Evolution Not Random

[Belz...]

No. I'm saying that a mutation is an event, and selection is the outcome of a series of events.

Sometimes, we speak of death as a selection event. We could speak of birth or mating as a selection event, but that isn't really accurate. Unlike mutation, there is no specific moment when "selection" happens. After a long time, we can say which mutations were selected, and we can point to events (births, deaths, matings) that contributed to the selection, and we can say to some extent how a particular mutation contributed to its own selection.

I would think that you could consider succesful reproduction as a selection event.

 

[Belz...]

I was a genetic counselor, and this is a little hairy-- [FONT=Verdana, Arial, Helvetica, sans-serif]although the adjective frequently applied to the process of "mutation" is "random", the probability of genetic change is not identical for every nucleotide in a genome. DNA sequence affects DNA structure. There are hot spots for crossing over and methylation and places that are highly conserved and very resistant to mutation. Moreover, although it appears to us that mistakes just happen--that does not mean there is not a physical reason for the copying errors (radiation for example--older eggs and sperm have more point mutations and nondisjunction of chromosomes.)

The only thing I would add to this are the more recent discoveries that much of the supposed random mutations are not quite as random as first thought. Evolution would of course had to have been random initially. But something else to consider here is that natural selection selects variability. Then given more time, controlling those mutations has an advantage so mutation control mechanisms are selected.

Then if you wanted to get even more speculative, (though still evidence based) natural selection selected organisms which exerted control over the direction of evolution of other organisms. And now natural selection has selected organisms that are in the early stages of beginning to very specifically control their own evolution. So natural selection moves the organisms toward less and less randomness. Again, not ideology, it is what the evidence reveals.

I just love smart women...

 

[Belz...]

It seems that people are having a hard time understanding the difference between a chaotic systems and random (or stochastic) systems.

By your definition, Mijo, everything is random. Perhaps you don't understand your own posts.

Ever wonder how stem cells seem to differentiate between different types so efficiently to form organs, and that we are therefore not just a random homogenous blob of various and sundry cells? Must be some kind of cooperation/plan going on, huh?

No. You're wrong, Iamme Tumble.

 

[Meadmaker]

I would think that you could consider succesful reproduction as a selection event.

It's an event. That particular event may or may not happen, based on all sorts of external circumstances. I cannot look at an organism and, based on its genome, determine whether or not that particular "selection event" will happen. On the other hand, I can use that genome to give a probability that such an event will occur. In other words, since probability is involved, the events are random.

 

[Meadmaker]

By your definition, Mijo, everything is random. Perhaps you don't understand your own posts.

If I may interject, everything that involves the physical world does indeed have a random component, per mijo's definition. Technically, the position of the Earth is partly influenced by the particular quantum states of some gas atom in the stratosphere.

The question is whether or not it's useful to consider it. I can make all sorts of useful predictions in the field of celestial mechanics without ever resorting to calculations involving probability.

When it comes to evolution, I can make some rather vague, general, statements about evolution without resorting to probability. I can say "organisms will grow more adapted to their environment". However, if I want to say anything useful numerically, I'll have to use random numbers.

Back in that other long thread on the subject, I asked for an example of any evolutionary model that made a useful prediction and didn't involve a random number. I don't recall ever seeing such an example.


Now that doesn't mean that "evolution is random" is the best description for all occasions. It's far too simple to be that useful.

 

[sol invictus]

Back in that other long thread on the subject, I asked for an example of any evolutionary model that made a useful prediction and didn't involve a random number. I don't recall ever seeing such an example.

Start from some creature. It's modeled by a bunch of attributes. Now give it offspring, and each child has slightly different attributes, determined by some deterministic rules - for example child 1 has attribute 1 increased, child 2 has attribute 1 decreased, child 3 has both 1 and 2 increased, etc. Each child then has offspring, which follow the same deterministic rules as the first generation. Then also at each stage you apply a fitness cut - less fit individuals have fewer or no offspring.

Let that go, and you'll end with a population of fit creatures, which (if the offspring rule is comprehensive enough) will probably be identical to the results of a random model.

 

[sol invictus]

There's another aspect of this that really hasn't been touched on. If we use mijo's naive definition, every real process is random. But we could try to make things more interesting by asking how random. Something totally unpredictable, like whether a fair coin flips to heads or tails, is very random, but whether it lands on its edge is not very random. But it's the same coin... so not only is everything random, but how random it is depends on which question you ask, which aspect you're interested in.

