Evolution thoughts: directed evolution

[Eos of the Eons]

The biq question is how the organism gets certain cells to do things that are not in the immediate interest of those cells. Cancer, methinks, is when this doesn't happen as planned, i.e. the cells no longer have the greater interest of the organism "at heart."

First look at the causes of cancer. At the end of each cause is a gene that is no longer coding right. All the back up systems have failed as well. You have mutation, radiation, viruses, etc. The organism doesn't get certain cells to do anything when it comes to cancer, right...the cells are no longer growing right in spite of all best efforts to prevent that.

Some people are more prone to cancer because they are doing their bodies no favors by inactivity or a bad diet. Your body needs health in order to fight off and prevent mutations resulting in cancer. Others people may be more prone to cancer because they don't have as good back up systems (melatonin in the skin) or whatever.

That is how cancer can run in the family or not...depending on the cause. If it is genetic you will see that run in the family.

Skin cancer is interesting...if very white skin runs in the family you will see a higher incidence of skin cancer because you don't have the natural protection that darker skinned people have. This is genetic because of the white skin, but also you have to have the exposure to the sun's radiation ultimately causing the mutation that leads to cancer.

Moles and other skin 'problems' don't necessarily require exposure to radiation...so those causes of cancer will run in the family as well.

Now, you get viruses like HPV and it's not genetic. You can't pass cancer caused by HPV onto your offspring. Maybe some kind of behavior linked to personality will get passed on to make a person more prone to getting HPV though. Or it may just be learned behavior as well.

What I'm trying to say is that environment and genetics play out in what ultimately causes something to be bad or good and have it passed on or not to the next generation.

If a certain personality (recklessness) trait is passed on, then a person may die before they have children. However, if recklessness causes a person to be 'braver' and can fight off a lion, then they will survive to pass on their genes better than a guy who will be petrified with fear.

But if you dye your hair blonde, then your kids won't be born blonde. If blondes do have more fun though, and children learn to dye their hair, then there will be more children as a result of successful breeding becaue of a learned behavior.

So, evolution will drive the survival of the fittest, no matter what type of fitness helps a person to survive. It may be a trait a person is born with-physical or mental, or it may be learned behavior as a result of intelligence you are born with. Humans don't survive by having nails to fight off predators.

It's all very complex. But, you cut off a dog's tail...that's not genetic. Genes must be involved in the evolution of desirable traits. If a dog with no tail survives better, then its offspring that don't have their tails cut off will end up dying off. Most of the 'evolution' that humans drive in plants or animals will cause the organism to die off in nature. How successful will a seedless watermelon be in the wild?

 

[c0rbin]

The immune system example is interesting. Now figger a way to pass that to the next gen!

I hear tell that breast feeding plays a large part in this.

 

[BillHoyt]

I hear tell that breast feeding plays a large part in this.

In a limited way, yes. It confers limited immunity to the infant for some diseases for about 6 months or so. After that, large proteins can no longer pass through to the bloodstream intact. From that point on, the infant's immune system is on its own and the monoclonal antibodies must be built based on the disease challenges encountered.

Cheers,

 

[BillyJoe]

About the closest thing to a mechanism for directed evolution is the Baldwin effect. This applies to organisms in which learned behavior is a significant factor. It basically says that in such organisms, some aspects of behavior are determined at run-time, rather than at design-time (sometimes called 'phenotypic plasticity').

So if there is some really cool behavioral adaptive peak nearby in design space, any lucky individuals who start out with factory settings that happen to be closer to the combination that produces that adaptive behavioral trick will be more likely to stumble on it within their lifetimes, at which point (though not before) it becomes visible to selection. Subsequent generations will then start out with even fewer connections that require setting, and so will have an even better chance of hitting the right combination sooner, etc.

By partially side-stepping the combinatorial explosions that tend to occur around the number of connections involved in complex behaviors, even a slight amount of help from this effect could streamline the process considerably (but probably doesn't go far enough to qualify as 'directed' -- a concept which cannot possibly be meaningful beyond the scope of local optima anyway).

You seem to be saying here that the Baldwin effect is non-Lamarckian although it certainly looks Lamarckian. I would agree. I bought this up in another thread a while back but it was immediately labelled Lamarckian by another poster who I think was BillHoyt. But he has not jumped on you so I am probably wrong about that. But then again you have explained the Baldwin Effect better than I did so perhaps he's giving it a second thought.

BillyJoe.

