Wednesday 10 September 2014

Reconceptualizing natural selection as primarily concerned with counteracting entropy (e.g. mutation accumulation)

Reconceptualizing natural selection as primarily concerned with counteracting entropy (e.g. mutation accumulation)

Michael A Woodley and I have developed a new conceptualization of the fundamental process of natural selection. It draws on diverse perspectives such as a couple of years of conversations about the cause of decline in intelligence (g) since Victorian times (especially the importance of mutation accumulation due to the massive decline in child mortality rates -, the work of WD Hamilton on the evolution of sexes, Graham Cairns-Smith's work on the origins of life, an old paper of mine on 'endogenous parasitism', recent discussion on the implications of the 'mouse utopia' experiment and more.  


The essence of the idea is that replication is not the main problem for living entities, indeed replication can be taken for granted.

The big problem for living entities is entropy; the main answer to entropy is natural selection, and the main anti-entropic mechanism of natural selection is massive overproduction of offspring with various means of selectively culling the most entropically-damaged offspring.

In terms of genetics; this could be understood in terms of saying that something like mutation-selection balance is the essence of natural selection


On Graham Cairns-Smith's conception of life, natural selection is built-into reality – NS is simply a part of the world – certainly part of chemistry (e.g. crystal propagation), probably physics too. In evolutionary terms, there is no dividing line between organic and inorganic, alive and dead.

Therefore, Life (in the sense of replicating entities subject to natural selection) is to be found pretty-much everywhere – not just in biology.


So the main problem for a living entity is not to make copies of itself, replication can be 'taken for granted' as a kind of universal phenomenon, but the main problem is to counteract entropy.

Entropy, that is to say random damage to replicating entities from many sources and copying errors during replication, is inevitable. Indeed, entropy will tend to degrade any identifiable structure, and any form of organization. Over time, the tendency is for all structure and organization to be returned to randomness.

This means that all lasting structures and all types of organization must overcome entropic degradation.

So, any actually-observable entity has already solved the problem of entropy to the extent that it is indeed observable! Any structured or organized thing which exists has solved the problem of entropy such that it at least came into existence, was sustained long enough to be observable by us, and - unless it is unique - has some mechanism for making more-of-itself: for replicating.


Furthermore, entropy affects all replication - so there are errors in replication that enter in the transmission of information between parent and offspring.

In a nutshell, this means that there is an unavoidable, intrinsic and cumulative entropic tendency for the fitness of any naturally selected lineage to decline to zero - to extinction, to non-life.

An example of this would be the tendency for mutations to occur in each parent organism, to be transmitted from parent to offspring - with new mutations occurring during the replication process, and for such mutations to accumulate generation upon generation until extinction.

Indeed this is not just additive accumulation, but there is a tendency towards a positive feedback cycle, in which mutations damage functionality which leads to more and additional, and uncorrected mutations.

So, the suggestion is that the fundamental problem for any entity is not replication, but combating entropy.


One implication is that the basic function of a molecule like DNA is not a matter of achieving replication - because replication would already have been happening, and can be taken for granted; DNA (and its evolution) is primarily about a mechanism of more-accurate/ less error-full replication.

So, the main question for living things is: how is entropy controlled?

And the main answer is: Natural Selection.

The context of Natural Selection (NS) is thus massive over-production of offspring (and spores, seeds, ova, sperm etc), and the strongly-selective reproductive-culling of offspring to eliminate accumulated entropic damage (such as mutations).

Thousands or millions of offspring (etc) may be generated, and selectively eliminated.


So, the context of the intrinsic decline of fitness in all replicating entities means that the main thrust of evolution by natural selection is simply to maintain fitness - to prevent extinction from intrinsic entropic tendencies - and not to improve fitness, nor to evolve adaptations.

So, this is a Red Queen phenomenon (in which there is running fast just to stay in the same place). Natural selection is necessary to maintain fitness in the face of the entropic tendency for fitness to disappear.

In other words, the phenomenon described as mutation-selection balance is not a specific, contingent, occasional circumstance: but the normal and indeed primary nature of natural selection as it applies to a genetic organism.

