'Nature' Paper Refigures the Evolution of Altruism

RichardPrins · #1 by **RichardPrins** » Feb 27, 2010 4:25 am

(PhysOrg.com) -- In 1871, Charles Darwin puzzled over the evolution of altruism. "He who was ready to sacrifice his life, as many a savage has been," he wrote in The Descent of Man, "rather than betray his comrades, would often leave no offspring to inherit his noble nature."

To this day, biologists debate about how altruistic behaviors evolve and persist. Sterile ants faithfully tend their queen with no chance of reproducing themselves. Vervet monkeys scream to other monkeys about approaching predators, drawing attention to themselves and risking their own safety. Bees lay down their lives to defend the hive.

"Why do they do that?" asks University of Vermont biologist Charles Goodnight. "Doesn't natural selection drive animals to behaviors that increase their own chances of survival, not those of others?"

This question underlies the decades-long debate -- sometimes rancorous -- between two camps of scientists. On one side, are those who argue in favor of "kin selection," in which individuals are altruistic to those who share their genes. In defending the hive, a self-sacrificing bee increases the chances that the genes she shares with her sisters will get passed down.

On the other side, are those who argue in favor of "group selection," (or, in its modern form, "multilevel selection") in which altruism arises from being part of a group. The self-sacrificing behavior of the bee persists and spreads across generations because the whole hive, a group, competes more successfully, leaving more offspring than others.

In the February 18 edition of the journal Nature a team of 18 scientists, including UVM's Goodnight, show that the two traditional approaches are actually mathematically equivalent.

One in the same
How can this be? In order for kin selection to be important, the related kin have to be in groups that preferentially confer altruistic behaviors on each other. In order for group selection to operate, the members of a group have to be closer kin than they are to other groups. The two ideas are close enough that they can actually be converted to each other mathematically. This understanding has been stated in technical research articles for more than 30 years, but the broader scientific community has not often recognized it.

"What we did in this paper was take the equations of a group that was very strongly kin selectionist and we worked through them and translated them back into classic equations," says Goodnight. "and they're the same."

"It is remarkable that kin selection has been widely accepted and group selection widely disparaged," says Michael Wade, a biologist at Indiana University, and the lead author on the paper, "when they are actually equivalent mathematically."

Evolution at all levels
A good bit of the fight between kin and group selection proponents is a product of history. (What fight isn't?) In the 1960's, some ideas about group selection were introduced that, in cartoon fashion, looked something like birds choosing not to reproduce for the good of their fellow birds. "There was this big rash of 'for the good of the group', naďve versions of evolution," says Goodnight.

But birds can't choose not to reproduce, nor can bacteria choose to be less virulent -- because it's good for their group. "Evolution doesn't work that way; evolution works by who leaves the most offspring," says Goodnight. Richard Dawkins and many other theorists largely dismantled this first wave of group selection ideas, and kin selection was ascendant. But in recent decades a new group-selection camp -- including Goodnight, David Sloan Wilson and others -- has emerged.

"The point is that evolution can work at many levels: the gene level, the cell level, the organismal level, the group level," Goodnight says, "and it probably works on all these levels at once."

The new paper in Nature considers the evolutionary mechanisms that would lead some parasites to have reduced virulence. From the kin selection (or individual-level selection) perspective, as presented in an earlier Nature paper by Geoff Wild at the University of Western Ontario and his colleagues, this lower virulence can be explained entirely by individual selection -- no group effect needed.

But Goodnight and his colleagues make a mathematical rebuttal, sketching out in their paper an argument for why two forms of opposing group selection -- "within-group" versus "among-group" -- are needed to explain how this seemingly disadvantageous trait nevertheless evolves in the whole parasite population.

"Those of us working on multilevel selection models have started seeing kin selection as subset of multilevel selection," Goodnight says, "The debate should no longer be whether it's individual or multilevel selection. The debate is how strong is each level of selection?" Or, as their paper concludes, "it's time to put the anachronistic debate between single-level and multilevel selection behind us."

