Evolution is a non-random directed process

[ProfMoriarty]

The notion of Intelligent Design is often raised as a potentially acceptable alternative to evolution. The claim is often made based upon the following supposition:

– evolution is a random process, so how could it have produced the complex organisms we see today - but an Intelligent Designer could have intervened and directed evolution to produce all the world's creatures and man.

However, evolution is not a random process.

The issue I want to raise for discussion is:

"Darwinian evolution is a non-random directed process. "

This is strictly in accord with both Darwin's original discussions in On The Origin Of Species, and also with current evolutionary theory, and is what enables evolutionary theory to be able to explain adaptive change and speciation.

[otseng]

f(x, y) = A(y) such that

A(0)=x
and
A(n+1) = A(n) + sin(A(n))

For sufficiently large y (Infinite is ideal, but 10 or so is often sufficient for a close enough approximation), and 0<x<2*Pi (6.28....)

f(x. y) = Pi (3.14159...)

That is interesting and I did confirm it with my handy dandy Excel spreadsheet. But isn't restricting x to 0<x<2*Pi a limiting example? It seems like when x > 2*pi, the equation then becomes random.

Just a note, I realize my f(x,y) example is a highly simplistic analogy. And there probably could be be counterexamples to my generalization.

[ENIGMA]

f(x, y) = A(y) such that

A(0)=x
and
A(n+1) = A(n) + sin(A(n))

For sufficiently large y (Infinite is ideal, but 10 or so is often sufficient for a close enough approximation), and 0<x<2*Pi (6.28....)

f(x. y) = Pi (3.14159...)

That is interesting and I did confirm it with my handy dandy Excel spreadsheet. But isn't restricting x to 0<x<2*Pi a limiting example? It seems like when x > 2*pi, the equation then becomes random.

Just a note, I realize my f(x,y) example is a highly simplistic analogy. And there probably could be be counterexamples to my generalization.

When it is above 2*Pi then it will converge to the closest odd multiple of Pi to x, except when x is an even multiple of pi (within whatever computational margin for error), in which case it remains stuck at its initial position.

To get a less limited example:

f(x, y) = A(y) such that

A(0)=x MOD (2*Pi)
and
A(n+1) = A(n) + sin(A(n))

f(x. y) = Pi (3.14159...) when A(0) does not equal 0.

[ST88]

I'll get around to the details of this but it requires a bit of thought. However, the general concept is this:

The mathematics in this thread is making this liberal arts major's head spin. Before I make a fool of myself by making assumptions about this, allow me to start on the qualitative discussion. If I am understanding this correctly, "directed" evolution has something to do with the environment dictating which mutations get to survive and which ones fail. But more specifically, there is an optimal variation which gets to survive within each mutation generation. In other words, in the ABCD example, the capital letters were somewhat predetermined by the environment & the organism that happened to have this pattern is the one that was most likely to pass the survival test. Is this correct?

[ProfMoriarty]

Is this correct?

Yes, more or less, although of course both my argument and yours are gross simplifications of the real situation. In reality, as natural selection can select against all genes in the genome that would exhibit an adaptive effect, there are so many combinations that it would be impossible for a population to fill all of them.

But put simply, and to answer otseng's question, evolution is non-random because selection will always direct evolutionary adaptive change towards a local peak in the adaptive landscape. If the environment changes, so that different combinations of genes are more adaptively optimal, then the direction of adaptation will change to move towards this new combination. If mutation was not random then there would be no variability for selection to select for in the new direction.

Selection of course isn't a "thing". It is just the name for all of the environmental factors that put constraints upon the viability of an organism. So less well adapted organisms have a harder job finding food, mates, not getting predated etc., and will tend to pass their genes on to fewer offspring than better adapted organisms. So in terms of genetics selection directs populations towards having higher ratios of those genes which are currently able to provide the highest adaptive benefits, and in terms of phenotype etc. the population will be seen to become better adapted to it's environment over time.

As otseng says

Selection should not be able to affect in any way gene mutations/combinations. It can only choose from the resultant genes that were produced.

.

This is true. Selection cannot direct for particular mutations to appear. However, the mutations in the offspring are derived from the genome of the parents. For instance if a particular set of genes determines that the members of the population should all stand approx. 4 feet high, then the offspring would be expected to inherit variations which would make them 4 feet +/- a couple of inches as well. You would not expect the offspring to range from mouse size to elepephant size. So selection has a random choice to select from, but that random choice is based upon the results of the previous rounds of selections. In this way selection can indirectly influece the range in which those variations or mutations are able to occur.