I think that's part of the problem here. Some aspects of evolution - like whether a bird with green feathers and a red beak will evolve - are very unpredictable starting from primordial soup, but others - like whether eyes or wings will evolve - might be rather predictable.

 

[Meadmaker]

Let that go, and you'll end with a population of fit creatures, which (if the offspring rule is comprehensive enough) will probably be identical to the results of a random model.

But there are a few problems with the model. First, the only way you can make the deterministic offspring rulel "comprehensive" enough is to include a large sample with all possibilities considered. In large populations, it will be indistinguishable from the random model.

However, at small populations, the results will indeed be different. The deterministic model will show inevitable convergence toward the population of fit creatures, with the fit creatures showing specific, predictable, traits. The random model will show probable convergence toward fit creatures. Depending on the specifics of the model, the random model might show extinction, or it might show something "fit", but with a range of possible fitness, i.e. there are several possible "fit" creatures that can fill a niche. The random model will show different possible creatures filling the niche.

Which one is more accurate?

In practice, when creating numerical models, biologists use random numbers because those models are more accurate than deterministic ones.

 

[jimbob]

There's another aspect of this that really hasn't been touched on. If we use mijo's naive definition, every real process is random. But we could try to make things more interesting by asking how random. Something totally unpredictable, like whether a fair coin flips to heads or tails, is very random, but whether it lands on its edge is not very random. But it's the same coin... so not only is everything random, but how random it is depends on which question you ask, which aspect you're interested in.

I think that's part of the problem here. Some aspects of evolution - like whether a bird with green feathers and a red beak will evolve - are very unpredictable starting from primordial soup, but others - like whether eyes or wings will evolve - might be rather predictable.

Thinking about whether a system is significant or not has set me thinking about the normal distribution. This is a classic example of a system that can be described as a predictible part, and random effects, lots of independent interactions that create the classic bell-shaped curve.

Is it random? It depends on what you are wanting. The V2 attacks on London were inaccurate, but there was no defense against them except passive ones. IIRC, the first few fell a little short, so it was decided to play up reports of those that fell long, and play down reports of those that fell short.

The aiming point did shift further short. Whether a V2 hit Southern England wasn't random. Whether it hit a particular house was.

It depends on which you consider significant...

 

[Skeptic Ginger]

In doing a bit of research looking at this thread argument I found this interesting observation.

Order in Spontaneous Behavior

According to Laplace, randomness is only a measure of our “ignorance of the different causes involved in the production of events.” [1]

Brains indeed do throw the dice–but by refuting the notion of stochasticity our results imply that they have exquisite control over when, where and how the dice are thrown [86].

[1]Laplace PS (1825) Essai Philosophique sur les Probabilités. Paris: Gauthier-Villars.

[86]Barinaga M. Neuroscience: Neurons Put the Uncertainty Into Reaction Times. Science. 1996;274:344–340. [PubMed]

And from Wiki on randomness^WP

Some mathematically defined sequences, such as the decimals of pi, exhibit some of the same characteristics as random sequences, but because they are generated by a describable mechanism they are called pseudorandom. To an observer who does not know the mechanism, a pseudorandom sequence is unpredictable.

 

[jimbob]

However Skeptigirl, the best guess from most physicists is that quantum mechanical events are truly random. To grossly simplify an example, it isn't just that in principle we can't know a particle's position and momentum, but that the particle itself can't either...

(excuse the anthrpopmorphism).

This is what Sol Invictus and Frederik have been saying in a side part. (ETA of this thread)

 

[Meadmaker]

Not to put down Laplace, but he isn't exactly on the cutting edge of science.

ETA: More than that, he came from an era where discovering the mechanisms of the world was considered a main goal of science. I can't recall who said, during that era, that if he only knew all the initial conditions, he could predict all the events of the future.

Quantum mechanics and chaos theory proved he was wrong.

ETA: It was Laplace himself who said it

"We may regard the present state of the universe as the effect of its past and the cause of its future. An intellect which at a certain moment would know all forces that set nature in motion, and all positions of all items of which nature is composed, if this intellect were also vast enough to submit these data to analysis, it would embrace in a single formula the movements of the greatest bodies of the universe and those of the tiniest atom; for such an intellect nothing would be uncertain and the future just like the past would be present before its eyes."

 

[sol invictus]

But there are a few problems with the model. First, the only way you can make the deterministic offspring rulel "comprehensive" enough is to include a large sample with all possibilities considered. In large populations, it will be indistinguishable from the random model.

Maybe, maybe not, but in any case it is not random by any definition, and it is a reasonable model for evolution (albeit maybe not the best one).

