Sleep on it: How the brain processes many experiences – even when 'offline'

Starting with my undergrad but fully committing to it by grad school, I (and several of my friends who went through similar programs in math/cs) have a strategy that uses this. If, for example, I had a new problem set in a math course, I would bash my head against it in the evening for an hour or two. I'd make an honest attempt but move on from problems quickly if I got stuck. I'd rarely get much done. Then, I'd do my best to get a good night's sleep (at least 7.5 hours quality sleep). In the morning I'd try the problems again first thing after coffee, and frequently found that I could do a significant portion of the problems, or at least make headway. This might be biased by the fact that I'm really much more of a morning person to begin with, but I know several people who use this strategy.

Hippocampal replay was the main subject of my dissertation. It has been studied primarily in rodents, but there have been a lot more human studies in the meantime.

My PhD proposal was to suggest that cognitive fatigue is an adaptive construct. Rather than reflect a depletion of glucose and that people can't function anymore, cognitive fatigue is a suggestion for the agent to go 'offline' and replay.

Two of my collaborators wrote an extremely influential paper writing down a Q-learning equation for replay: https://www.nature.com/articles/s41593-018-0232-z

Admiral Raymond Spruance of the World War 2 US Navy was known as very calm in a crisis; he was very serious about getting enough sleep so that he was well-rested during any battles. When other officers would stay awake for 36 to 48 hours at a stretch, he would read a novel and get sleep because he knew he had enormous responsibilities that needed him at his best.

He also walked 8+ miles a day, even when at sea he would make sure he walked around the ship, usually with some other officers to discuss any pertinent issues of the day. Walking is great for turning over problems in your mind, or even just daydreaming to give your subconscious mind "space to work".

I’ve been lucky enjoy to have been able to live around my unconscious. I made a conscious decision to allow my unconscious to guide me through life.

The original idea to do this came from Le Corbusier[1] who once described his process of working as being a phase of collecting details on a project, a phase of doing something else (allowing his unconscious to work on the project) and finally he would sit down and complete the project.

The disadvantage is that I never know when inspiration hits and when exactly I will get something done. It’s important to be organised and have everything written down is my approach.

Also I give myself time and room to explore possible solutions from seemingly unrelated areas - a kind of zen navigation[2] for project work.

[1] https://en.wikipedia.org/wiki/Le_Corbusier

[2] https://www.goodreads.com/quotes/667285-he-had-a-tremendous-...

Something I've been wondering about - but have been unable to find any solid research on - is if it would be "optimal" to sleep immediately after any learning/training session, whether it be mental or physical, instead of just resting while still awake.

If sleep is the best state for the body to be in to consolidate memories, reduce fatigue, etc., then it would seem logical to try and be in the sleep state as much as possible.

Obviously the difficult part is actually being able to fall asleep on command without using some kind of pharmaceutical, but I do think falling asleep quickly is something that can be learned:

https://www.inc.com/melanie-curtin/want-to-fall-asleep-faste...

Over the years, I've learned to recognize when I'm stuck on a problem in a way that will be resolved by sleeping on it.

I can go to sleep confident that in the morning I'll probably figure out what I just can't get a handle on right now.

Related:

The Committee of Sleep: How Artists, Scientists, and Athletes Use Dreams for Creative Problem-Solving – https://en.wikipedia.org/wiki/The_Committee_of_Sleep

Car drive: Bash head against wall all day. Give up. Drive home in zoned out state and realize solution (or at least the unseen issue).

The Book The Sleep Solution or Why We Sleep touches on this. It also claims that REM sleep is a core part of this experience. The analogy that I found most useful is that REM sleep writes data from the cache to hard disk. In the process it cleanses that data reducing the emotional overload attached with it.

Say you just experienced grief. Dreaming about it is an essential in the process of getting over it.

REM sleep is inhibited when you use sleeping pills or alcohol.

Interesting, as I would divide my dreams up 25/25/25/25 between mundane daily experiences, the same or very similar dream, totally new situations and finally the surreal ones where you magically end up in random places and random situations.

Always figured it's a blend of taking past and recent experiences and re-ordering them, with a hint of hypothetical scenarios for the future.

The "flicker" sounds like a good A/B test for the hypothetical while the replays are good for memorisation.

I can often tell when I'm dreaming when it comes to the more hypothetical ones due to dodgy physics or whatever (try nipping my face and can't feel anything) so I'm probably buggering that process up a bit. Maybe an INTJ trait. I remember doing that as a kid and being able to fly, but nowadays often I can't get beyond 10 metres above ground. Been rate limited.

I've often though about sleep in RPG terms. Sleeping gives me 1 experience point to put into any skill, and I get to choose which skill gets the experience point by what I work on during the day.