If environmental conditions favour larger body size, the offspring of 4 feet + will have an advantage, however slight, and over time the entire population will move towards greater body size. If it favours smaller body size, then those of 4 feet - would have the advantage and the population would tend towards smaller body size.

Suppose that this population becomes split into two because a change in the course of a river imposes a barrier between them, and that on each side of the river the conditions favour different body sizes. Over perhaps 100,000 years the two populations could easily vary so much from the initial body size that if they were reintegrated through another change in the river they would be physically unable to mate. This would mean that although the geographical barrier was removed there would still remain a biological barrier, and this would keep the gene pools separate. In this case the two populations would have to be regarded as two distinct species. This scenario, which is called allopatric speciation is generally regarded as one of the most important engines of large scale evolutionary change.

[ST88]

Suppose that this population becomes split into two because a change in the course of a river imposes a barrier between them, and that on each side of the river the conditions favour different body sizes. Over perhaps 100,000 years the two populations could easily vary so much from the initial body size... is called allopatric speciation is generally regarded as one of the most important engines of large scale evolutionary change.

If I remember correctly, this is the reason behind the differences between old world and new world monkeys. That's quite a body of water that now separates them. No new world monkeys have opposable thumbs. I believe they are primarily arboreal -- vs. old world monkeys of which many are not arboreal -- so there is no selection pressure. However, orangutans are arboreal and do have opposable thumbs. As I understand it, the primary opposable-thumb function is use of tools, with inline thumbs more useful for climbing and swinging.

I realize that this is a vast oversimplification of a complicated problem. But I wanted to know if you more or less agreed with the idea that the opposable thumb was an example of this "directed" evolution.

Now, is it your contention that the current designs are the most efficient designs for the purposes that various evironments imply? Is the bilateral, four-legged design the most efficient design for land-dwelling, lung-breathing creatures? Is the bilateral, six-legged design the most efficient for inspirator air-breathers? Couldn't it be an accident that insects have three pairs of legs instead of, say, four?

[Jose]

Now, is it your contention that the current designs are the most efficient designs for the purposes that various evironments imply? Is the bilateral, four-legged design the most efficient design for land-dwelling, lung-breathing creatures? Is the bilateral, six-legged design the most efficient for inspirator air-breathers? Couldn't it be an accident that insects have three pairs of legs instead of, say, four?

I don't buy the idea that the current designs are the most efficient for their current environments. Rather, they are good enough to get by in the current conditions. It's important to remember that mutation can only change what already exists. Therefore, if the ancestor of land-dwelling, lung-breathing creatures was a bilateral, four-legged creature, then the land-dwelling creatures would inherit this basic body plan. Evolutionary changes to that body plan would be hard, since the bilateral symmetry is set up as early as the fertilized egg, and determines how embryology works. There has been some tweaking of the later stages of development, in things like the length of the tail (or internal coccyx, as in us), the presence or absence of limbs (snakes), or the use of the forelimbs as wings rather than legs (birds, bats). Perhaps a different body plan would have been much more efficient, but the basic material wasn't there to build it.

As for insects, it's useful to compare them to millipedes, crustaceans, and arachnids. These guys all have different numbers of legs. As we understand the molecular biology, the millipede-types were the ancestor. Mutations affecting segment-development genes resulted in the loss of legs from the posterior segments. With insects, we've come to 6 legs. Most winged insects have 4 wings, but in flies, the posterior wings have been reduced into a little nubbin. Maybe the last pair of legs will go sometime, too. Again, though, the starting material was bilaterally symmetric, with a particular body plan. The only possibility was to modify it.

So, I'd say it's partly accident, and partly the simple consequence of modifying what already exists.

[otseng]

When it is above 2*Pi then it will converge to the closest odd multiple of Pi to x, except when x is an even multiple of pi (within whatever computational margin for error), in which case it remains stuck at its initial position.

I think this math example is actually an appropriate illustration of this discussion. (Though of course biological evolution cannot be formulated into pure mathematics)

Let's take x as representing mutations. y as natural selection. And f(x,y) as the result of mutation and natural selection. In your first formula, f(x,y) has the property of converging to certain values after many iterations. But still, when x is random, the result of f(x,y) is still random, even though it is a limited result set.

This is my argument against evolution being non-random. The resut might be directed, but the result is still dependent on the random number x.

