This story by Kate Baggaley originally appeared in Van Winkles, the publication devoted to sleep.

At one point in Marcel Proust’s “In Search of Lost Time,” the narrator breaks with habit to take some tea and a madeleine cookie. The moment he tastes the “warm liquid mixed with the crumbs,” he’s overcome with joy. The flavor of tea and cake summons memories of his youth, a visceral link to the morning ritual he shared with his aunt.

As Proust’s narrator realized, our senses are powerful keys to memory. It turns out this is true in sleep as in wake. While we can’t actually learn new information as we doze, researchers are hard at work on a process that commandeers our senses to help the snoozing brain strengthen or throw out what it’s already taken in. The technique, called Targeted Memory Reactivation (TMR), uses noises and odors to amplify the memory-strengthening processes that happen during sleep. And if it proves as effective as scientists believe, we may be able to use it to enhance our ability to absorb new learning, loosen the grip of traumatic memories or even alter how people think.

“The potential applications are huge,” says Susanne Diekelmann, a researcher at the University of Tübingen in Germany who has been probing TMR for several years.

In short, it could allow us to explore and seize control of the ties between sleep and memory. “Memory is important for identity and quality life, so if TMR can help memory, it may help more broadly,” says James Antony, a top neuroscientist at Princeton University. “Because in a lot of ways what we’re doing is…trying to collect memories.”


Every day we’re bombarded with information — ads, conversations, tweets, snaps, scenery, Slacks. By the time we slip under the covers, we have amassed a string of fragile, newly minted memories, which are built from connections between neurons. Many of these memories, like what we eat for lunch each day, fade with time. Others are strengthened and converted into long-term memories. This process, called memory consolidation, appears to occur chiefly during slow-wave sleep. The underlying process is called neural replay. Learning a new fact or face (called memory-encoding) corresponds to a distinct pattern of neural firing. Later, during sleep, our slumbering brains replay that same neural firing pattern from earlier. It’s an automatic game of learn-and-repeat which, the theory goes, helps implant memories firmly in the brain.

“It’s like your brain is practicing and rehearsing these memories, says Penny Lewis, a neuroscientist and sleep scientist at Cardiff University in Wales whose work focuses on sleep and memory consolidation.

For each memory, the pattern looks a little different. But these webs overlap. “Every piece of knowledge that you have is related to other things,” adds Antony. So when our sleeping brain replays memories, it doesn’t just strengthen them — it embeds new information you’ve picked up into existing networks of knowledge. Recall, for instance, your last trip to the beach. You might remember waves crashing along the shore, the grit of sand against your toes, the emotions stirred up by your companions. These fragments are stored in different parts of the brain. “It’s likely that when a memory is replayed in that way it’s a full multisensory experience,” Lewis says.

With TMR, we may be able to shape these processes by using scents or noises (initially experienced when we’re awake and learning new information) to spark replays. The right scents or sounds can boost the process of memory consolidation, helping to strengthen — or even alter — memories.


Scientists first got an inkling that sensory cues could influence memory during sleep in the 1980s when they saw greater memory activation in rats who’d been shocked two times — during a learning task and then again during sleep. Then, in 1990, a team in Canada paired a ticking clock sound with a logic task. “The idea was to ‘remind’ the subject to ‘work’ on consolidating the memory of our little task,” even though they probably had more important memories that needed to be consolidated, recalls Carlyle Smith, one of the world’s leading sleep researchers and the director of the Trent University Sleep Research Laboratories in Ontario.

Smith wanted to gather evidence that memory consolidation occurred in sleep, which many scientists doubted at the time. He was onto something, but his work was mostly ignored because it preceded the breakthrough discoveries about replays underlying scientists’ current understanding of sleep and memory.

In 2007, Björn Rasch, a biopsychologist and colleague of Diekelmann at the University of Fribourg in Switzerland, revived TMR using a version of the memory game Concentration. While people learned to associate pictures of cards with certain locations on a computer screen, the scientists pumped in the scent of roses. This odor, wafted into the background, became linked in people’s minds with what they were learning. Those who were re-exposed to the aroma during sleep had an edge in recalling placement of the cards.

Intrigued, the team continued to probe the technique, which they found could make memories more stable, and do so more quickly, than they’d otherwise be. Various other scents, they discovered, worked equally well to cue memories. But they weren’t yet sure how much information could be pinned to a single odor, or if a smell might lose its memory-triggering power over time.

When certain sounds were presented again during slow wave sleep, subjects remembered locations for the associated images more accurately.

It wasn’t long before scientists discovered that sound, too, is an effective trigger. In one experiment, subjects were asked to memorize the grid coordinates of an image, such as a cat or kettle, that had been paired with a relevant sound, such as a meow or whistle. When certain sounds were presented again during slow wave sleep, subjects remembered locations for the associated images more accurately.


Though TMR is in its infancy, researchers have already seen great progress in what it can do. And the potential, right now, appears boundless.