 

[BillHoyt]

You seem to be saying here that the Baldwin effect is non-Lamarckian although it certainly looks Lamarckian. I would agree. I bought this up in another thread a while back but it was immediately labelled Lamarckian by another poster who I think was BillHoyt. But he has not jumped on you so I am probably wrong about that. But then again you have explained the Baldwin Effect better than I did so perhaps he's giving it a second thought.

BillyJoe.

BillyJoe,

Yes, that was me, and yes, I still stand by it. The "Baldwin effect" is not a useful construct for evolutionary theory. Baldwin originally proposed it long before the neo-Darwinian synthesis. It is still referenced in psychology today, but most notably in the computer sciences. Almost never (maybe never?) in biology.

The "plasticity" Baldwin refers to is a phenotypic characteristic. Any underlying genotype that creates a mores plastic phenotype would probably be favored evolutionarily, but it is still the genotype that is, ultimately, selected. Hence, the "effect" is a non-starter in evolutionary theory. It still comes down to the genes.

Cheers,

 

[Dymanic]

Originally posted by BillyJoe

You seem to be saying here that the Baldwin effect is non-Lamarckian although it certainly looks Lamarckian.

Right.

It isn't Lamarkian, because nothing the organism learns is explicitly passed to its offspring; but it looks Lamarkian, because the higher number of preset connections is. Selection keeps raising the bar, so the entire population moves toward the adaptive peak. It might be thought of as a 'scouting' of local design space by individuals in advance of actual changes to the genome.

Originally posted by Bill Hoyt

It is still referenced in psychology today, but most notably in the computer sciences.

Yes, that's where I encountered it.

Almost never (maybe never?) in biology.

Dang. I didn't know that; thanks for bringing me up to speed.
Is is completely useless in biology then? It seems like such a lovely idea.

 

[BillHoyt]

Dang. I didn't know that; thanks for bringing me up to speed.
Is is completely useless in biology then? It seems like such a lovely idea.

Many a lovely idea has been discarded along the way, I'm afraid. Baldwin proposed it as something different from Lamarck and different from Darwin. This was long before we understood the significance of genes and their essential role in inheritance.

The Baldwin effect claims that behavioral plasticity somehow paves the way for evolution. The problem is we now know that the genes control the behavioral repertoire. So the plasticity is a phenotypic characteristic; a result of the combination of the genes and the environment. Those with this plasticity had to have had the genes permitting the behavior in the first place, then. So it comes down to the genes again, and that comes down to neo-Darwinian selection.

Cheers,

 

[Dymanic]

Before I can proceed to the acceptance phase of the grieving process that the passing of this lovely idea will initiate for me, I guess I need to work through the denial phase.

Originally posted by Bill Hoyt

The Baldwin effect claims that behavioral plasticity somehow paves the way for evolution. The problem is we now know that the genes control the behavioral repertoire. So the plasticity is a phenotypic characteristic; a result of the combination of the genes and the environment. Those with this plasticity had to have had the genes permitting the behavior in the first place, then. So it comes down to the genes again, and that comes down to neo-Darwinian selection.

I don't see Baldwin as denying that genes control the behavioral repertoire; but accomplishing that means hitting the right combination among (especially for complex behaviors) absolutely astronomical numbers of possibilities, each (or most, or many) of which will produce a different behavior.

What exactly does it mean to say that those with this plasticity had to have had the genes permitting the behavior in the first place? Grasping for an analogy here, isn't that something like saying that because the alphabet provides the necessary plasticity, the complete works of Shakespeare are therefore explicitly contained within it?

I'm also struggling with the idea of plasticity being a phenotypic, rather that genotypic characterisic. I'm assuming that plasticity refers to the number of connections that are not preset, but left subject to environmental influence. Wouldn't that number be genetically determined?

 

[BillyJoe]

Bill,

The "plasticity" Baldwin refers to is a phenotypic characteristic. Any underlying genotype that creates a mores plastic phenotype would probably be favored evolutionarily, but it is still the genotype that is, ultimately, selected. Hence, the "effect" is a non-starter in evolutionary theory. It still comes down to the genes.

Ah, I see your point.

Yes, it all comes down to genes. Agreed. But the point about "plasticity" was that it "seems" to suggest evolution along Lamarckian lines but, in fact, it is non-Lamarkian precisely because it is coded for in the genes. With "plasticity" organisms seem to evolve within their lifetimes by changing with environmental change and then passing on these seemingly evolved characteristics to the next generation. It looks Lamarckian but it isn't.