But put simply, and to answer otseng's question, evolution is non-random because selection will always direct evolutionary adaptive change towards a local peak in the adaptive landscape. If the environment changes, so that different combinations of genes are more adaptively optimal, then the direction of adaptation will change to move towards this new combination. If mutation was not random then there would be no variability for selection to select for in the new direction.

I'm having a hard time seeing the logic. I can agree that selection can have a directive effect. But, it seems like what you are saying is that this directive effect also is able to remove the random component of mutations.

Mutations would have no idea what has a higher/lower chance of surviving in an environment. It would be a purely random event. Because of this alone, it would make the final product a result of a random occurence.

Let me ask also, what are the implications if evolution is not random?

[perfessor]

... when x is random, the result of f(x,y) is still random, even though it is a limited result set.

This looks self-contradictory. If the results were random, all outcomes would be equally likely. The selection pressures ensure that this is not the case by placing constraints on the set of results (i.e. the individuals who are lucky enough to survive and reproduce). Unless you have a different definition for "random".

Mutations would have no idea what has a higher/lower chance of surviving in an environment.

True.

It would be a purely random event.

False! Again, the selection process is non-random, so the result is "ordered" - that is, specific traits which were randomly distributed among even a tiny portion of a population, could end up being predominant in the gene pool after a suitable number of iterations.

Let me ask also, what are the implications if evolution is not random?

I'm going to answer this as though you had said, "...if mutations are not random", because I would say that evolution is in fact not random. But if mutations were directed, it would be evidence for some sort of design - it would indeed be interesting, if it could be shown.

[Jose]

Take a peek at the simulation I mentioned earlier. You don't have to do it, because there are links to a couple of filled-in sheets. If you read the "rules," you'll see that mutation is random. All that changes is the "selection pressure." This should provide a glimpse into the mechanism: even with random mutation, selection can give rise to apparently-designed outcomes.

Take a peek at the simulation I mentioned earlier. If it doesn't answer the question, what do you think are the reasons?

[otseng]

... when x is random, the result of f(x,y) is still random, even though it is a limited result set.

This looks self-contradictory. If the results were random, all outcomes would be equally likely.

I think the problem is a lack of a definition of random.

I think of random as not being able to predict the outcome with certainty. When something is random, you can only give the odds of a certain outcome. When something is not random, you can definitively predict the outcome.

So, in the mathematical formuals previously given, there is no way to predict what f(x,y) would be without knowing x.

However, here is an example where you can predict f() without knowing x: f(x,y) = 1. No matter what x is, you can predict the outcome with certainty.

The selection pressures ensure that this is not the case by placing constraints on the set of results (i.e. the individuals who are lucky enough to survive and reproduce).

Having a limited result set doesn't prove that it is not random.

Suppose f(x,y) = sin(x*y). The result set is limited, but it is still random if x is random.

Again, the selection process is non-random, so the result is "ordered" - that is, specific traits which were randomly distributed among even a tiny portion of a population, could end up being predominant in the gene pool after a suitable number of iterations.

Is there an objective criteria for which the result is ordered by natural selection? How can one determine what would cause something to survive or not? If one can objectively do this, then I can see how it can remove the random component of mutations. In order to do this, one must be able to show that only certain mutations will result in a chance of survivial in a given environment prior to a mutation occuring.

So, for instance, someone gives me a coin one-by-one. I have made a rule that I will only keep quarters. The other person continually gives me a coin randomly - a penny, a dime, a quarter, a nickel, a nickel, and so on. By the end of the day, I'll only have quarters. What I end up with in my pocket is not random because I have made an objective rule to select what I want. The rule is clear and is defined prior to receiving the coins.

Suppose I decide that I will keep what I like at that moment of receiving the coin. Could be a penny, could be a nickel, depending on my mood that second. What I would end up with in my pocket would be a random set of coins.

So, in order for evolution to not be random, one should be able to predict what the final result would be without knowing what will result from a mutation.

I'm going to answer this as though you had said, "...if mutations are not random", because I would say that evolution is in fact not random. But if mutations were directed, it would be evidence for some sort of design - it would indeed be interesting, if it could be shown.

I would agree that if mutations are not random, then it would be evidence for design.

However, I was more questioning the entire process of evolution being non-random. If evolution (mutation + selection) is not random, what are the implications?

simulation I mentioned earlier. If it doesn't answer the question, what do you think are the reasons?

Unfortunately, I cannot read it since my firewall blocks port 16080. The firewall only allows the standard http port 80.