Last year, MIT neuroscientists Xu Liu and Steve Ramirez manipulated the memory of a mouse. In a fascinating and mildly troubling breakthrough caused by a laser and the protein channelrhodopsin, they “activated” fear memories in a mouse. The impetus, says Ramirez, was the awful feeling of a break-up, the desire, Eternal Sunshine-style, to erase the bad associations with his ex. Says Ramirez, “I realized, maybe that’s a little bit lofty for now. So what if we could start off by going into the brain of a mouse and just find a single memory to begin with? Could we jump-start that memory back to life, maybe even play with the contents of that memory?”
Well, as psychologist and criminologist Elizabeth Loftus knows, humans already have extremely faulty memories; so what happens when the technology for manipulation on the neural level becomes possible? Could there be a future therapy for people with depression or PTSD — or heartbreak? What might that mean for the courtroom of the future? And do people really want to have their memories manipulated, or do our memories make us who we are?
To tackle these questions, TED asked Loftus, Ramirez, and Liu to join in a phone conversation about the future of memory manipulation. Below is an edited transcript of the conversation.
A question for the group: Why is human memory so malleable?
Elizabeth Loftus: Whatever your theory is, why would Darwin or God or whoever have made us with memories that are so malleable? That must serve some sort of function, and one of the functions that malleable memories can serve is to be able to correct errors that creep in. You can update your memory with accurate information, and that is certainly a benefit.
It’s also thought that sometimes we might be better off with distorted memories. There are spontaneous memory distortions — we remember we gave more to charity than we really did, or we got better grades than we really did, or we voted in elections that we didn’t really vote in. These allow us to feel better about ourselves, and maybe live a little bit of a happier life.
Xu Liu: I think there are probably some protective mechanisms in the brain. Sometimes when very traumatic events happen, people say they kind of black out. Probably that’s one of the ways the brain is trying to protect us from those really traumatic memories by blocking them or sometimes even changing them.
EL: Xu, you’re getting perilously close to a folklore I’ve been fighting for a while. The idea that we repress horrific brutalization and are completely unaware of it is an idea that’s been floating around for a long time. But I have not found any credible scientific support for the idea that you can, for example, be raped for ten years and have absolutely no awareness that this happened, and go into therapy and reliably recover it later. So while I do think that sometimes unpleasant and awful memories can be distorted or changed in a way that allows us to feel a little better, I don’t know about the blockage of repeated brutalization.
Steve Ramirez: I was actually wondering, Elizabeth, if you happened to know: Is it easier to induce false memories in people with depression, whether through suggestibility or hypnosis? Or are they more immune to them?
EL: I’d love to know the answer to that, but we have not knowingly run people with depression through our memory distortion experiments. Recently, we published a paper where we took individuals with extraordinarily good autobiographical memories — they can almost remember every day of their adult life — and we found that even this special group of people was susceptible to memory contamination. We thought if any group were going to be immune, it would be them. They weren’t. It raises the question: Is anybody?
XL: Memory is also connected to the way we imagine the future. We tend to have a more positive bias in our memory, but somehow, studies show, the memories and imaginations of people with depression are probably more realistic than “normal” people.
EL: Yes, I’ve seen some recent articles on the fact that many of the same brain structures are involved when we imagine a future event as when we try to recall a past event — so people are hypothesizing that there’s a connection between those two activities. Maybe this malleable memory system allows us the flexibility to anticipate alternative futures that we can contemplate and then grapple with in advance.
SR: Right. If I remember correctly, there are a handful of studies implicating the same structures involved in forming memories and projecting yourself into the future. People with certain kinds of damage to the memory-forming areas of the brain show profound deficits in being able to project themselves into the future. That’s a really neat interplay.
EL: So where are you going next with the mice?
SR: It’s all over the place. With our first couple of papers, we wanted to provide a tool that could activate discrete memories, or could even tinker with them a little bit to play with the contents of memories. But this team of just Xu and me has blossomed into more than a dozen people asking whether we can do those same kinds of manipulations with recent memories and old memories. And rather than sticking with fear memories, could we also do this with pleasurable memories? So that’s one of the next steps.
One of the reasons I asked you about depression patients is because I’m working on a project now to try to reactivate positive memories in animals with depression. This is to see what the interplay is between a brain that’s in a depressed state and a positive memory that might potentially rescue that depressed state.
EL: That’s fascinating. This may only be tangentially relevant, but we have asked whether it’s easier to distort somebody’s memory in a negative versus a positive direction. We finally have some good evidence that people are more likely to adopt a false memory when it’s a positive memory — and more likely to do so if it also paints you in a more positive light. I don’t know if this is true for people with depression, but it might be a terrific treatment for depressed people. If you can succeed in doing that, it would be enormous.
SR: Yeah. Because there are plenty of treatments for depression, and yet their efficacy is low. We have a pretty good idea of what certain drugs are doing to the brain; I think the goal with this new project would be to do more of a cognitive intervention. We know sometimes they work really well in people. Sometimes they even outperform drug treatments.
But as to how they work mechanistically, the brain has still been a black box with that kind of treatment, so providing an analogous or roughly parallel animal model would be great, because we can actually go in and causally dissect these circuits.
EL: What kinds of criticisms do people throw at you?
XL: One of the first things people ask is, ‘are you sure you are manipulating a memory per se, or are you only manipulating a sensory representation?’ For example, people ask whether you can reactivate the same sensory areas and reactivate a response. That’s a very good question. So far, our answer is that we think we are manipulating memory, but not a sensory representation per se. If it’s purely sensory, then it doesn’t matter whether you activate the same cells a day after the memory or a month after the memory; the response should more or less be the same.