But, okay, if Baldwin really did mean it as something other than Darwinism, then I guess we drop him into the basket.

regards,
BillyJoe

 

[BillHoyt]

I don't see Baldwin as denying that genes control the behavioral repertoire; but accomplishing that means hitting the right combination among (especially for complex behaviors) absolutely astronomical numbers of possibilities, each (or most, or many) of which will produce a different behavior.

Baldwin actually had nothing to say about genes. He wrote his paper in 1896, long before genes were known and a few years before Mendel's work began to gain wide recognition. The question is this: does the Baldwin effect recognize something different than neo-Darwinian theory.

What exactly does it mean to say that those with this plasticity had to have had the genes permitting the behavior in the first place? Grasping for an analogy here, isn't that something like saying that because the alphabet provides the necessary plasticity, the complete works of Shakespeare are therefore explicitly contained within it?

Baldwin posits a social/learning role in evolution. To be part of evolution, this learning must be represented as a difference in alleles. In other words, before the new behavior was "acquired", say 10% of the population had the alleles that permitted them to learn the behavior. If it becomes adaptive, then, over time, the allele's frequency must increase. We must be able to see it rise to 40% or 75%.

Let's try it as a reductio ad absurdum. Let us say this new learning immediately spread throughout the population, within a single generation. That means they were all already capable of "acquiring" it. Hence, there will be no evolutionary change in the next generation.

I'm also struggling with the idea of plasticity being a phenotypic, rather that genotypic characterisic. I'm assuming that plasticity refers to the number of connections that are not preset, but left subject to environmental influence. Wouldn't that number be genetically determined?

Baldwin's comments on plasticity are over 100 years old, and, therfore, in light of modern understanding, a bit muddled. His "plasticity" was along the lines of the old (and largely discarded) notion of the tabula rasa. He claims that plasiticity is a lack of preset behavior and that that is selected for. But unless adaptive behaviors are substituted for this lack of behavior, there is nothing to select for. Therefore, for the plasticity to be selected, something must have been learned. Although he framed it in genotypic terms, he was necessarily arguing that the phenotype gets selected. Not merely a blank-slate, but a blank-slate on which was writ something new and adaptive.

Cheers,

 

[Dymanic]

I wouldn't say that the Baldwin effect says anything that contradicts neo-Darwinian theory. The Question seems to be whether it says anything substantive enough to be useful, i.e., whether there is any advantage in the exploration of design space by flexible behavior in individuals over exploration of design space by variations in hardwired behavior at the level of the genome. If I'm going to let go of this idea, it needs to be based on its lack of consistency or substance, rather than reminders that its author knew nothing of genes or neurons (charges which could be leveled with equal validity against Darwin himself).

Spiders don't learn to build webs by trial and error, nor by watching other spiders build webs; web building is driven by a specific pattern of neural connections that come preset from the factory -- efficient, but inflexible.

Many young animals, ducklings being a popular example, 'imprint' on their mothers, making them easier to find among the chaos of a large herd by recording nuances of smell (or whatever). Some of the neural connections that govern this behavior cannot possibly be made until certain information from the environment is available, so the pattern is incomplete initially. But this plasticity is a one-shot deal; once the final connections have been made, they are essentially just as irreversable as the spider's entirely pre-wired set. In humans, one pronounced example of this effect is in language aquisition.

Perhaps the sets of connections which produce the behaviors in the spider and the duckling are little more than predefined nodes with specific ranges of acceptable values. The duckling's plasticity may amount to little more than optimization within the bounds of a very limited domain. Even the phenomenal language aquisition capabilities of young humans appears to follow this pattern.

Letting neurons 'free associate' makes possible a huge range of results, but mostly unpredictable results; at this point it is necessary to place another item on the assumption stack -- that relationships between a number of neurons can produce a virtually infinite set of possible behaviors (in the same sense that the alphabet produces the works of Shakespeare). So the only behaviors that can really be said to be specifically coded for would be the pre-wired ones. Now, maybe this wouldn't work; maybe randomly diddling the connections would be so unlikely to produce any adaptive behavior at all that at least the relationships (if not the specific values) have to be predetermined. But then, randomness at the level of that predetermination would be subject to the same problem!

So let's say that even in the more complex organisms, some degree of pre-wiring guarantees at least a minimal level of adaptive behavior, and some number of connections are left available for postnatal fine-tuning, resulting in organisms which adapt well to a wide range of environments (though maybe not individually very flexible about moving from one environment to another).