I haven't always succeeded, but I try to at least work-on or study something I care about before bed.

In undergrad, I would tackle problem sets or a programming course problem as soon as I could. I would wrestle and fight, sometimes I made progress, usually not.

Night of sleep and going to $dayjob to pay for school, ideas would just manifest themselves in my head. Happened dozens and dozens of times.

I once had the following experience as a kid: i learnt a skill (in that case, juggling), completely in my sleep. I clearly remember trying it in my dream, failing, in what was pretty much a "dry-run" or a simulation, where maybe several hours of actual practice were emulated within a few minutes of dreaming. When I woke up, I try, and it worked, I had captured the actual coordination/reflexes necessary for it to work. I also solved several problems/issues on a few occasions but this experience was the most stunning because it resulted in getting an actual physical skill that since hasn't gone away.

In college especially before a test my brain was on study mode and would not turn off. Most often I couldn't sleep because I was trying to solve a problem. There was no problem it was just nonsense and no solution. I was going through the motions of it but it was pointless. I wish my brain actually would have worked on actual problems while I slept.

edit: I also recall when I was a naive late teen early 20s "song writer" (wannabe) I would often wake up with lyrics. They were also nonsense. Literally. I had a notepad and wrote what I thought was great but it was just gibberish.

Sleeping to solidify memories and concepts is like when you were a kid and you just slept off the entire 8 hours car journey. Not only you get the sleep, not only you get the result, but you also didn't even notice it happen

Barbara Oakley's "Learning How to Learn" course describes this processing as "diffusion". It's amazing how much just doing something else for a while when I get stuck on something can help, be it a drum beat I'm having trouble replicating or a query I'm having trouble writing.

(Course is https://www.coursera.org/learn/learning-how-to-learn)

Rick Hickey made a great talk about that: https://www.youtube.com/watch?v=f84n5oFoZBc

Since this is processing of past events and future possibles during sleep, would it be fair to hypothesize that animals that sleep actively (appear to dream) are conscious when awake?

This is what the hammock is for in hammock-driven development, isn't it? ;)

--

1: https://youtu.be/f84n5oFoZBc

this phenomenon is easily experienced while fighting an Elden Ring boss... very often players will spend 2-3 hours fighting a boss, learning the patterns and trying their best... and sometimes its just impossible. but then you sleep and boot up the game the next day and you get it first or second try. Its a really well known thing in the DarkSouls community and it is common advice.

A way to improve it is to use ZettelKasten note taking to detail the problems in tackling the problem so that when you wake up you can review those notes before re-examining your approach to the problem...

I started using this approach about 20 days ago...came up with new software architecture beyond cleanarch due to this new using of those techniques.

The ideal work schedule for me is work first thing in the morning for 2 to 3 hours.

Eat lunch with people, ideally people who are technical enough to discuss work. Extend lunch to an espresso afterwards.

Quick nap in the afternoon, any errands/chores, answer any emails, schedule meetings in this afternoon slot.

Cook dinner, eat with family.

Work another 2-3 hours before sleeping.

Ok can someone explain how they reached this conclusion that the brain is able to separate, merge, or drop experiences during sleep within 1 second? The rat experiment mentioned doesnt explain how they are able to interpret brain signals and map them to prior experiences.

So the brain will bundle fragment of experience, shrink or extend them and replay or even preplay it during the sleep.

Sounds like a superpower in daily life, but general routine in sleep.

Magical

Alternative access without paywall: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6203620/

Are we unconscious when we sleep or do we merely lose the capacity to remember? How could we tell the difference?

I have a huge problem with getting quality sleep, thank you to the op!

This reminds me Henri Poincaré's "Science et méthode" (1908)

I am wondering if we should be paid for consistent sleep time

It's what LLMs are missing. They don't sleep. If only they could sleep, they would be able to make novel experiences stick past the context buffer. Sleep is like fine-tuning on previous interactions.

What if...

Combining the topics of manipulation, gaslighting, intermittent reinforcement, and the brain's processing during sleep reveals how these psychological tactics might affect memory and emotional regulation. Manipulation and gaslighting distort a person’s reality, creating confusion and emotional distress. Intermittent reinforcement keeps the individual emotionally hooked, much like a brain's parallel processing of unrelated experiences during sleep. These experiences can disrupt the brain's ability to properly consolidate memories and manage emotions, potentially causing maladaptive neural patterns. Over time, this can lead to cognitive dysfunction, impaired decision-making, and an increased vulnerability to further manipulation.

After much thought I have decided that thought is central to this whole thing. Anybody who disagrees is dreaming.

EDIT Satire is lost on you people.

This level of complexity and sophistication somehow "randomly happened", they said