The genetics of NS is not primarily about evolving new genes but primarily about preserving from (entropic) destruction what are already-successful genes. It is about preventing the intrinsic tendency towards corruption/ degradation of an already- known-to-be-successful genetic recipe.


In sum, NS happens in what are ‘already fit’, already replicating, massively-over-reproducing, entities. Lacking which, fitness inevitably regresses to zero.

So – the main thrust of evolution (NS) is to maintain fitness. (A Red Queen sort of thing.)

The usual method for combating entropic damage is massive overproduction of offspring, and therefore Disposable Offspring.

The usual, background situation was that replication was not a problem, and sufficient offspring survival could be taken for granted in the immediate short –term – the problem was the distal long term of a few generations ahead at the point when the tendency was for mutation accumulation to destroy fitness.


Short term fitness, over the next few generations, was not the major problem - since there was such over-production of offspring; therefore long-term fitness beyond the next few generations is THE major problem.

Therefore, because in this conceptualization, natural selection implicitly looks-forward several generations,  so this is not about the single organism and its fitness but is instead a 'group-ish theory' kind of selection process.


With such a concept, it is trivially easy for individuals to ‘sacrifice’ their own fitness to some degree, or even the fitness of the immediate next generation or two - when the longer term fitness of the group of descendants is significantly enhanced. (This is a consequence of short-term replication being 'taken for granted' due to the context of massive over-production of offspring.

There is a very low cost to ‘adaptations’ which somewhat lower individual fitness if there is a fitness advantage in the next few generations – because the next few generations are almost guaranteed – they are not the big problem. The big problem is entropy, hence mutation accumulation – and that takes a few generations.

A pay-off two or three generations down the line is therefore almost-directly selected for, in the sense that the short term costs are trivial in the context of massive overproduction and a world of replication-not-a-problem.


Sexual reproduction was, by this account, relatively easy to evolve; because it enabled better control of entropy (see the work of WD Hamilton - but replacing/ adding to 'parasites' with spontaneously occurring entropic damage), e.g. the purging of mutations, the purifying of the gene pool of second, third etc –generations of offspring.

What gets naturally selected is therefore fitness down-the-line.

In a sense, the proximate locus of natural selection is a few generations ahead; specifically the future generations whose fitness would have declined to zero absent the operations of natural selection.

(This is different from the main emphasis of the conceptualization of mainstream selfish gene theory - which only very seldom allows for the possibility of long-term 'group' fitness advantage overcoming a significant short-term fitness disadvantages. In our view the short-term disadvantages are trivial in effect in a context of 'replication taken for granted' and the usual situation of massive overproduction of offspring.)


The new conceptualization of NS implies that ‘competition’ with other living things is mainly about preventing the accumulation of entropic damage. Competition with other living things, including other members of the same species, is primarily about the purging, purifying or culling of a large majority of offspring - as the primary method for removing what would otherwise be fitness, lethal accumulations of mutations.


Note added 11 Sep 2014:

New concept of Natural Selection: RES
1. Replication
2. Entropy
3. Selection

1. Replication is taken as given, 2. entropy tends to degrade structures and organization to stop replication (extinction of lineage), 3. natural selection controls the entropy - lineage is maintained.

By contrast -

Traditional concept of Natural Selection: ERS
1. Entropy
2. Replication
3. Selection

1. Entropy generates variants 2. some of which replicate, and 3. some of these undergo natural selection to expand and create a lineage.


Note added 18 September 2014

Another way of conceptualizing this is to regard 'mutational meltdown' as a universal process, which always threatens extinction  - and therefore requires mechanisms and process to overcome this intrinsic tendency.

Mutational meltdown was first described as a threat for small populations of asexual organisms; later this was widened to sexual organisms and then to large populations - so mutational meltdown has gone from being a specific case to probably a universal possibility.

In effect, I am suggesting that the primary functional necessity for living things is avoidance of mutational meltdown - and all actually observable living things have solved this problem to the extent of their lineage surviving long enough to be observable.


Bruce B. said...

Greg Cochran mentioned something about this in a comment on his blog recently. He mentioned it like is it was obvious, not some new discovery.

Bruce B. said...