We found that when we activate these cells over a short period of time, we can cause the recall of the memory — but if we reactivate them a month later, they don’t do anything any more. This is consistent with the so-called “systems consolidation” hypothesis: short-term memory involves the hippocampus, but long-term memory is somehow transferred to other areas, like the cortex. Someone else in the lab tried to activate cortical neurons a month later, and indeed we saw that memory gets recalled.
SR: One of the challenges we got in regard to our false memory paper is that we need an operational definition of a false memory. People weren’t necessarily convinced that what we claimed was a false memory truly was one.
The relationship between false memory, like the kind we created by manipulating a mouse, and false memory in humans is at best unclear. We wouldn’t directly try to adopt our animal model as one for human false memories, because it’s slightly different in terms of how these kinds of memories are formed. What we did was to artificially reactivate the memory of a particular environment and then redirected a feeling of aversion to that memory by sending the mouse mild foot shocks, in order to form an artificial association between the two.
So Elizabeth, is it accurate to say that most false memories in humans are not really totally reconstructed? In other words, the false memory doesn’t come out of the blue but recombines elements of previous memories into a memory that’s no longer a bona fide representation of the past? It sounds almost like you’re ungluing elements from certain past experiences and then regluing them onto a particular memory, so that memory becomes updated in a way that is no longer representative of that actual event.
EL: That’s a good description of what often happens with humans when they are developing what we think of as a false memory: They are taking bits and pieces of experience from different times and different places and combining them together to construct what feels to them like a memory.
We too get a question: ‘How do you know that you really have a false memory, as opposed to what we’d call ‘demand characteristics’?’ I don’t know if you have this with mice, but you do get human subjects who want to tell you something because they want to please you, because they figure out what it is you’re looking for and they want to help you.
So we’ve developed a number of ways of deciding — or convincing ourselves — that we’re really dealing with a false memory. One interesting way is to put people into what they think is a completely different experiment at a different time much later and see how they behave. So if we plant a false memory that you got sick eating a particular food and then later on, we put you in another experiment, you’ll think it has a different purpose. Then, if you have a chance to eat that food that made you sick, and you don’t eat it as much, it convinces us that this really is a memory operating.
Another thing psychologists have used to prove to themselves that the subject really believes it’s their memory is, when you put them in a new situation, will they tell someone else that they have this memory? If they do that, when there’s no original experimenter around to please, that also helps convince us.
But I don’t know if there’s anything analogous there for mice. I mean, you obviously can’t ask them.
XL: We have to test that by observing their behavior and by looking into their brains. Something good about working with animals is we can look into the cells and see which areas of the brain are actually activated by this memory. By looking at that, we can more or less tell whether the animal is recalling a traumatic event or not. So that’s one advantage of using animals, but again, that’s just based on observation. We cannot tell exactly what the animals are thinking, unlike with humans.
EL: As we get better at planting false memories and controlling behavior, what are the societal implications? Should we ever be doing this for people? Should we ever ban the use of this mind technology? I don’t really have the answers to that.
SR: By giving TED Talks about memory manipulation, I think the three of us can agree that we are promoting this conversation and this dialogue with everybody, that this is a two-way street of insight that needs to happen so that we can make proper legislation.
When the Human Genome Project was started in the 1980s, everybody thought that this dystopian future and Gattaca and genetically engineered kids would happen in the next couple of decades — and that clearly didn’t come to be. In fact, we started that conversation 20-25 years before the genome was even sequenced, so once it was sequenced, that meant there was a 25-year head start on that conversation. Such that — while not perfect — there was the proper legislation to prevent, for example, certain jobs or certain CEOs from discriminating based on genetic background.
By starting this conversation now about memory manipulation technology and implanting false memories, we can have a head start — on the order of decades — before this kind of technology reaches the realm of humans, so that we’re prepared. My hope would be that if something like this were to reach escape velocity and was somehow applied to humans, it would be in a clinically relevant setting. The same way that doctors don’t give antidepressants to an entire population, they just give them to those patients who are debilitated by depression, maybe this technology would just be made available for the population that’s debilitated by things like PTSD or depression.
EL: That reminds me of memory-dampening drugs, like propranolol. Supposedly, you give propranolol to individuals who have had trauma and it dampens the memory. We did a study not long ago in which we asked people to imagine that they had gone through a trauma, they were beaten up on their way home from work and robbed, then they went to the emergency room and the doctor offered them this drug that would reduce their chances of post-traumatic stress disorder. To my shock, 80 percent of people said they didn’t want the drug.
XL and SR [together]: Oh wow.
EL: Their memories are their identity; they don’t know the side effects; what other memories is it going to take away? We’ve repeated that study several times now, and some people just don’t want to have their memory dampened, even when it’s painful and even when they’re told it will help minimize their risk of getting post-traumatic stress disorder. This has taught me that people cherish their memories even when they’re painful and harmful.
And even when they’re fake!
EL: Even if [the memories] are fake and they believe they’re real, they’re going to want to cling to them. So that’s just going to complicate the policy decision-making in this area.
SR: Interestingly enough, that was raised repeatedly in the comments sections for both of our TED talks. A lot of people were saying that they did learn from the heartache of a break-up, that there are mistakes they would no longer make, and it really threads together and unifies their sense of being emotionally “now.”
EL: I appreciate all the times I’ve had a romantic entanglement and a broken heart, and it’s not too painful to think back on it. But on the other hand I would really like to die of an overdose of morphine. I don’t want the pain and I would frankly rather blot out death. It seems we really are complicated creatures.
This article is published as part of our “Questions Worth Asking” series. This week’s teaser: “Should we redesign humans?” See also a playlist of talks featuring thoughts on this topic from the likes of Juan Enriquez and Anthony Atala — and an admittedly incomplete tour of the history of biomaterials.