Hot take: Sleep and memory go together like Sundays and self-loathing, which is to say quite well. Now, researchers from The Netherlands and Switzerland have concrete evidence that deep sleep does a much better job strengthening our memory of information that's related to stuff we already know than for wholly novel facts and faces.
Slightly more lukewarm take: During sleep, our brains largely stop processing the world around us, and instead focus on a) converting newly learned information and experiences into longterm memories, a process called memory consolidation, and b) shedding garbage thoughts, a process called neural pruning.
The mechanism underlying memory consolidation is called neural replay. The experience of learning new information (for me: "banal" rhymes with "canal," not "anal") corresponds to a distinct neural firing pattern. During sleep, our brains "replay" those distinct neural firing patterns, thereby enhancing our memories for those experiences. (Neurons that fire together wire together, and don't ever forget it.)
Scientists developed a way to use scents and sounds to "cue" certain pieces of information during sleep, meaning manipulate which experiences our deep-sleeping brains replay, downplay and cast aside as refuse. It's called Targeted Memory Reactivation (TMR).The gist of TMR is pretty simple: Study volunteers learn something new — a list of foreign words, perhaps — while they're awake. While they're studying their assigned German verbs, researchers either play a distinct sound (brrring, brrring) or diffuse a smell (roses). Then, when volunteers reach stages 3 and 4 of sleep, researchers re-play or re-diffuse that same sound/scent, triggering neural replay and strengthening the memory.
TMR is still a research-only procedure (meaning it's not approved as a treatment in any clinical context), and it doesn't work all the time and might not work for that long. But, it's getting better and becoming better-known. Potentially, TMR has broad-ranging use — for helping dementia patients relearn forgotten information, helping people with trauma disorders forget scarring experiences and correcting deep-seated, perceptual biases that lie at the root of the prejudicial -isms.
TMR works, experts say, because of the way our brains integrate new information into existing knowledge. When we recall an experience, we typically draw on multiple senses. A full memory of say, walking into your own surprise party, might include the scent of beer, the sight of a particular friend throwing his arms up, the sound of tone-deaf people bellowing the happy birthday song, and so on. Those sensory cues (i.e., the beer scent or the off-key "happpppy birthday too youuuu") aren't all stored in the same place in the brain. Instead, they live in overlapping neuronal networks littered across the brain. Re-exposure to that beer smell can trigger a memory for the whole birthday-surprise experience, since the experience breaks is essentially a web of interconnected memory fragments.
Let's say that, today, you have to learn to sing two songs in French — the happy birthday song and something you've never heard before. Once you learn them, researchers could use TMR to "cue" them during sleep, meaning they'd replay the first stanza of each song a handful of times while you're snoozing in order to help you form long term memories of them. But neuroscientists have long theorized that deep-sleep cueing would do a lot more to help you commit "bon anniversaire a toi" to memory than the other, entirely new song. A new study, published in Nature, confirms their hunch.
The study didn't involve learning songs. Instead, participants were shown made-up words (like "snookle") paired with pictures of objects. Half of the objects were familiar (a chair) and half weren't (a spiky, 7-sided container). Participants studied all of the word-picture pairs (snookle + chair; veisel + 7-sided container) while they were awake and immediately got tested on them. Then, researchers "cued" some of the words while participants were asleep, and retested them, again, on all of the words (without the pictures to help) the following morning.
So, the sleep-cued words included made-up words they had and hadn't remembered during the initial testing, for both familiar and unfamiliar objects. In the AM, participants showed much better recall for cued words only when they'd originally been paired with familiar objects, regardless of whether or not participants had remembered them in the first test.
The conclusion: Sleep-cueing helped people learn information if they could connect it to prior knowledge of some type, but not if they couldn't.