Here’s the Cochran line (characteristically memorable):

“This notion that ongoing natural selection is not the default – that it only happens on national holidays or whatever – is fairly common among biologists. Obviously untrue, because you can’t even have things stay the same without ongoing selection – otherwise mutations and drift would gradually ruin everything. Only selection lets horseshoe crabs outlast mountain ranges”

Bruce Charlton said...

@BB - Yes, and I was banging on about it on that blog in June 2013!

Nicholas Fulford said...

And the extension of your thesis is that inhibiting Natural Selection through enabling naturally non-viable offspring to live long enough to reproduce accumulates mutations that should have been culled leading to greater entropy and less fitness further down the road.

Add to this the fact that the naturally fit - those who would not have been culled by Natural Selection - are overwhelmed by the reproduction of the less fit, and a recipe exists for a "hard correction" at some point down the line.

Reproduction by the less fit also leads to a resource drain as more effort is spent to allow the less fit to live and live longer, and breed, et cetera.
Add to this the tendency of many who are fit to put off or not reproduce, and the species becomes even less fit - though sustained by artificial means.

I sense a positive feedback loop here, and not a nice one.

Dennis Mangan said...

Why wouldn't a better DNA repair mechanism, for example, be a more thrifty way of ensuring against entropy? The culling of masses of offspring when a few extra DNA repair systems (enzymes) would do the job seems far more wasteful, and DNA repair more fitness producing. Organisms already have these mechanisms, so doubling up on them wouldn't be difficult.
(This is all assuming that I've understood your argument correctly.)

Anonymous said...

It seems like it is essential for people in advanced societies to employ advanced medical technologies like genetic engineering to slough off deleterious mutations every few generations as well as correct current deleterious mutations that cause disease.

Benjamin I. Espen said...

Have you considered a simulation that might flesh out this idea? One might be able to construct a model to illustrate the concept.

Adam G. said...

Have you come up with a mathematical way to express this insight? There should be a way of formalizing it.

Bruce Charlton said...

@satndingagainst... In volume 2 of The Narrow Roads of Gene Land, WD Hamilton actually tackles the full implications of the rather vague general idea of correcting mutations using some kind of technology. It is in the first place much more difficult (and a much more remote possibility) than is usually recognized. Secondly, because of the scale of the problem, its consequences are far less acceptable than many would accept. For example, the human genome becomes, in effect, the property of the state - and in practice would be frozen at one particular variant so people would become clones - all new mutations would need to be eliminated because they *might* be deleterious in some unknown way.

@Mangan - That has a quantitative effect on reducing entropy/ culling some mutations, which is presumably why such mechanisms evolved. But none are 100 percent effective, plus of course the repair mechanisms are themselves subject to entropy - so you would need a repair mechanism for the repair mechanism (and so on...)

Mark Yuray said...

Yes! Absolutely perfect! Entropy is diametrically opposed to humanity, which is complexity, differentiation and life. We ARE the anti-entropy.

pyrrhus said...

@Mangan--Good point, but NS is an absolutely mechanical /mathematical system that is inherent in the physical Universe. So Nature has to operate within that entropic background , regardless.

Thomas said...

How do you square this with a loving Father? Why would he setup humanity to be dependent on high-infant-mortality (which seems terrible and tragic) to thrive?

Bruce Charlton said...

@Thomas - That is a valid question (although outwith science) - but perhaps ultra-high child mortality is to be seen in the context of nature as a whole, and its vast wastage and suffering.

About half of humans conceived die as zygotes, embryos or fetuses before they are even born (I mean naturally).

There is a primary theological sense in which dying is - indeed must be - the most important thing we ever do. My understanding is that - for some reason I cannot fathom - dying is necessary to spiritual progression. As a (theoretical) Mormon I believe our pre-mortal spirits were incarnated and then they die.

The problem of suffering is complex - some suffering is necessary some is good for us - but there is a residuum of spiritually-unnecessary and excessive suffering in the world. I believe that God cannot altogether prevent this - partly because of human 'free will' and partly because there are some natural processes (earthquakes, asteroid collisions etc) which 'just happen' - God is in the universe, the universe is not in God.

Those sufferings God cannot prevent God does heal. He cannot make them as if they had not happened, but He can - and does - wholly heal their ill effects.