Individual preference
Goodnight's colleague in the UVM biology department, Sara Cahan, agrees with this conclusion. But she doesn't agree with everything in Goodnight's paper -- and is more in the kin selection camp. "Charles and I really enjoy one another -- I respect Charles very highly -- but we do tend to argue a lot," she says, with a laugh. (Perhaps it's no wonder the students she and Goodnight had in their co-taught graduate seminar "Levels of Selection" called the course "Crossfire.")

"In this case of virulence, and in many cases where this altruism argument has been battled," she says, "the trait of interest is an individual-level trait. And if it's an individual-level trait, we really need to think about it as an individual-level adaptation -- regardless of what kind of selection, group or otherwise, has acted it."

"This debate is far from over," says Charles Goodnight.

More information: Paper: http://www.nature.com/nature/journal/v4 ... 08809.html

DanDare · #2 by **DanDare** » Feb 27, 2010 5:31 am

Its interesting. I have been strongly in the Selfish Gene camp since Richard published the book. Often, I couldn't understand how many of the attempts at describing a "group selection" system were actually different from gene selection.

Now I think Kin selection is seen to be foundational but there are some auxilary mechanisms at work that have groupish effects. Its going to make for a very interesting read.

monesy · #3 by **monesy** » Feb 27, 2010 7:48 pm

Kin selection accounts for altruism between organisms that have a high coefficient of relatedness. On the other hand, group selection often attempts to account for what reciprocal altruism accounts for -- altruism which is not necessarily associated with Hamiltonian kin selection. The mulitlevel selection theory, championed by David Sloan Wilson, is interesting because it takes Hamilton's Rule, and expands it so it is also applicable to groups that may not consist only of kin.

Kin Selection:
Hamilton's rule is rb > c, where r is the coefficient of relatedness, b is the (reproductive) benefit gained by the recipient of the altruistic act, and c is the (reproductive) cost cost incurred by the individual performing the altruistic act. If the product of the degree of relatedness and the benefit is greater than the cost, then altruism is a favorable strategy.

Multilevel selection theory:
This is simply an expansion of Hamilton's rule, such that (rbk + be) > c, where r is the coefficient of relatedness, bk is the (reproductive) benefit bestowed to kin, be is the (reproductive) benefit bestowed to the group, and c is the (reproductive) cost incurred by the altruistic individual.

Source.

'Nature' Paper Refigures the Evolution of Altruism

"Why do they do that?" asks University of Vermont biologist Charles Goodnight. "Doesn't natural selection drive animals to behaviors that increase their own chances of survival, not those of others?"

This question underlies the decades-long debate -- sometimes rancorous -- between two camps of scientists. On one side, are those who argue in favor of "kin selection," in which individuals are altruistic to those who share their genes. In defending the hive, a self-sacrificing bee increases the chances that the genes she shares with her sisters will get passed down.

On the other side, are those who argue in favor of "group selection," (or, in its modern form, "multilevel selection") in which altruism arises from being part of a group. The self-sacrificing behavior of the bee persists and spreads across generations because the whole hive, a group, competes more successfully, leaving more offspring than others.

But there is a third camp, alluded to by DanDare, which are the proponents of selfish gene theory. Selfish gene proponents argue that the best way of viewing evolutionary selection is at the fundamental unit of selection: the gene. Instead of focusing on (1) selection at the individual level (as in kin selection), (2) selection at the group level (as in group selection), or (3) selection at both levels (as in multilevel selection), one can simply look at selection at the most fundamental level -- the gene -- and account for what we observe as kin selection, as well as what we observe as reciprocal altruism. This is not to say that Hamilton's rule, or Wilson's expanded version of Hamilton's rule are not applicable. If the fundamental unit of evolutionary selection is the gene, these relationships would still naturally arise. Wilson's argument is that group selection is inherently accounted for in the same model that describes kin selection. But kin selection, and group selection (and, for that matter, any other known mechanism of evolution that may not even deal with altruism) would inherently be accounted for when one looks at evolutionary selection at the level of the gene, as opposed to looking at evolutionary selection at other levels such as the individual or the group.

