Volker Korz and Julietta U. Frey of the Leibniz Institute for Neurobiology in Magdeburg, Germany, show exactly that effect. By implanting electrodes into the hippocampus, a critical memory center of the rat brain, they found that synapses there could be strengthened more permanently by subjecting rats to a stressful or cognitively challenging experience (finding their way through a maze). This challenge triggered stress hormones, and long-term potentiation induced by electrical stimulation did not fade as quickly as it would have otherwise. The researchers were later able to block this consolidation by using drugs that interfered with the neurotransmitters and hormones. The Frey group and others have also shown that a permanent increase in synapse strength can be undermined by adding drugs that block the synthesis of synapse proteins--in effect dissolving the memory.

Faster Forgetting

One approach to treating PTSD, then, would be to administer such blocking drugs immediately after a traumatic event. Such treatment would prevent the short-term memories from consolidating. There are drugs currently approved for use in humans that act on the receptors of these proteins, such as propranolol, used for certain cardiac patients. [For more on propranolol's memory-blocking effects, see Can We Cure Fear? by Marc K. Siegel, page 44.]

This approach would offer no help to the many PTSD victims whose horrible memories have already consolidated into permanent memory, however. But other methods hold promise. One possibility would be to speed up a psychological approach known as extinction. Therapists ask a patient to recall a stressful event repeatedly under safe and calm conditions. The repetition seems to inform the brain that this memory is no longer linked to a dangerous situation and therefore can be allowed to fade.

Lab mice that at one time were shocked in their cages while hearing a certain tone and later froze in fear when the tone sounded again eventually forgot the bad experience after the tone was subsequently heard at many intervals with no consequences. Yet in 2002 experiments by Beat Lutz, now at the Johannes Gutenberg University in Mainz, Germany, showed that mice genetically engineered to lack receptors in the brain for cannabinoids--molecules that resemble the active ingredient in marijuana--are not able to forget as rapidly. The theory is that the brain's own marijuana calms the neural circuits involved in fear, allowing mice to relax more quickly when they learn that no electric shock follows the tone. If there were ways to boost the cannabinoids only in the amygdala, the brain's fear center, this increase might help people with PTSD put their bad memories behind them more quickly. I should note that this targeted approach is not possible simply by smoking marijuana.

Understanding sleep may provide another avenue for erasing bad memories. A growing body of evidence suggests that memory consolidation continues off-line while we sleep, in part because sleep involves periodic surges of some of the same hormones and neurotransmitters that are aroused in stressful and novel situations. In 2001 Kenway Louie, now at New York University, and Matthew A. Wilson of the Massachusetts Institute of Technology found in rats' hippocampi--vital for learning facts--that the same firing patterns recorded while the rodents were learning recurred during REM sleep.

In 2004 positron-emission tomography studies performed in humans by Philippe Peigneux of the University of Lige in Belgium showed that hippocampal areas activated while subjects learned their way around a virtual town were reactivated during subsequent non-REM sleep. Moreover, greater brain activity during sleep correlated with improved scores on tests of the subjects' route-finding ability the next day. This work illustrates that memory consolidation requires sorting through fresh memories, integrating them with other memories, and shuttling them to different brain regions for permanent storage. Short-term memories deemed dispensable are discarded.

Genes Convert Short-Term to Long-Term

How might all these observations help us devise a strategy for erasing memories? There may be a common thread: genes. In 2004 Chiara Cirelli and her colleagues at the University of WisconsinMadison found increased activity in about 100 genes during sleep. Some of the genes are the same ones that are turned on when short-term memories are converted into long-term ones. The study by Lutz on rats and cannabinoids also confirms that different individuals have different levels of genetic predisposition to fear and PTSD. (One implication may be that some of the alcohol and drug abuse of such people is perhaps an attempt at self-medication, albeit with inappropriate chemicals.)

Scientists have known since the 1960s that turning on genes was somehow involved in making memories permanent, because genes tell cells to produce proteins, and new proteins must be synthesized in neural networks within minutes of an experience for it to be coded as a memory. Using a mild electric shock as punishment, Bernard Agranoff of the University of Michigan at Ann Arbor in the mid-1960s trained goldfish to swim to one side of a tank when a light switched on. If Agranoff first injected a drug into a fish that blocked protein synthesis, the fish learned the task just as fast, but when tested three days later, it behaved as if it had never encountered the situation before. The fish could once again relearn the task just as rapidly as other fish, but the short-term memory was never converted into a long-term one.

Many other scientists have achieved similar results using a variety of animals and learning situations. But only recently have we pinned down just how genes direct memory formation. For a protein to be produced, a stretch of DNA inside a neuron cell's nucleus must be transcribed into a portable form called messenger RNA (mRNA). It travels out into the cell body, where its encoded information is translated into a protein. Experiments show that blocking the transcription of DNA into mRNA or blocking its translation to a protein allows short-term memory to stick but impedes long-term retention.

