Gregory Bateson has a good piece in one of his books (it might have been "Steps to an Ecology of Mind") where he derives from cybernetic principals the result that Darwinian evolution works better than Lamarckian.
The key insight is that we can think of meta-evolution driving the form of the "first-order" evolution, meaning we evolved to evolve Darwinian-ly. (In practice meta-evolution cannot be separated from evolution, there is only one.)
This implies that we should expect limited Lamarckian evolution anywhere it leads to greater fitness than Darwinian.
From this POV epigenetic mechanisms are expected.
(I still look askance and the so-called "Central Dogma" of biology. Why on Earth would you assume that genes can't be changed by soma? If there was adaptive advantage to being able to edit DNA then it seems like cells and viruses would have it.
As I typed that my brain said: CRISPR stoopid. SO, yeah, duh.)
> This implies that we should expect limited Lamarckian evolution anywhere it leads to greater fitness than Darwinian.
Not necessarily... evolution is good at finding local maxima, but not necessarily global ones if there's a deep valley in between. Transgenerational epigenetic inheritance is hard to evolve (compared to more Darwinian/DNA sequence-based evolution, it requires extra mechanistic steps and typically the effect size is small), so most of the time it isn't employed. If you look at enough cases where Lamarckian evolution might be advantageous, you'll find a few where it's employed to a limited degree, but in any given situation, it probably won't be.
The key insight is that we can think of meta-evolution driving the form of the "first-order" evolution, meaning we evolved to evolve Darwinian-ly. (In practice meta-evolution cannot be separated from evolution, there is only one.)
This implies that we should expect limited Lamarckian evolution anywhere it leads to greater fitness than Darwinian.
From this POV epigenetic mechanisms are expected.
(I still look askance and the so-called "Central Dogma" of biology. Why on Earth would you assume that genes can't be changed by soma? If there was adaptive advantage to being able to edit DNA then it seems like cells and viruses would have it. As I typed that my brain said: CRISPR stoopid. SO, yeah, duh.)