DanDare · #4 by **DanDare** » Feb 28, 2010 2:23 am

Thanks for that monesy, that was very clear and sorted out some confusion for me.

Where do things like interspecies co-operation fit into this, for example domestication?

monesy · #5 by **monesy** » Feb 28, 2010 6:34 am

DanDare wrote:
Where do things like interspecies co-operation fit into this, for example domestication?

It depends on the scenario. You see, interspecies co-operation (often called mutualism) is not always regarded as altruism--and paradoxically enough, reciprocal altruism isn't always regarded as altruism either. For instance, mutualism that results in mutually beneficial fitness gains (+/+) is not technically altruism, and there is no mystery to why such behavior is evolutionarily advantageous for both players. Altruism, per se, is a special form of mutualism that has a fitness cost to the donor, and a fitness benefit to the recipient (-/+). Moreover, strong altruism decreases the absolute fitness of the donor, whereas weak altruism decreases the relative fitness of the donor. Terms, terms, terms...

Interspecies co-operation (in the sense of altruistic mutualism) is commonly explained using game theory within the paradigm of reciprocal altruism. IMO Dawkins does a very good job in The Selfish Gene explaining the evolutionary advantages of reciprocal altruism, with respect to the fundamental unit of selection (i.e., the gene), using simple layperson's instances of cooperators vs. defectors scenarios (chapters: You Scratch My Back, I'll Ride on Yours, and Nice Guys Finish First.).

While domestication isn't really a form of altruism, altruistic behavior does occur in domesticated animals. For instance, domesticated elephants and horses have both been used extensively in warfare and certainly incur potential costs to their fitness so as to the benefit their riders. Domesticated dogs have been observed to risk their own life to protect the lives of their 'adopted family'.

Viewing the gene is the fundamental unit of selection effectively enables us to dissolve (or perhaps the better word is "UNIFY") the notions of group selection, kin selection, reciprocal altruism, as well as other mechanisms of selection. These forms of selection are arguably just manifestations of a more fundamental selection at the level of the gene. In the case of altruism, selfish genes drive 'unselfish acts' in order to increase their chances of surviving and replicating. Even if the donor with a given gene should die, a copy of the same gene could very likely exist in the surviving recipient. To put things in perspective, the similarities between mouse and human genes have an average of 85% similarity.

DanDare · #6 by **DanDare** » Feb 28, 2010 3:44 pm

Yes, terms, terms, terms 8-)

but I see how it makes it easier to discuss succinctly.

Is the fact that I like it that my cat purrs when I pat him a detail that fits in this picture somehow? Is there a genetic advantage to the gene pool that I represent, in this pleasure response that appears to be built in to me?

monesy · #7 by **monesy** » Mar 01, 2010 12:33 am

DanDare wrote:
Is the fact that I like it that my cat purrs when I pat him a detail that fits in this picture somehow? Is there a genetic advantage to the gene pool that I represent, in this pleasure response that appears to be built in to me?

Cats purr under different circumstances and there is no question that we respond to this behavior. While we often assume that purring by wild and domesticated cats is an expression of pleasure, but that is not always the case as cats also purr when they are under duress. Purring is thought to have a intra-specific communication function, and possibly even a physiological function in stimulating muscle and bone growth/repair. Interestingly, a recent study has shown that domesticated cats alter their purr so as to make solicitations from humans, possibly capitalizing on "a mammalian sensitivity to acoustic cues relevant in the context of nurturing offspring." Hence, purring appears to have been co-opted by cats for inter-specific communication with their human co-dwellers (I hesitate to use the word "master" in a human-cat relationship).