Now individuals venture forth with their preset connections set to various values, none of which being sufficient to produce an adaptive behavior which happens to exist in local design space, but some of which start out closer. A miss being as good as a mile, those individuals have no advantage over the others -- except that they are more likely to stumble on the right combination earlier in their reproductive lifetimes, at which point they begin to enjoy differential success. This gets reported back to the genome in the form of the greater numbers of their progeny (who inherit their preset connections). As those progeny come to dominate the population, the process repeats, the population always moving closer to the adaptive peak (paying a price in behavioral flexibility at the same time, btw -- the whole deal seems to be based on trading flexibility for more efficient hard-wiring).

It must be assumed that the adaptive trick is expressible using the available neural alphabet, so nothing says the others won't find it too -- just that more prewiring makes it easier to find (more likely to stumble upon). In fact, in animals smart enough to learn by imitation, an individual might not get much reproductive mileage out of a new trick before the others picked it up.

 

[Soapy Sam]

Oh Bill.
"Tabula rossa?"

 

[BillHoyt]

Oh Bill.
"Tabula rossa?"

Ooops. Thanks for the correction!

Cheers,

 

[BillHoyt]

Originally posted by Dynamic:
I wouldn't say that the Baldwin effect says anything that contradicts neo-Darwinian theory. The Question seems to be whether it says anything substantive enough to be useful, i.e., whether there is any advantage in the exploration of design space by flexible behavior in individuals over exploration of design space by variations in hardwired behavior at the level of the genome. If I'm going to let go of this idea, it needs to be based on its lack of consistency or substance, rather than reminders that its author knew nothing of genes or neurons (charges which could be leveled with equal validity against Darwin himself).

I mentioned genes only to indicate that Baldwin did not have this information at his disposal, and therefore didn't comment on them. I also agree that neo-Baldwinism, if you will, does not contradict neo-Darwinism. (His original papers, however, presented it as something different from both Darwin and Lamarck.)

Flexible behavior is clearly an adaptive advantage. I see no difference in our opinions there. What it comes down to for me, though, is whether there is anything left of Baldwinism to be meaningful in evolutionary theory terms.

Spiders don't learn to build webs by trial and error, nor by watching other spiders build webs; web building is driven by a specific pattern of neural connections that come preset from the factory -- efficient, but inflexible.

Agreed.

Many young animals, ducklings being a popular example, 'imprint' on their mothers, making them easier to find among the chaos of a large herd by recording nuances of smell (or whatever). Some of the neural connections that govern this behavior cannot possibly be made until certain information from the environment is available, so the pattern is incomplete initially. But this plasticity is a one-shot deal; once the final connections have been made, they are essentially just as irreversable as the spider's entirely pre-wired set. In humans, one pronounced example of this effect is in language aquisition.

Agreed.

Perhaps the sets of connections which produce the behaviors in the spider and the duckling are little more than predefined nodes with specific ranges of acceptable values. The duckling's plasticity may amount to little more than optimization within the bounds of a very limited domain. Even the phenomenal language aquisition capabilities of young humans appears to follow this pattern.

Agreed.

Now individuals venture forth with their preset connections set to various values, none of which being sufficient to produce an adaptive behavior which happens to exist in local design space, but some of which start out closer. A miss being as good as a mile, those individuals have no advantage over the others -- except that they are more likely to stumble on the right combination earlier in their reproductive lifetimes, at which point they begin to enjoy differential success. This gets reported back to the genome in the form of the greater numbers of their progeny (who inherit their preset connections). As those progeny come to dominate the population, the process repeats, the population always moving closer to the adaptive peak (paying a price in behavioral flexibility at the same time, btw -- the whole deal seems to be based on trading flexibility for more efficient hard-wiring).

It must be assumed that the adaptive trick is expressible using the available neural alphabet, so nothing says the others won't find it too -- just that more prewiring makes it easier to find (more likely to stumble upon). In fact, in animals smart enough to learn by imitation, an individual might not get much reproductive mileage out of a new trick before the others picked it up.

There are several things here with which I disagree. A few of them are simply wrong in evolution-theoretic terms. "A miss being as good as a mile": the equations of theoretical population genetics incorporate Darwinian fitness as a value between 0 and 1. The reason for that is that a miss is not as good as a mile. Small advantages are always measured in small differences in reproductive success. The duck pair that produced 100 ducklings in their lifetimes has a 1% advantage over the pair that produced 99. Secondly, speaking in terms of adaptive peaks is almost certainly wrong in evolution-theoretic terms, but that is a much longer discussion. Suffice it to say, there are no peaks that can exist (except momentarily) in a temporily- and spatially- varying environment.

