"Life without memory is no life at all," so said Luis Buñuel, the surrealist filmmaker. But do we really want every memory? Joel Barish didn't. Remember the guy played by Jim Carrey in Michel Gondry's 2004 film, Eternal Sunshine of the Spotless Mind? He wanted the memory of Clementine Kruczynski (Kate Winslet) extracted, traces of her wiped clean. But surely this was a film fantasy and is not possible in real life. Or is it?
In 2000 my lab published a paper that jump-started a wave of research and speculation about memory erasure. The studies were done in rats, not people, and were focused on a very specific kind of memory. But the work generated heated discussions in the scientific and lay press about the broad implications of memory erasure in humans, including questions about the ethics of memory manipulation. Bush's panel on bioethics weighed in with a scathing comment, but its views were based on their emotional reaction to the very idea of memory manipulation rather than a careful analysis of the scientific discoveries and their implications.
Within the last couple of years, there have in fact been significant scientific advances. Announcements in the press of successful "memory erasure" have given those who suffer from traumatic memory hope that they may one day be able to lead normal lives again. Each such announcement leads to a flooding of my inbox with requests by people who want various memories erased, ranging from memories of annoying ex-spouses to debilitating memories of rape, torture or horrible accidents. So can human memory really be erased? And, if so, should we do this?
A little background. My lab has studied emotional, especially fear, functions of the brain for the past 25 years. We do this mostly in rats using Pavlovian fear conditioning, a procedure in which a sound is paired with a mild foot shock, and the next time the rat hears the sound it acts scared (it freezes) and has highly aroused physiological responses (blood pressure, heart rate, and respiration increase and stress hormones flow). These are automatic (involuntary) responses elicited by the stimulus and help the body defend against the danger that the stimulus warns about. The sound is able to elicit these responses days, weeks, moths and even years later because an association is learned between the sound and the shock and this is stored in the brain as a memory. The association is formed in a region of the brain called the amygdala which receives sensory inputs from nerve pathways that process the sound and the shock. Neurons there integrate the tone with the shock to form an association between the neutral and the harmful events. When the neutral stimulus occurs later, it activates this association, which in turn activates output connections of the amygdala then control behavioral and physiological responses.
Long ago it was discovered that the formation of a long-lasting memory requires the synthesis of new proteins in the brain. The proteins allow the conversion of a temporary or short-term memory into a persistently stored long-term memory. So well established was this idea that it was considered a dogma in neuroscience. It was thus only natural that we would ask whether protein synthesis in the amygdala is necessary for the creation of long-lasting fear memories. It surprised no one that it was. This is just the beginning of the story.
Part of the memory consolidation dogma was that a memory is consolidated only once. Thus, each time the memory is retrieved, it is the original memory that is activated and remembered. But in spite of this strongly held assumption, other research, largely ignored for decades, suggested that memories become labile when retrieved and have to be restored, or re-consolidated, via protein synthesis in order to remain available for later retrieval. If something is done to them during the time when they are labile and before they are reconsolidated they are subject to disruption -- that is, the memories might be erased, or made permanently inaccessible, during this time.
This takes us to the experiment we did in 2000. Karim Nader, a postdoctoral researcher, carried out this study. He created a fear memory in rats, and allowed enough time for it to be fully consolidated into a long-term persistent memory. He then had the rats retrieve the memory by exposing them to the sound. Right after that, he injected the protein synthesis inhibitor into the amygdala. The rats had short-term memory for a few hours but no long-term memory the next day, or any other day after that.
Four years later, Joel Barish had his memory of Clementine erased by using pictures to elicit memories of her and then having his brain zapped as each memory was retrieved. This is very similar to what we did in the rat studies. But can this be done in people?
Nader's study led to a massive tide of research. Some scientists tried to dismiss it (since it went against the idea [dogma] that a memory is only consolidated once). But the data poured in from around the world. Hundreds and hundreds of studies were performed in many different animals (worms, bugs, rats, mice and even people) showing that memory could be interfered with after retrieval. Not all the studies used protein synthesis inhibitors. Drugs that interfere with the process of protein synthesis indirectly also seem to do the trick. We have a long way to go before we fully understand what is going on mechanistically, but it seems clear that memory, or at least some kinds and aspects of memory, can be altered after they are retrieved (reactivated).
So where does that leave us? Can we erase memories in people? To answer this question we need to make a distinction -- fear memories created through associative conditioning are not the same as conscious recollections of the frightening experience. When you form an association between a neutral stimulus and danger (a street corner where you were hit by a car), later exposure to the neutral stimulus not only automatically elicits behavioral and physiological fear responses, it also retrieves a conscious memory of the experience. In contrast to conditioned fear memories created in the amygdala, conscious memories crucially involved the hippocampus and other areas of the cerebral cortex (the wrinkled out layers of the brain).
Most research showing that fear memory can be disrupted after retrieval in animals has involved the automatic kinds of memory rather than the conscious kind. However, in recent studies of humans, where effects on conscious memory can be easily assessed, the physiological responses to the tone but not the conscious memory of having been conditioned to the tone were blunted. This suggests that even if it is possible to "erase" conditioned fear memories, conscious fear memories may not be altered significantly. This is very promising for therapeutic purposes since it might allow a reduction in the emotional upheaval elicited by trauma-related stimuli without interfering with the conscious memory of the trauma. With emotional arousal to the traumatic memory weakened, it might be easier to engage in successful therapy about the traumatic experience.
Above I just said the emotional memory is weakened. Why didn't I say "erased"? It is very hard to scientifically prove that something doesn't exist. All we know is that memory is weaker after treatments that block reconsolidation. It might be erased from the brain or it might still be in the brain but inaccessible. Some studies suggest that fear memory is erased since it doesn't recover after the treatment but whether erasure occurred is an interpretation.
Memory erasure remains a possible but unproven hypothesis. But the blunting of unwanted emotional responses elicited by stimuli from the past is definitely possible and has broad implications. We are nothing without our memories, but sometimes they also make us less than we could be. The job of the scientist is to push the frontiers of knowledge. Society then has to decide what to do with that information. Although some ethicists argue that memory should not be tampered with, every special date and anniversary, every advertisement, every therapy session, every day in school is an effort to create or modify memory. Tampering with memory is a part of daily life. If we take a more realistic view of just how much we mess with memory, the dampening of memories that produce emotional responses in traumatized individuals might seem less malevolent.
Now the disclaimer. I am not a physician or therapist, and I can't blunt your memory of past emotions. Even if I were qualified, there would still be roadblocks. The drugs that are used in animals have for the most part not been proven safe for use in humans. The one drug that is safe, propranalol, may have only weak effects. My lab published a paper in the magazine Science a few months ago revealing a new approach that doesn't depend on drugs, but this idea in the early stages of exploration. New discoveries will likely take us to the point where it will be possible to blunt some forms of memory. When that happens, physicians and therapists will have to make difficult decisions about how the treatment should be used. In the meantime, everyone involved should welcome the broadest possible debate on this challenging but promising possibility.