TMR seems to boost motor skills as well as explicit memories. Antony and his colleagues set up an experiment using a game resembling “Guitar Hero.” Participants learned two different melodies, one of which was cued during sleep. Subsequently, they were able to play the sleep melody more accurately than the other one.

And we might be able to use TMR to change memories, too. Recently, Diekelmann and her cohort reported that TMR helped men figure out hidden patterns, transforming their intuition into understanding (women didn’t receive as significant a boost, but most TMR studies reveal no difference across genders).

When Penny Lewis of Cardiff University cued “melodies that people had tapped out earlier, people became both swifter and better at articulating the tune in writing.

And when Batterink and company asked people to unscramble words assembled according to a made-up grammatical system, TMR seemed help them intuit the system’s rules. “These are pretty small effects,” Batterink says. “It’s not like they’re twice as good.”

But still, it’s inspiring. Scientists are even investigating whether TMR can influence how emotional memories are consolidated. Lewis, for instance, is digging into whether or not triggering upsetting memories during sleep can make them less distressing.

“We’ve got some very nice data suggesting that it can,” she says.

Lewis and her colleagues present people disturbing pictures (such as car crashes) paired with a sound cue. After sleep, people seem to find these images less troubling if they were reactivated using TMR. One explanation for this phenomenon might lie in a neurotransmitter called noradrenaline, which is greatly diminished during REM sleep. It helps control our bodies’ physiological fear responses, such as increased heart rate, sweating or pupil dilation.

Lewis is digging into whether or not triggering upsetting memories during sleep can make them less distressing. “We’ve got some very nice data suggesting that it can,” she says.

“The idea is that if you replay these memories at that time, you actually can’t have the emotional response,” Lewis says. “So replaying it without the physiological response might help people to kind of decouple the memory from the emotion.”

It’s not hard to imagine TMR as a treatment for trauma disorders including PTSD. It’s a tantalizing prospect, but for now, even in early trials involving those with upsetting memories, it’s too early to draw any conclusions, he says.

“We need a good database before we can transfer this to patients, because it’s not quite clear what that will do,” Rasch says. “Does that actually help people, to reactivate this traumatic memory during sleep, or does it make things worse?”

Rasch and his colleagues are also investigating how TMR might facilitate learning foreign-language vocabulary.

And while TMR seems to be most helpful at enhancing memories that are still being consolidated, it may prove helpful in helping Alzheimer’s patients remember the way home or recognize family members. Physical therapists could also capitalize on TMR to help people recover motor skills after a stroke or injury.

It’s also well within the realm of possibility that TMR could even influence how people think. Say you’re at a party, and spot your colleague across the room. You wave; they don’t respond. You must judge for yourself whether they’re snubbing you or just didn’t see you. We run across similarly ambiguous situations all the time. Some people, such as those with anxiety or depression, are more likely to put a negative spin on them.


Despite the arsenal of applications for TMR, its influence on memory is subtle, and it’s unclear how long the effects last. And other kinks have to be worked out.

For instance, how do individual replays appear in the human brain?

“We don’t even know what the signal looks like,” Antony says. He and his colleagues are using TMR to illustrate memory replay in people by measuring their subjects’ reactivation of previously viewed images with electroencephalography (EEG).

“Smell is a very powerful trigger because it’s directly connected to this brain region that is related to memory, whereas sounds…have to stop at the relay station first.”

Sound and smell have both been successfully used as cues in TMR experiments, and each domain has its own advantages. Smell, for starters, is no stranger in the world of sleep science. There’s some evidence that herbs such as lavender can help us relax and drift off, and that smell can affect the emotional content of a person’s dreams.

Indeed, the brain’s olfactory system is closely connected to the hippocampus and to emotion-related areas such as the amygdala. Odors also typically don’t wake us up; they’re processed during sleep without disturbing it.

The same is not true of sounds, which are routed through the thalamus. “[Smell] is a very powerful trigger because it’s directly connected to this brain region that is related to memory, whereas sounds…have to stop at the relay station first,” Pereira says.

Precisely calibrating sounds for TMR is tricky, too. Noises can easily disturb sleepers. And yet, “if you play them too quietly you’re not entirely sure that they’re being registered,” Batterink says. “So it can be a bit of a delicate balance between getting the volume just right on a individual basis.”

On the other hand, sound is more successful at cueing motor-skill performance than odor; brain areas involved in auditory processing are strongly connected to motor areas, Diekelmann says.

Compared to smells, sounds are also easier to control at home, meaning scientists could conjure up a nearly unlimited library of them. “If we wanted 60 different odors that are really distinctive from each other, that would be much harder to arrange,” Lewis says.

An odor is a blunt tool that acts on all memories (learned during experimental study sessions) equally. Sounds allow a more specific, fine-grained approach that can pinpoint individual memories.

What’s more, sounds can be played and processed more quickly than odors can. An odor is a blunt tool that acts on all memories (learned during experimental study sessions) equally. Sounds allow a more specific, fine-grained approach that can pinpoint individual memories.

