Evolution and Grammaticalization (2)
phonosemantics at earthlink.net
Wed Mar 1 04:30:58 UTC 2006
First apologies for how far off linguistics this will be getting...
Again, things ain't that random. Free or bound. Black or white. Tastes great or less filling. What happened to intermediate states, all the nice gray areas and fuzzy boundaries? I have this sort of argument with a Chomskyoid friend of mine all the time.
OK- if things were completely random then anything could mutate into anything. But that's not what we see. There are limits. There is relative freedom between the limits when that's what's been selected for. Take for example conserved parts of genes, which code for important functional molecular groups in proteins. Conservation implies some sort of form/function fit which living organisms literally cannot live without. Histone proteins, which bind DNA into chromatin for compact storage and now also apparently for periodic organization as in my previous post, are hardly variable across all eukaryotic organisms. Sure they can mutate beyond these limits, but then its bye-bye baby! Other proteins, such as serum albumins, vary wildly in their particular 3-D conformation and most of their individual peptide identities. Except for the ones that bind molecules and ions that need to be carted around the bloodstream. The shape of the molecule is just not that important, so selection allows a great latitude here.
Most proteins are somewhere in between- they have the structure they need to function (the business ends, regulatory sites, docking sites for attachments to various pieces of architecture or other parts of more complex multi-protein machines), or to fold into their proper shapes (with the oil-loving side chains in the center, and the water-loving ones on the periphery), and so on.
Shapes and charges are often complementary- both within-unit and between. Parts of larger protein complexes have to interact- therefore if one part changes its conformation due to a mutation, another in its partner unit may undo the significance of the original one. And so on, across the board of the 'proteome' as its now being called. In spirit rather like a gigantic multidimensional and dynamic jigsaw puzzle, whose pieces' borders shift around a bit, but complementarily. Even so, the flexibility of the system is rather large. My growth hormone will work in a fruit fly in a pinch, thank you very much.
The vast majority of such point mutations either kill the organism outright, make its reproductive success less likely, or are simply neutral. Sometimes mutations are simply out of the control of the cell- radiation, chemical damage, a kink in the DNA at the wrong place and time, etc. At other times these (and other types of) mutation(s) seem to be facilitated by the organism. This isn't 'intelligent design', its regulation of repair and reproductive genetic shuffling. In other words, the organism actually has some (unconscious but based on tried and true 'memory') control over where and when mutations take place. The repair of DNA can be scaled up or down in accuracy or rate. This often seems to happen in reaction to outside events having to do with size of populations, resource availability, moving to new environments, etc. It adapts. The events of meiosis are also very interesting in terms of error correction.
Although nobody has looked yet, I'm betting the system has an internal model of itself- whether it is centralized or distributed who can say? First the existance of such a thing would have to be demonstrated.
But most point mutations have little to do with evolution at the level most biologists are interested in. From a systems point of view a hydra is little different from a hamster. Sure the hamster has more bells and whistles, and a bigger staff cellwise, but this comes far more from regulation changes than from the particulars of individual protein structure generated by DNA point mutations. In the jigsaw analogy I'm making, one can make and break links between pieces, change the nature and degree, spacing and timing of interactions. The basic parts- the mortar and bricks- are still essentially the same. In eukaryotic DNA, much of the regulation comes from the 'junk' DNA that comes in between actual protein-translated genes. And the 'junk' makes up the vast majority of the DNA (think of it as 'dark matter'). Most of the 'junk' (interesting how easily one can dismiss what one isn't currently examining, isn't it?) is made of repeated patterns of DNA, sometimes longer, sometimes shorter. Far from random. If your random mutation hypothesis were correct, then one would expect to find the 'junk' cumulating mutations. But this is not so. It appears that such mutations (though they MUST occur) are weeded out, either by death, failure to reproduce successfully, or by repair mechanism in cell.