Lastly, you are describing the opposite of the Baldwin effect here. You are describing an adaptive success conferred by the very "presets" countervalent to Baldwin's "plasticity".

Cheers,

 

[Dymanic]

Originally posted by Bill Hoyt

The reason for that is that a miss is not as good as a mile. Small advantages are always measured in small differences in reproductive success.

What I mean by 'a miss is a good as a mile', in this sense, is that we are talking about a behavior that emerges only once a certain threshold of complete connections has been reached; anything less, and there is no adaptive behavior whatsoever (fitness value = 0). Since the presets determine only some of those, the advantage they offer is zip until the others have been established (the advantage they offer is a shorter path to the completion of the full set).

You are describing an adaptive success conferred by the very "presets" countervalent to Baldwin's "plasticity".

The adaptive success is conferred by the presets. That is what saves the whole concept from being Lamarkian. Over time, in an organism in a very stable environment, I guess you would expect to see the plasticity disappear altogether. But as long as the presets alone are not enough to produce the behavior, plasticity could shorten the path to that point.

Or maybe the 'effect' is just a conceptual artifact; a product of a certain way of looking at things?

 

[BillHoyt]

What I mean by 'a miss is a good as a mile', in this sense, is that we are talking about a behavior that emerges only once a certain threshold of complete connections has been reached; anything less, and there is no adaptive behavior whatsoever (fitness value = 0). Since the presets determine only some of those, the advantage they offer is zip until the others have been established (the advantage they offer is a shorter path to the completion of the full set).

Dynamic,

First, I want to correct something I posted before. Fitness values do not range between 0 and 1; they must be simply greater than zero. My error.

With that correction, I'll assume you meant w=1.0, which means no selective advantage. With this you are hypothesizing that these successive connection changes, by random drift alone, are both retained in the genome and accrete! The far more likely scenario is that the first connection change conferred some advantage. (Although you put a pro-evolutionary spin to it, you are claiming the same thing creationists do to argue against evolution. "Half an eye is of no use". )

The adaptive success is conferred by the presets. That is what saves the whole concept from being Lamarkian. Over time, in an organism in a very stable environment, I guess you would expect to see the plasticity disappear altogether. But as long as the presets alone are not enough to produce the behavior, plasticity could shorten the path to that point.

I see that the success in your scenario is conferred by the presets. With this hypothesis, though, you are arguing against Baldwin's plasticity.

Cheers,

 

[Dymanic]

Originally posted by Bill Hoyt

(Although you put a pro-evolutionary spin to it, you are claiming the same thing creationists do to argue against evolution. "Half an eye is of no use". )

The untenability of the creationist position lies in a refusal to recognize that half an eye is still more eye than no eye at all, where anyone with half a brain can see that even a few light-sensitive cells might spell the difference between avoiding a predator and being eaten. This is different. If ten connections are required to produce a behavior, having nine of them will not procuce the behavior any more than will having two.

I see that the success in your scenario is conferred by the presets. With this hypothesis, though, you are arguing against Baldwin's plasticity.

The success is conferred by the presets, but (until the point is reached where the behavior is entirely hardwired) the plasticity is needed to extract their adaptive value (sort of).

With this you are hypothesizing that these successive connection changes, by random drift alone, are both retained in the genome and accrete!

This is where things feel fuzzy to me. Can you rephrase that? I think I'm saying that when changes in presets which, when used as a starting point for flexibles, produce the adaptive behavior, those get retained; it is the interaction between what is fixed and what is flexible that produces the result, but only what is fixed is retained.

I see the situation as analogous to the two-player game, Master Mind.

One player, the 'codemaker' creates a four-digit code, each digit represented by one of six colors. The other player, the 'codebreaker', tries to break the code by making a series of guesses, first randomly, then based on the results of earlier guesses.

So if the code is: RGBB, the guess: RYYG would get a score of: one direct hit (for the R in the right place) and one indirect (for the G -- right color, wrong place).

Now, a really lousy strategy is to simply go with a fresh guess every time, ignoring the results of previous guesses. The chance of hitting the right combination is always the same: 1 in 1296. Ignoring the results of the experimentation the plasticity permits would be analogous to this, but against much tougher odds. I'm not prepared to argue that evolution has not faced -- and beaten -- tough odds, but I can see how the flexible experimentation might act as a sort of scoring -- the equivalent to an 'indirect' hit. (Or is it a 'direct' hit -- I'm getting all confused).