In practice, when creating numerical models, biologists use random numbers because those models are more accurate than deterministic ones.

Totally with you there.

 

[mijopaalmc]

Maybe, maybe not, but in any case it is not random by any definition, and it is a reasonable model for evolution (albeit maybe not the best one).

Except the large-sample orderly behavior of random systems has everything to do with their being random and nothing with their being deterministic.

 

[sol invictus]

Except the large-sample orderly behavior of random systems has everything to do with their being random and nothing with their being deterministic.

Huh?


Incidentally, do you have an answer to my question yet?

Please provide one single example of a physical process (not a mathematical abstraction) which is not random by your definition.

 

[mijopaalmc]

Except the large-sample orderly behavior of random systems has everything to do with their being random and nothing with their being deterministic.

Huh?

It means exactly what it say, namely, the mathematical definition of of randomness allows for the behavior that Meadmaker was describing (know as convergence of random variables) precisely because random variable a random (in the sense that they are defined on probability measures).

Incidentally, do you have an answer to my question yet?

Please provide one single example of a physical process (not a mathematical abstraction) which is not random by your definition.

I have answered your question; you just didn't like the answer. The variation in the final conditions can be derived from the variation in initial conditions, whereas the variation in the final conditions cannot be derived at all from the variation in initial conditions, because there is often no variation in initial conditions. For instance, identical twins are genetically identical but their appearances can differ in ways that cannot be derived from their identical genetics.

 

[jimbob]

Does anyone here actually disagree with any of the following statements?

1) Over geological timescales, random events have significantly altered the course of evolution, so that there was nothing inevitable about the origin of humanity, or indeed anything akin to humanity. In that respect, the evolution of humanity was just one of many possible occurances. This does not make humanity "special" any more than it makes cockroaches "special".

2) Adaptations to a particular environment are inevitable, providing something can reproduce in that said environment; what form the adaptations will take is not inevitable. Saying that, some adaptations have such a strong selective advantage that they are very likely, the classic example being sight, and its convers the losing of sight in cave-dwellling animals, due I suppose to the (obviously non-marginal) energy cost associated with it.

3) Evolution is most rapid when an environment is changing, or has recently changed; this is because the "normal" organism is further from the "optimum", so a change from this is morel likely to be "better". If the organism is fairly well optimised, any change is more likely to move away from the optimum.

4) Organisms, as they evolve, change the selective environment for other organisms, so whichever adaptation occurs first (due to random mutations) will alter the subsequent course of evolution. This means that in a "new" environment (e.g. after the KT impact), the initial direction of the evolution of the surviving organisms as they fill the many vacated niches, will be more random than in an environment that has ben stable for a long time.

 

[sol invictus]

It means exactly what it say, namely, the mathematical definition of of randomness allows for the behavior that Meadmaker was describing (know as convergence of random variables) precisely because random variable a random (in the sense that they are defined on probability measures).

Yeah... and?

I have answered your question; you just didn't like the answer.

Really? Where!? Not in this thread!

Celestial mechanics, fluid dynamics, ballistics.

Hmm... those sound like mathematical models, don't they?

So the equations for the N-body system in which the only force acting upon the masses is gravity does not fully describe celestial mechanics?

"N-body system...only force...is gravity" - hmmm, not exactly a real system, is it?

Look my point is that, at mesoscopic scales where matter consist of >10¹⁰ particles, the quantum level randomness averages out, giving the appearance of deterministically predictable systems. Similarly, when a population is large that the inverse of the mutation rate, the random effects of drift, mutation, and selection averages, giving the appearance of deterministic behvior in evolution.

Oh, so physical systems are random after all!

I said the way celestial mechanics, fluid dynamics, and ballistics are most commonly and completely described is deterministic. Last time I checked, the Navier-Stokes equations were deterministic, as were Kepler's Laws and Newton's Laws.

[my emphasis]

So it was idealized mathematical models you were talking about all along?

I have answered your question; you just didn't like the answer.

Make up your mind - I'm getting dizzy.

 

[sol invictus]

Does anyone here actually disagree with any of the following statements?

I agree with 1) and 2). 3) and 4) are probably correct too, but I'd quibble with the wording a little (particularly 4))

Incidentally - has anyone tried to connect the punctuated parts of evolution to periods of rapid environmental change? I mean it's pretty obvious that a big comet that wipes out 90% of species will have a major effect on evolution for a while, but have people tried to see that in more detail in a model, or what the effects of rapid but less catastrophic changes would be?

 

[Belz...]

For instance, identical twins are genetically identical

Unless there's a mutation during the initial division...