The irreducibility of natural selection

[MarxSchmarx]

Superficially, Darwinian natural selection appears to be a wholly "self-contained" theory. In particular, once one premises that:

1. Entities X differ in their survivorship
2. These differences are heritable between X_t and X_{t+1}

Then, whatever we happen to be talking about (biological organisms, computer programs, societies??), a sort of optimization routine should follow, all other things being equal.

That such a "principle" of biology can be articulated, therefore, appears to in essence contradict the notion that "biology" can be reduced to chemistry. Yet the reduction of biology to chemistry, and ultimately to physics, appears necessary for the unification of the project of modern science. What do comrades make of this?

[ÑóẊîöʼn]

Pragmatically speaking, I don't see this "unification" that you talk about at all. Attempting to use the "language" of atoms and molecules to describe macroscopic organisms, let alone entire populations, strikes me as incredibly unwieldy, even if we assume it could be done as a matter of practicality in the first place.

Nature operates at different "levels" - from the subatomic through to the macroscopic all the way to the cosmological. This isn't to say that there is no overlap - indeed, discoveries in particle physics, dealing with the smallest lumps of matter possible, can have profound implications for cosmology, the study of the largest object known.

Reductionism and holism are both, by themselves, insufficient. Can there not be a rational synthesis?

[Luís Henrique]

Superficially, Darwinian natural selection appears to be a wholly "self-contained" theory. In particular, once one premises that:

1. Entities X differ in their survivorship
2. These differences are heritable between X_t and X_{t+1}

Then, whatever we happen to be talking about (biological organisms, computer programs, societies??), a sort of optimization routine should follow, all other things being equal.

This is valid only for biological organisms - nor for computer programs or societies, which do not mutate or reproduce in the same way as biological organisms.

Luís Henrique

[MarxSchmarx]

True, it seems that there are two levels of discussion here. On the one hand, there is the issue of what is the material basis for all that happens biologically. By this token, it is rather trivial that biology degenerates into chemistry and hence to physics, because all biological entities are just arrangements of molecules and ultimately atoms/subatomic particals.

On the other hand, the forces that operate in biological evolution, and the "laws" invoked appear quite distinct from the laws physics, at least superficially. It is fair to ask whether these are, in fact, distinct. For instance, "variation due to unequal crossing over" is in some sense "random", but it ultimately can be attributed to the rules of statistical physics (or, at least, so I suspect). Taking for the moment this as a valid analysis (for the sake of argument), is the generation of this variation itself distinct from physics? Probably not. So then what of the other operating principles of biological change?


Pragmatically speaking, I don't see this "unification" that you talk about at all. Attempting to use the "language" of atoms and molecules to describe macroscopic organisms, let alone entire populations, strikes me as incredibly unwieldy, even if we assume it could be done as a matter of practicality in the first place.

It may not happen within our life times, though I suspect with powerful enough computers it probably can be predicted to a very limited extent. And once this quantum computing thing is figured out why the sky seems the limit in terms of showing how larger scale phenomena can be reduced to the fundamental laws of physics.



Nature operates at different "levels" - from the subatomic through to the macroscopic all the way to the cosmological. This isn't to say that there is no overlap - indeed, discoveries in particle physics, dealing with the smallest lumps of matter possible, can have profound implications for cosmology, the study of the largest object known.
Reductionism and holism are both, by themselves, insufficient. Can there not be a rational synthesis?

A fair question, and I think this is, to some extent, what the problem presented boils down to. Ultimately the problem is not unique to biology, but can be found in any number of fields, including, I am guessing, physics. A reductionist would disagree, because, almost by definition, that is what their program is about. Holism itself finds few "pure" adherents, although from a pragmatic standpoint some synthesis of the two has been fruitful. Nevertheless, one has this lingering doubt raised by the reductionist that such pragmatism is only temporary, a provisional "alliance of convenience" until such time as a more reductionist theory emerges.



This is valid only for biological organisms - nor for computer programs or societies, which do not mutate or reproduce in the same way as biological organisms.


Why does it matter precisely how the mutation occurs? For instance, a computer program can be set to "mutate" in the sense that it undergoes some change, random from the perspective of the fitness function used. This is directly analogous to how mutations occur among organisms, at least as far as the question of adaptation (or lack thereof) is concerned.

[ComradeRed]

This is valid only for biological organisms - nor for computer programs or societies, which do not mutate or reproduce in the same way as biological organisms.

Luís Henrique Well, that depends on whether one models society as "evolving" (in some sense) or not.

In some sense, "evolution" is something with a feedback loop that just uses the feedback to optimize performance.

I think that human societies have a similar feedback loop, but that's a matter of debate (I guess).

But with regards to computer programs, you are correct, they're just algorithms - nothing evolutionary there.

Just a predetermined sequence of machine code.