Cues aside, if TMR can break out of the lab, other practical considerations remain at large. For at-home use, scientists would have to develop a gadget to identify a person’s sleep stage and then deliver a cue at the right time. People become habituated to the sound, per co-author Rebecca Spencer, of the University of Massachusetts Amherst.

Smith, who envisions TMR being used for studying or sports, says, “The idea awaits some kind of very simple device or app that can do these things, although I think it is quite possible.”

Rasch and his colleagues are working on developing such a device, which would record a person’s sleep and then automatically present words at the right time and right volume. He expects to have proof of principle in a year or two, at which point he’d find collaborators to help design the gadget.

“That would probably involve electrodes or some kind of brain recording device…much simpler than in the sleep lab,” he says.


Messing with memory is dangerous business and every technology, no matter how well-intended, can be used for nefarious means. When the time comes for TMR to emerge from the lab, the potentially sinister uses must be recognized.

One of the biggest concerns comes from the fact that TMR can influence how people think. And Rasch’s research into how it can shift interpretation biases (how we perceive someone is acting towards us) is only the beginning.

“You can use TMR for good to help people decrease their bias but you can also just as easily…try to convince people that these ideas are better, or these people are better”

Antony is cautious about extrapolating from this work. “It is promising, but there hasn’t been a direct link shown that this is going to cure racism,” he says.

“You can use TMR for good to help people decrease their bias but you can also just as easily…try to convince people that these ideas are better, or these people are better,” says Pereira, who has coauthored a paper reviewing the promise and pitfalls of TMR research.

TMR-brainwashing is unlikely, according to researchers. But there are other concerns: Imagine you’re in a hotel room and, while you sleep, someone pipes in the scent of a particularly buttery croissant. As you head down to the breakfast bar the next morning, you see the flaky pastry for sale. Sub any product or even idea in place of croissants and you get the idea.

“I think the main problem is…if you’re not informed about what is playing when you sleep,” says Rasch.

Should the most common incarnation of TMR be tools to boost learning, the risk of malicious applications, per Antony, is probably low. The more important concern, he says, is making sure people know the relative costs and benefits.

“We still don’t understand exactly how this works,” Pereira says. “You don’t know what’s going to happen and so you shouldn’t just go around trying it on your own at home.”

A few negative consequences of TMR experiments have appeared. In Lewis’s melody experiment, for example, people who received TMR did worse than others at performing or deciphering the second, un-cued sequence. This indicates that memory replays might be a limited resource within the brain. 

“If you prove that this is a safe and effective technique to enhance cognition, then how do you decide who gets access to it and why?”

Many experiments suggest a tradeoff, where reactivating and thus boosting one memory comes at the cost of cementing another. If TMR does result in a zero sum gain, users might fail to reinforce memories on which they aren’t focusing. Students might, say, plump up Spanish vocabulary at the expense of equations they learned in Trig.

Another concern is that TMR users might accidentally create false memories. This already happens when we transfer recollections to long-term storage; reactivations preserve or generalize the gist of an experience, but single details may crumble away. It’s possible that TMR would spur this process.

And, as with other potential tools to boost cognitive function, scientists have reason to be cautious. “We do something with the brain in a state where conscious control is limited,” Rasch says. “We have to consider [that].

Because we still don’t know for sure if TMR has any long-term health consequences, Rasch says he would not recommend it yet on a larger scale, or in people with conditions like PTSD. But for simple, individual use, such as vocabulary learning, he says there probably isn’t much danger.

At the onset, TMR could likely be expensive and confer even more advantage to people with money to burn.

“If you prove that this is a safe and effective technique to enhance cognition,” says Pereira. “Then how do you decide who gets access to it and why?”


Before the question of access and intent are even debated, years of research still need to be completed before scientists start spelunking into the subconscious to extract, enhance or alter memories. And per researchers, the future TMR presents is ultimately hopeful, even if it is laced with more than a little trepidation.

Imagine this scenario: Twenty years from now, a terrible event takes place for which you are present. Maybe it happened on your way home one night or while you were traveling. Whatever the case, it shook you deeply and you find yourself haunted by terrifying memories. You could engage in standard talk therapy, but you could also choose to make an appointment with a more specialized doctor.

During your session, the doctor would play a quick tone or diffuse a whiff of fragrance. That sound or scent emerges again as you’re enjoying an ordinary night’s sleep in his office, triggering and rewiring the memory as you dream. You’ll awake, free of the wicked remembrance as though it was nothing more than a mole that needed to be lasered. In the morning, you’ll sign a few papers shake the doctor’s hand.

What could have scarred and shaped the brain has been removed. It’s not even a memory of a memory — it’s simply gone. You’re free to move forward, unhindered, into the future.

Photos via Shutterstock (1, 2, 3, 4, 5, 6), Paul Bence/Flickr

Van Winkle’s is a new website dedicated to exploring how sleep affects and informs our lives, both at night and during the day. Sleep may account for one-third of our time, but it influences us around the clock. Whether it’s sleep as related to science, health, family, pets, sex or travel, we’re eager to learn more.