But why do genes turn on to transcribe DNA to begin with? Neuroscientists have found that when two neurons fire together strongly and repeatedly, calcium enters their nuclei and tells the genes to transcribe mRNA. New proteins are made from the mRNA that cement the short-term synapse connections into long-term memory networks.

No Fear

If we could find a way to block the synthesis of new proteins from genes immediately after a traumatic event, memory would not be consolidated and the horrible visions should fade away. Shock treatment, more formally known as electroconvulsive shock, seems to have such an effect in lab animals. But as with protein-synthesis inhibitors, the shock must be delivered at just the right time. And like the neuralyzer in Men in Black, the shock does not remove well-established memories (those about 24 hours old), because they are already consolidated.

Nevertheless, perhaps a similar mechanism could erase an experience already ingrained in long-term memory. Clues can be traced back to a 1968 study published in Science by James R. Misanin and others at Rutgers University. They found that a consolidated memory could be erased if a lab rat was shocked right after being forced to recall the experience. Shock treatment could make you forget your fear of dogs if you received it right after you revisited the old Dugan place. In the rat, recalling the memory had somehow made it vulnerable to disruption. This phenomenon has been termed reconsolidation.

Research over subsequent years expanded this result to other types of memories held in various parts of the brain in different animals and in response to a wide range of drugs. These include gene and protein-synthesis inhibitors as well as neurotransmitters and chemicals that affect them, called neuromodulators. Recent research by Joseph E. LeDoux and his colleagues at New York University has shown that microinjecting protein-synthesis inhibitors into a rat's amygdala can block memory reconsolidation if applied shortly after the memory is recalled.

A few neuroscientists are beginning to extrapolate this animal work to humans, who perhaps someday could take medication during an attack of PTSD to disrupt reconsolidation and thus undermine the bad memory. In 1994 James L. McGaugh, Larry Cahill and their co-workers at the University of California, Irvine, showed that propranolol, which blocks beta-adrenergic receptors, will make people forget an emotionally stressful story more quickly, even though their memory of a pleasant story is not affected. Small-scale experiments with this drug in 2002 by Roger K. Pitman of Harvard Medical School were performed on people immediately after a traumatic experience in the hope that reducing the stress response would attenuate consolidation. After three months, the patients who were treated with propranolol were less likely to experience post-traumatic stress disorder than those who were not. The next step will be to see if the same drug could help people who already have PTSD, by asking them to take it when they experience an attack.

Erasing memory during the consolidation phase is science fact, not fiction, as anyone who has suffered amnesia can attest. I was left with a 90-minute gap in my memory after a fall from my bicycle. Although I never lost consciousness, my abnormal brain activity after hitting the pavement interrupted the normal consolidation of all short-term memories into permanent ones. Erasing all memory from a certain time interval could be troublesome, but it is likely that emotional memories could be selectively dissolved because they are coded into memory by a special mechanism. And their terrible, vivid recurrence in PTSD makes them uniquely vulnerable to treatment.

Of course, there could be a dark side to memory eradication. Some people fear that successful techniques could lead to a new kind of mind control. Brainwashing, a literal translation from the Chinese xi nao, changes a person's convictions through intensive indoctrination and propaganda, often using isolation, humiliation or torture. But never before has there been a way to truly wash away memory. If scientists succeed in developing a way to cut victims free from haunting memories, they will have forged a double-edged sword. There are memories we would rather not have and memories others might prefer us to lose. Memory-suppressing drugs could conceivably be used in espionage and warfare to free covert agents and their governments from incrimination or by criminals on their victims to eliminate witnesses. Considering the crude brainwashing methods practiced today by cults, political tyrants, criminals and armies, tomorrow's treatment to relieve suffering would, unfortunately, also appeal to the evil aspects of human nature.

These possibilities may or may not come readily. Memory reconsolidation is a well-established phenomenon, but some scientists argue that older memories develop deeper and broader roots as they become connected with other experiences and memories, and they may be hard to purge. Others think that memories must be renewed when they are recalled to integrate them into the brain's subsequent experience and are therefore transient and open to disruption.

Either way, complications could be a problem. For example, because memories are the physical remodeling of synapses, it is possible that each new memory might impact synapses encoding an old memory. Erasing a painful image from the past could alter other images the brain needs for the present. In Men in Black, Will Smith's character warns his partner not to flash the neuralyzer again, saying, You ain't gonna be happy till you fricassee somebody's brain. It is doubtful that any memory-erasing technique would cause brain damage, but in the bargain for a second chance to choose our memories, are we lifting the lid on Pandora's box?