>How about this: Cell NETWORKS do "train" in the sense of artificial
>neural networks training on exemplars.
Yes, but this brings us into the multicellular realm. I was thinking about
single-celled organisms (i.e. the prototypical cell) in particular. The reason I
wonder is because the cell is exactly the prototype of autopoiesis. If the cell
is the only organism which does not have to "train" to stay fit, then all other
kinds of organisms are not directly autopoietic. "Training" sounds a little
un-autopoietic. An autopoietic system is by definition a perpetual motion
machine and should therefore not have to train to self-maintain, or at least I
will have to modify my view of autopoiesis if this is the case.
But I agree that training is a fundamental aspect of multi-cellular
organisms.
Onar.