Is this altruistic behavior under the definition in my last post? Likely not, but you can probably come up with specific scenarios in which altruistic behavior could develop. Even if altruism is not present, mutualism certainly is. The cat certainly benefits from purring provided that you respond to its solicitations, but what is the benefit to you? I think that you will likely benefit emotionally from the bond you form with your cat, which could help make you a emotionally balanced person--a desirable trait for a mate to have. A woman may observe your nurturing behavior toward your cat, and undoubtedly regard this as a positive trait for a father to have. Hence, it is possible that you may indirectly incur a fitness benefit from the pleasure you receive from your cat's purrs, and you may incur a more direct (even if accidental) benefit in fitness by your reaction (and reciprocation) toward your cat's purrs.

DanDare · #8 by **DanDare** » Mar 01, 2010 7:49 am

monesy wrote:I hesitate to use the word "master" in a human-cat relationship

I don't see why, its obvious that the cats are.

monesy · #9 by **monesy** » Mar 01, 2010 7:58 am

True that!

zoon · **#10** by **zoon** » Mar 01, 2010 1:55 pm

A link to the original paper by Wild, Gardner and West here, and to the reply by Wade, Wilson et al with the reply to that one here.

Not that I’ve read them, let alone followed the maths. I think these two groups have had an ongoing argument for some years. My suspicion is that D.S. Wilson is particularly attached to group selection because it has fewer worrying implications for humans than kin selection.

monesy · **#11** by **monesy** » Mar 04, 2010 6:18 am

Those are some pretty interesting papers--you should read them. They certainly help to add context to the OP.

Squawk · **#12** by **Squawk** » Mar 06, 2010 10:37 pm

Hey guys, I've read the thread, going to read the links in due course. This has been extremely interesting for me not least because I was involved in a debate on altruism just yesterday, with the debate centered around group vs kin or gene selection. I'd like to make sure I've got this correct and that my understanding is in accord with the papers in nature.

First, let me make a couple of definitions for clarity. These are given in basic terms, I'm no evolutionary biologist.

Fitness: The probability of a given individual passing on genes to future generations (ie, having kids).
Altruism: Any act by organism X that serves to reduce it's fitness whilst increasing the fitness of organism Y, where X and Y are of the same species group and/or species.

The only way I can make sense of this is to consider selection as acting at the level of the gene, thanks to Richard Dawkins and the selfish gene which I found compelling, and it leads me to a conclusion that I want confirmation of, or a refutation of.

I think that, by definition, true altruism must always be deleterious. I cannot conceive of any selection pressure that can select in for true altruism.

I can think of lots of selection pressures that would result in high levels of co-operation and reciprocal altruism but in all cases this would not meet my definition of altruism unless each altruistic act is considered in isolation. The fitness of a given organism should be measured over it's lifetime.

In all cases I can conceive of group and kin selection it seems that both are actually generalisations of gene selection, rather than being mechanisms in and of themselves.

I think true altruism can only be considered a misfiring of genes that have evolved for reciprocal altruism and co-operation because on the face of it true altruism would always be deleterious, a situation that must be the case if selection really does occur at the gene level.

monesy · **#13** by **monesy** » Mar 06, 2010 11:54 pm

Squawk wrote:
Fitness: The probability of a given individual passing on genes (ie, having kids).

That would be Hartl's definition of finess. Much of the time fitness is an actual measurement, and you should also note that there is a distinction between relative fitness and absolute fitness. See common definitions here. You also need to understand the concept of inclusive fitness, as it is very important concept in this discussion. Moreover, you may apply these definitions to alleles, as you have a gene-centered view of evolution.

Altruism: Any act by organism X that serves to reduce it's fitness whilst increasing the fitness of organism Y, where X and Y are of the same species group and/or species.

Yes, except it is not a requirement that organism X and Y are part of the same species, or even the same group.