Since as I said the bulk of interesting evolution comes from changes in such DNA (changes in the population sizes of the repeat variants, their placements, their N-mer number), that doesn't leave much room for randomness, at least from the point mutation POV. Perhaps you meant the outcome of the struggle to occupy space on the DNA wasn't predictable (these repeats may be originally viral in origin, by the way- remember what I said about integration of structurally reduced parasites into the host? Other larger parasites are now regularly being found to modify the behaviors of their hosts to increase their own reproductive fitness- maybe this is the origin of split genes and junk DNA in the first place, as eukaryotes were cobbled together by otherwise mindless ancient viruses for their own use). External context, though, may end up canalizing such changes in regular ways- something perhaps similar in spirit to grammaticalization, by the way. Neither complete randomness nor complete predictability.
Now, are all cell-induced mutations or other modifications (such as methylation of nucleotide bases and similar marks added or subtracted from proteins, polysaccharides, lipids etc., all of which also play very important large roles I've ignored here- for instance a new theory that polysaccharides form a system of mailing addresses that direct proteins and other molecules to the right places at the right times) always random? NO. Because the DNA itself can be opened and closed up in controlled fashion, and because there are proteins and other molecules which can recognize sequences, plain or tagged, the cell actually can pick and choose to a certain degree what its modificational targets will be at any given time. I'm not implying intention here- its all automatic- but a sufficiently complex multilevel hierarchical dynamical system optimized to dam holes in the dyke may start to seem like it has intention.
If you want to invoke the butterfly effect (which isn't what you were getting at but I couldn't resist- and if Laura Dern would lend me her hand now I'd be most gratified!) maybe that might start to resemble randomness because we have no access either to initial conditions or a God-like overarching perspective (but watch out for quantum and relativistic effects). That supernova over there in Orion a couple of thousand years ago may turn your grandchildren blue as the cosmic rays reach their mark- but it is not likely to turn them into bats.
Now it may be (trying to give this post some FUNKNET-relevant linguistic content) that similar-flavor system regulatory effects happen historically intergenerationally as well as during the lifetime of a language user. Such things as regular sound changes, some of which act together (chain effects, series effects, such as in Tai-Kadai tones, etc.) come to mind. Some sort of matrix like thing going on, kinda sorta? What about at higher levels hierarchically? H/D order harmonies, for instance, or prosody? Never 100%, but interesting nonetheless. And often competitive to boot. I read years ago that over phylogenetic time different groups/populations of nerve cells have vied for physical hierarchical position and concomitant resources/connectivity in the nervous system- akin to what gets recapitulated in ontogeny. Something like this go on in language acquisition/processing?
Now your added bonus feature:
The genetic code itself has evolved to minimize the impact of point mutations while at the same time giving enough variety of translation products- it is 'degenerate'. Because of this some mutations give the same peptide unit
But it goes further- the entire system of 'cubies' in the 4x4x4 matrix is arranged such that even when the coding shift doesn't give the same unit, it often gives a near equivalent (in terms of relevant side chain chemical/physical or torsional properties). Ok so I don't have oranges today- here's a tangerine...
It also turns out (unpublished work) that when you slightly (and motivatedly..) rearrange the order of nucleotides on the matrix axes the system becomes even more balanced and geometrically clean- with the cube center-crossing diagonals representing opposing properties such as positive or negative charge, shapes and sizes of side chains, etc. All the known code variants (in archaea, for instance) seem to take advantage of this design, as do the stop/start signals. Oops, did I call it a design? Pattern, how's that?
In a purely arbitrary system one would not see such well-balanced patterning. Was this system designed? Do I think so? Not hardly! But given the bits and pieces available and reproducible at the time, and the larger physical context of early earth, this is the optimal coding system currently surviving life hit upon- there may have been others.
I guess in the end it really all boils down to issues of generativity/production versus filtering/selection, and how much the various levels of System have over each- where, when, how much, fidelity, etc. I say biological systems have far more control than you think they do- that we are NOT merely kluges. Whether we have control over overall directions of change ('drift') is an open question.
More information about the Funknet