It’s the premise of every third sci-fi thriller. Man wakes up to his normal seeming life, but of course it isn’t. At first, just the little things are off – the dog won’t eat and the TV keeps looping some strange video – but whatever. A few cuts later, with only his granddad’s rusty brass knuckles and a steely-eyed contempt for authority, our hero reveals a conspiracy that kicks up straight to the top. There were deals. Some blackmailing. A probe or two. But in the end, what’s most important is that everything he thought he knew was wrong. Because the scientists (Noooo!!) got to him with one of those electrode caps and rewrote his memory. Everything – the job, the daughter, the free parking – is a lie.
The dramatic ploy works on us because memory seems inviolable, or at least, we desperately hope that it is. We allow that our memories may fade and fail a bit, but otherwise, we go on the sanity-preserving assumption that there is one reason why we remember a particular thing: because we were there, and it actually happened.
Now, a new set of experiments, led by MIT neuroscientists Steve Ramirez and Xu Liu in Susumu Tonegawa’s lab, shows that this needn’t be the case. Using a stunning set of molecular neuroscience techniques (no electrode caps involved), these scientists have captured specific memories in mice, altered them, and shown that the mice behave in accord with these new, false, implanted memories. The era of memory engineering is upon us, and naturally, there are big implications for basic science and, perhaps someday, human health and society.
Although the techniques these investigators used to manipulate memory involved a jaw-dropping sampling from modern neuroscience’s bag of tricks, the essential strategy is easy to understand. Basically, you need a way of labeling neurons that were active during a specific experience, and a switch to operate them.
Enter designer mice. The mice in Ramirez and Liu’s experiments had been genetically modified so that when their neurons were highly active (and therefore presumably encoding ongoing experiences) those same neurons would produce a molecular label, as well as a molecular ‘ON’ switch. The label caused the neurons to glow red, and the switch was the now famous, and likely-future-Nobel-landing molecule Channelrhodopsin, which renders neurons light-activated. In these mice, then, the scientists could quite literally see recent experiences that had been written to specific brain cells. Even more impressively, they could activate those very same neurons in behaving mice by shining light on them, re-awakening whatever fragments of experience those cells had presumably encoded.
A final, and key feature of Ramirez et al’s labeling system was that it could itself be switched on or off, under control of the common antibiotic doxycycline. If doxycyline was given to the mice in their diet, the labeling process was snuffed out. If doxycycline was removed though, labeling was unimpeded. This was critical for labeling memories formed only during specific, experimenter-defined time windows.
In their main experiment, the researchers removed doxycycline for a short spell as mice explored a novel arena, allowing the neurons representing that arena –especially those neurons in a brain area called the hippocampus – to become labeled and light-activatable. The mice were then given doxycycline again to stop the labeling process. In this way, the experimenters had given themselves a literal biological handle on something that seems hopelessly subjective: a mental representation of a particular experience, at a particular time.
Impressive as it is to label a mental representation like this, this was still only the first half of Ramirez and Liu’s experiment. What they were really interested in is memory engineering – whether this specific representation could be brought on line again artificially and modified. In other words, could they make mice recall something that had in fact never occurred?
Their hunch was that the specific but otherwise unremarkable memory of the arena could be re-tooled and loaded with novel emotional content. To do this, the experimenters moved the mice to a new setting, shone light onto the rodents’ brains to re-awaken the memory of the previous arena, and paired this with a series of electrical shocks to induce fear.
Later, when the mice were returned to the original arena, they showed a much higher rate of behavioral freezing – the innate, ‘paralyzed with fear’ reaction that most mammals show in response to frightening situations. In other words, their brains had been tricked into thinking that the arena was a frightening place associated with shocks, even though no shocks had ever been given there. To rule out the more mundane possibility that the mice were just generally more fearful after the shocks, the researchers showed that fear reactions were specific to the original arena, and absent when mice were tested in new arenas.
As their final act of memory manhandling, the experimenters asked whether they could not only lay down a new false memory, but also if these memories would guide overt behaviors. This involved a slight variation on the previous experiment. Mice were placed in a two-room arena, and the mental representation of one of these rooms was selectively labeled using the same technique as above – that is, restricting doxycycline when present in one of the rooms, but not the other. As before, the memory for this room was awakened in a new context by light stimulation, and simultaneously paired with an electrical shock. Finally, the mice were placed back in the two-room arena, and allowed to spontaneously explore it. The mice showed a strong preference for the non fear-conditioned room, indicating that falsely implanted memories can bias ongoing behaviors. These memories are folded into a mouse’s way of looking at the world.
Naturally, one wonders whether these techniques might someday find human applications. Perhaps it would be possible to rebuild particularly cherished and important memories that have deteriorated with age or disease? Or perhaps, more provocatively, some might even embrace the idea of falsified memory – artificially adding in happiness where there is only remembered pain, or subtracting out enduring despair that’s long outlived its usefulness. These are some ethically tricky situations, to be sure. At the same time, though, it’s hard to not sympathize with someone, say a war veteran or a rape victim, who might want the emotional content of a specific, life-destroying memory modified.
For now, these are all hypotheticals, given that memory engineering at the cellular level is currently only possible in highly engineered strains of mice who are tricked out with molecular beacons and switches, and whose memories are established in highly stylized ways. Nevertheless, these experiments should certainly give us pause. This is still a very long way off, but we’ve gotten a glimpse of how you might eventually modify the brain to modify more complex, narrative-like memories. It probably won’t be with electrode caps.
Are you a scientist who specializes in neuroscience, cognitive science, or psychology? And have you read a recent peer-reviewed paper that you would like to write about? Please send suggestions to Mind Matters editor Gareth Cook, a Pulitzer prize-winning journalist and regular contributor to NewYorker.com. He can be reached at garethideas AT gmail.com or Twitter @garethideas.