Comment by A_D_E_P_T

> If you need oxygen, then you need sleep!

Do plants sleep? Don't some insects, like flies, live without any sleep?

> why ancient ancestors started sleeping

I tend to believe that our ancestors didn't start sleeping, they started waking up ! the default pattern is sleep and conservation of energy, but you need to wake up to expend more energy for a short period in order to feed yourself efficiently

Hey, that's Hyrum's Law!

Sounds like my microservices

sounds like legacy code

That's what I still 'believe'. Wake-sleep algorithm [1] is a good start for speculation. I think brain needs to be in a different mode to reorganize its weights and to forget unnecessary things to prevent overfitting. In this mode we happen to be unconscious. I also believe dreams are just hallucinations caused by random noise input to the system. The brain converts noise distribution to a meaningful distribution and samples from that. I have zero evidence btw, but I believe these are related.

[1] https://en.wikipedia.org/wiki/Wake-sleep_algorithm

Curious how the zeitgeist changes, on a previous AI cycle we could thought sleep was required/generated by a semi-space garbage collection brain-LISP process :)

> sleep was the training of our neural network with all the prompts of the day

Periods of sleep certainly seem to be used in that sort of way, but that is an extra use evolution found for the sleep cycle once it existed rather than the reason sleep developed in the first place.

There are a number of things that seem tied to, or at least aligned with, our wake/sleep cycle that likely didn't exist when sleep first came about.

You didn’t need this study to realize that this was wrong: jellyfish and hydras also sleep despite not having a central nervous system. There are indications that sponges sleep too, despite not having any neurons (though obviously it’s somewhat ambiguous): https://www.science.org/content/article/if-alive-sleeps-brai...

It's not training as much as it's discarding bad examples. Sort of.

Rebalancing the weights.

Jesus christ, not even a biology thread is safe in the orange website.

I feel this too, and always wondered if it related to the glymphatic system [1].

This is the system that clears out metabolic waste from the brain which builds up over time, and it's theorized that during slow-wave sleep in particular, the slow waves help pump out this waste fluid through microscopic channels the open up.

AFIAK, there were some researchers that were wondering if a drug of some kind could force this to happen more quickly, thus cutting down the amount we need to sleep. (Probably a bad idea.)

1. https://en.m.wikipedia.org/wiki/Glymphatic_system

> it feels like there is poison built up in my mind, then sleep clears it up/feel better.

I'm not sure how common this is, but I feel this acutely after sustained mental exertion (e.g. reading informational material for a few hours). A deep 15 minute nap takes the feeling away completely without any grogginess.

Exactly! Going offline for hours every day in an adversarial world is positively nuts! The reason can't be idiosyncratic. No gentle gradient of comparative advantage can rationalize it. It must be something severe and nigh impossible to do any other way.

Furthermore, sleep is very specifically about the taking the brain offline: that's what deteriorates first in the absence of sleep and the tortured workarounds for animals that absolutely must avoid sleep (e.g. migratory birds) involve sleeping part of the brain at a time. Any explanation that isn't highly specific to the brain's responsibilities has the immediate hurdle of explaining this away, and for that reason I don't buy the mitochondrial explanation. Mitochondria are too universal and sleep is too specific to the brain. Energy is fungible, so I don't buy that nature wouldn't figure out the "trick" of having a subset of the mitochondrial population sleep at a time.

My money is on the "brain algorithm" requiring an online/offline phase, whether that's contrastive learning or memory consolidation or something else. There are lots of candidates for fundamental brain algorithms with the "feature" that they require an offline phase that cannot be incrementally worked around, and these fit the observations much better.

the interesting mystery is the mechanism of that dependency, not its evolutionary advantage.

Nah, I'd say the evolutionary advantage is the more interesting mystery. The mechanism is just an implementation detail, after all.

And by the way, if we tamper with something without understanding its purpose we risk messing something up.

It's interesting that sleep is controlled by mitochondria, but sleep is clearly involved in learning, and whatever algorithm for intelligence the brain does. Do those algorithms still work if you intervene at the level of the mitochondria? Or are the mitochondria just a good way of measuring elapsed time through energy expenditure? e.g. The algorithm needs a sleep phase to run roughly every x neural firings, or performance degrades and mitochondria were available as measuring devices when nature needed a way to guess how long the wake phase had been running.

Maybe you could intervene to prevent anyone from feeling tired, but would the learning algorithm still work? That part is still a mystery.

I don't mean it as an attack on GP, but no, I don't agree that this is pedantic. This happens constantly when science is popularized -- people read one article and leap to the conclusion that a problem has been revolutionized/solved/answered simply because they're reading about it -- and no, the HN audience is no better. Technophiles love a good scientific revolution story.

It's very much a fundamental misunderstanding of how science works. Almost nothing in science has an answer, and if you let your brain lock in that way, you forego the opportunity to ask interesting questions. It also leads directly to lots of downstream pathologies common in amongst laypeople (e.g. "The Science is Settled", which it almost never is).

> Going further, yes, this is a new theory among others, but afaik is the first one with strong evidence.

I am not an expert in this field, but others have evidence too. Particularly when asking "why" questions like this, the bar for proof is incredibly high.

On the contrary, this is such a common misunderstanding that it practically defines the meme of pop science.

Proposal of a hypothesis is not answering. Even if, decades from now and after many additional studies, scientific consensus settles on this hypothesis as "the answer", the first paper to speculate about the idea is still just a speculation. Moreover, if you're an outsider, the speculation is often an idea that's been floating around the field for longer than you've been aware of it.

Basically, just abandon your notion that there is "an answer" to any sufficiently complex scientific question, and you will be better off.

I've never heard that, it doesn't really make sense given what we know about REM sleep and slow-wave sleep, and the Wikipedia page on blinking doesn't mention that at all, not even as an old theory.

The paper proposes what is one level deeper, though.

Filling in the gaps: Mitochondria are less efficient due to electron leakage -> ATP gets consumed faster -> adenosine builds up faster

The first step is the new one.

That's almost exactly what people said about the appetite -- about the biochemical pathways which govern hunger, which are known to be massively redundant and overlapping.

But then Ozempic was released and it turned out there was a shortcut after all.

Which is not to say that such things are necessarily "healthy" or desirable, just that you can't rule out that biochemically-modifiable characteristics, however complex, have "one simple trick!" you can use to attain a desired end.