101–106 of 106 posts
Earlier quoted context omitted.
We haven't found anything in nature that resembles reverse-mode automatic differentiation, either in evolution or in neuroscience.
What would that look like. Is there an implementation of the finite difference method in biology somehow?
Earlier quoted context omitted.
I'd be surprised.
Yeah thanks, me too. So I'm wondering what point 22 is getting at. How could that even be possible?
Earlier quoted context omitted.
Yeah thanks, me too. So I'm wondering what point 22 is getting at. How could that even be possible?
Biology can use discrete methods; we have discrete neurons (and other cells), discrete chemical species like ATP, discrete amino acids, discrete genes made of discrete nucleotides, etc.
Point 22 seems to imply that the other finds it notable that that isn't happening.
Earlier quoted context omitted.
Biology can use discrete methods; we have discrete neurons (and other cells), discrete chemical species like ATP, discrete amino acids, discrete genes made of discrete nucleotides, etc.
Fur sure. But I don't get how that leads to an expectation that our biology is computing and utilizing derivatives in a computational graph. Point 22 seems to imply that the other finds it notable that that isn't happening.
Earlier quoted context omitted.
> Point #2 ("somewhat less fit... on average") is totally inaccurate if the parents are statistically average in the modern/Western world. I wonder if you've misunderstood the point. Offspring are expected to be less fit on average because -things can go wrong- (mutations, birth defects, etc). But selection is a counterweight to this.
Seemed to me that the author was referring to regression to the mean, as another commenter noted. De novo mutations have a negative effect, to be sure, but it is extremely weak on an individual level. In parents who are extraordinary, the effect of regression to the mean is going to be 20x to 40x stronger than the effect of de novo mutations. For instance, if you have two parents who are both 195cm tall, the regressi…
> That if you’re a life form and you cook up a baby and copy your genes to them, you’ll find that the genes have been degraded due to oxidative stress et al., which isn’t cause for celebration, but if you find some other hopefully-hot person and randomly swap in half of their genes, your baby will still be somewhat less fit ...
You're right that it's a relatively weak effect-- which is a good part of why the effects of variance and selection (incl sexual selection) win out and fitness doesn't decline with each generation.