Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mammals dream about the world they are about to experience before they are born

news.yale.edu

Post reply on HN
Print view View original

41–50 of 193 posts

Re: Mammals dream about the world they are about to experience before they are born

#41

Post by akomtu »

teruakohatu wrote: post #17
gojomo wrote: post #12

Something about this reminds me of the 'Winning Ticket Hypothesis' in artifical neural networks: that some 'random' initializations prime a network far better for later faster learning. From https://arxiv.org/abs/1803.03635 - the abstract: Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of infere…

> Are these innate patterns of activity priming mammalian brains in a similar way? A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning? Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains that cannot…

I guess it's firmware of some kind. Newborns already know how to breathe, how to control the heart rate and tons of other low level things. Legs movement is a small feature in comparison.

Re: Mammals dream about the world they are about to experience before they are born

#42

Post by stevenhuang »

dogorman wrote: post #36
taneq wrote: post #25

Earlier quoted context omitted.

I find it highly implausible that many people believe fetusus are "not alive". You might be referring to the widely held position that an early-term fetus isn't sufficiently developed to be considered a full-fledged human, and that any rights it might have don't outweigh those of the mother who is carrying it?

I've found that "fetuses aren't alive" is a common talking point espoused by people who aren't comfortable sitting in a moral gray area and want an simple excuse to resolve their cognitive dissonance. It's common enough that I can't help but wonder if that is the reason people are so resistant to the idea that fetuses might dream. It seems remarkably similar to "farm animals can't feel pain" . A transparent fib peopl…

I think you've confused the talking points.

I've heard the saying that fetuses aren't conscious, but I've never heard anyone, however fringe, say fetuses aren't alive.

Re: Mammals dream about the world they are about to experience before they are born

#43

Post by outworlder »

chrisseaton wrote: post #26
teruakohatu wrote: post #17

Earlier quoted context omitted.

> Are these innate patterns of activity priming mammalian brains in a similar way? A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning? Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains that cannot…

> Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains Have you ever seen a new-born baby's legs? They aren't even remotely strong enough to walk, no matter what the brain tells them. It takes those months to develop leg muscles. New-borns don't have them. It's absolutely not restricted by the brain size.

True. But the movements on all limbs are quite uncoordinated at first. It's also not restricted to muscle strength.

Re: Mammals dream about the world they are about to experience before they are born

#44

Post by emmelaich »

chrisseaton wrote: post #26
teruakohatu wrote: post #17

Earlier quoted context omitted.

> Are these innate patterns of activity priming mammalian brains in a similar way? A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning? Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains that cannot…

> Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains Have you ever seen a new-born baby's legs? They aren't even remotely strong enough to walk, no matter what the brain tells them. It takes those months to develop leg muscles. New-borns don't have them. It's absolutely not restricted by the brain size.

Well, the legs may be less developed since they're not useful because ... the brain size is not big enough?

Anyway, hard to conclude, just think your logic is missing a little.

Re: Mammals dream about the world they are about to experience before they are born

#45

Post by gota »

gojomo wrote: post #12

Something about this reminds me of the 'Winning Ticket Hypothesis' in artifical neural networks: that some 'random' initializations prime a network far better for later faster learning. From https://arxiv.org/abs/1803.03635 - the abstract: Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of infere…

I guess I can see some parallel the way you describe it, but it is not that tight of an analogy IMO. Correct me if I'm wrong but the Lottery Ticket Hypothesis mainly means that, for many of our tasks, we start with needlessly huge networks, and thus a good portion of the learning process is just the network learning to "cope with the noise" from the extraneous connections

Hence why the lottery ticket network (being a smaller version of the same initial network) can achieve similar accuracy (but not 'better') in much shorter time - because you took away all the confounding stuff. That kind of 'counterfactual' doesn't really translate to a biological domain, right?

Also - the baby brain is not really a 'pruned', smaller version of the adult brain, right? Neither is the fetus brain compared to the baby brain (which is more to the point of the article). So the analogy breaks down there too

I guess what the article describes is a bit like "pre-training" a full network with 'synthetic' experiences that quickly tune it for later. In that sense, I think its like these "dreams" are a 'distilled dataset' [1]. The question then is: how is this dataset being passed on from mother/father to child?

[1] https://arxiv.org/abs/1811.10959

Abstract:

> Model distillation aims to distill the knowledge of a complex model into a simpler one. In this paper, we consider an alternative formulation called dataset distillation: we keep the model fixed and instead attempt to distill the knowledge from a large training dataset into a small one. The idea is to synthesize a small number of data points that do not need to come from the correct data distribution, but will, when given to the learning algorithm as training data, approximate the model trained on the original data. For example, we show that it is possible to compress 60,000 MNIST training images into just 10 synthetic distilled images (one per class) and achieve close to original performance with only a few gradient descent steps, given a fixed network initialization. We evaluate our method in various initialization settings and with different learning objectives. Experiments on multiple datasets show the advantage of our approach compared to alternative methods.

Re: Mammals dream about the world they are about to experience before they are born

#46

Post by throw_nbvc1234 »

emmelaich wrote: post #44
chrisseaton wrote: post #26

Earlier quoted context omitted.

> Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains Have you ever seen a new-born baby's legs? They aren't even remotely strong enough to walk, no matter what the brain tells them. It takes those months to develop leg muscles. New-borns don't have them. It's absolutely not restricted by the brain size.

Well, the legs may be less developed since they're not useful because ... the brain size is not big enough? Anyway, hard to conclude, just think your logic is missing a little.

Except that children can walk with assistance (and crawl) before they're strong enough to walk on their own.

Re: Mammals dream about the world they are about to experience before they are born

#47

Post by chrisseaton »

emmelaich wrote: post #44
chrisseaton wrote: post #26

Earlier quoted context omitted.

> Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains Have you ever seen a new-born baby's legs? They aren't even remotely strong enough to walk, no matter what the brain tells them. It takes those months to develop leg muscles. New-borns don't have them. It's absolutely not restricted by the brain size.

Well, the legs may be less developed since they're not useful because ... the brain size is not big enough? Anyway, hard to conclude, just think your logic is missing a little.

But if you hold a child up before their legs are strong enough they will walk.

Re: Mammals dream about the world they are about to experience before they are born

#48

Post by gota »

emtel wrote: post #39
gojomo wrote: post #12

Something about this reminds me of the 'Winning Ticket Hypothesis' in artifical neural networks: that some 'random' initializations prime a network far better for later faster learning. From https://arxiv.org/abs/1803.03635 - the abstract: Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of infere…

It’s not just mammals. I’ve always been astounded that spiders know how to spin webs. Their mothers don’t teach them how!

Or this, which is mind-blowing:

https://www.smithsonianmag.com/smart-news/dragonfly-undertak...

Insects that have large migration patterns - all of the individuals die before coming back to the initial spot; only their descendants continue the cycle. How?!

Re: Mammals dream about the world they are about to experience before they are born

#49

Post by optimalsolver »

gojomo wrote: post #12

Something about this reminds me of the 'Winning Ticket Hypothesis' in artifical neural networks: that some 'random' initializations prime a network far better for later faster learning. From https://arxiv.org/abs/1803.03635 - the abstract: Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of infere…

How hard would it be to find random networks that don't need to be trained at all?
Post reply on HN