The only way I can make sense of this is to consider selection as acting at the level of the gene, thanks to Richard Dawkins and the selfish gene which I found compelling, and it leads me to a conclusion that I want confirmation of, or a refutation of.

You will get confirmations or refutations depending on what camp you ask. One can also argue that the selection may act at the level of the group, or the individual (as we see in the two papers), and it is very difficult to tease out where the selection is actually acting due to the fact that the maths are essentially equivalent. I, however, share your viewpoint that evolution by natural selection is best understood when one focuses at the level of the gene.

I think that, by definition, true altruism must always be deleterious. I cannot conceive of any selection pressure that can select in for true altruism.

Scenario: We both have gene X. We are in a situation where there is a high probability both parish unless one of us sacrifices our fitness.

In this scenario, NON-altruistism deleterious to gene X. Dawkins provides many examples of how and why altruism is selected for in The Selfish Gene, which introduces the idea that selection is best understood when we view it at the level of the gene.

I can think of lots of selection pressures that would result in high levels of co-operation and reciprocal altruism but in all cases this would not meet my definition of altruism unless each altruistic act is considered in isolation.

Sure, an altruistic act is a single act; however, altruistic behavior is a general behavior inclining an organism to behave altruistically (i.e., we are looking at several acts). When altruism is modeled, more than a single act is being taken into account. In fact, for reciprocal altruism to be an evolutionarily stable strategy, it is an absolute requirement that there is a high probability of multiple interactions. Game theory is commonly used to model these multiple iterations to determine if a given strategy is evolutionarily advantageous, neutral, or disadvantageous.

The fitness of a given organism should be measured over it's lifetime.

Actually, just one lifetime is not enough. Fitness is about VIABLE offspring, not just offspring, right? An organism could have multiple offspring, but if these offspring are all sterile, then the organism's fitness does not increase. Hence, we actually need to look at how a strategy affects the fitness of an allele over multiple generations.

In all cases I can conceive of group and kin selection it seems that both are actually generalisations of gene selection, rather than being mechanisms in and of themselves.

I could not agree more! However, it is difficult to show this. I think it is possible...just difficult.

I think true altruism can only be considered a misfiring of genes that have evolved for reciprocal altruism and co-operation because on the face of it true altruism would always be deleterious, a situation that must be the case of selection really does occur at the gene level.

As mentioned, from an evolutionary point of view altruism is certainly not always deleterious. To many organisms, it is an evolutionary stable strategy. In fact, evolutionary theory predicts altruism in species that are highly social, highly related, or both.

Squawk · **#14** by **Squawk** » Mar 07, 2010 12:41 pm

Thanks for the reply monesy. A couple more questions and observations.

monesy wrote:
Squawk wrote:
Altruism: Any act by organism X that serves to reduce it's fitness whilst increasing the fitness of organism Y, where X and Y are of the same species group and/or species.

Yes, except it is not a requirement that organism X and Y are part of the same species, or even the same group.

I can see how this applies in a general sense in that you can be altruistic or not towards another species, but if we are to consider selection pressures as they may apply in selecting for or against altruism I concluded that it would only matter intra-species, and if we are to discuss group selection then also intra-group. I recognise that this is a restriction, but it seems to apply in that sense.

monesy wrote:
Squawk wrote:
I think that, by definition, true altruism must always be deleterious. I cannot conceive of any selection pressure that can select in for true altruism.

Scenario: We both have gene X. We are in a situation where there is a high probability both parish unless one of us sacrifices our fitness.

In this scenario, NON-altruistism deleterious to gene X. Dawkins provides many examples of how and why altruism is selected for in The Selfish Gene, which introduces the idea that selection is best understood when we view it at the level of the gene.

I see that as a counter example, but let me expound upon it properly and see if it still holds water. I apologise if I get terminology wrong, please correct any.

I want to refer to pseudo-altruism to mean an act that appears to be altruistic but might not be, depending on where we consider selection to occur. For example, a mother sacrifices herself to save her 5 kids. I would refer to this as pseudo-altruism in that she gave her progeny the best chance of survival, although in everyday speak I would simply refer to it as altruism. I wish to do this here to see if I can find a selection pressure that would turn pseudo-altruism into true altruism.

Ok, two hypothetical populations.
Pop1: All individuals have genes giving equal levels of pseudo-altruistic behaviour
Pop2: Differential levels of pseudo-altruistic behaviour in each individual.

Then lets consider two individuals who happen to share a genotype, two twins, and put them into the situation you described above. If neither acts, both die. If both act or one acts, one survives and the other dies.

Now consider that these twins come from population 1. Regardless of actions, the mean pseudo-altruistic behaviour of the population will remain the same. If neither acted the population will be smaller by two individuals, if one or both acted the population will be smaller by one individual.

Further, lets postulate another group in which the situation is identical, but the mean level of pseudo-altruism is slightly lower, and lets suggest that these two groups are in competition with each other. With everything else being equal we do have differential selection of the groups, the group with higher levels of pseudo-altruism will survive.

When we view the unit of selection as the gene we note that the supposed altruistic behaviour actually increased the genes fitness. Only one copy of X, whatever X happens to be, is removed from the population. If altruistic behaviour doesn't occur two copies of X are removed from the population. Far from altruism, from the genes point of view altruism of this sort is selfish.

The population in which the "more altruistic" gene is present does better than the one in which it is not present, but this is an emergent property from X, the gene, acting in a selfish manner.

Lets go for a more realistic scenario and put our twins in population 2, one with differing levels of pseudo-altruism, but the twins have identical levels of pseudo-altruism. Lets take this even further to say that rather than one set of twins we have 10 sets of twins each encountering the same set of circumstances, but independently.

The result can only be an increase in pseudo-altruism within the population since any twins that did not act pseudo-altruistically were both wiped out, whereas only one of the pair was killed when acting pseudo-altruistically. Apparently altruistic behaviour is again exposed as selfish behaviour on the part of the gene when we note that the genes that acted altruistically increased their proportion in the population.

So far so simple, pseudo-altruistic behaviour will always be selected for when the two individuals involved share the gene/behaviour that is being selected.

The interesting cases emerge when we consider the case where the individuals making the choice do not possess equal pseudo-altruism, something which I'll get to after the next bit because you cover it.

monesy wrote:
Squawk wrote:
I can think of lots of selection pressures that would result in high levels of co-operation and reciprocal altruism but in all cases this would not meet my definition of altruism unless each altruistic act is considered in isolation.

Sure, an altruistic act is a single act; however, altruistic behavior is a general behavior inclining an organism to behave altruistically (i.e., we are looking at several acts). When altruism is modeled, more than a single act is being taken into account. In fact, for reciprocal altruism to be an evolutionarily stable strategy, it is an absolute requirement that there is a high probability of multiple interactions. Game theory is commonly used to model these multiple iterations to determine if a given strategy is evolutionarily advantageous, neutral, or disadvantageous.

That's pretty much what I was getting at but my wording was inadequate.

monesy wrote:
Squawk wrote:The fitness of a given organism should be measured over it's lifetime.

Actually, just one lifetime is not enough. Fitness is about VIABLE offspring, not just offspring, right? An organism could have multiple offspring, but if these offspring are all sterile, then the organism's fitness does not increase. Hence, we actually need to look at how a strategy affects the fitness of an allele over multiple generations.

Yeah, this is a case of my lacking terminology. I wanted to refer to fitness of progeny, but wasn't sure if you can say it like that. What are the chances of a given organism passing it's genes on into subsequent generations, of its genes entering the gene pool. A miswording on my part.

monesy wrote:
Squawk wrote:In all cases I can conceive of group and kin selection it seems that both are actually generalisations of gene selection, rather than being mechanisms in and of themselves.

I could not agree more! However, it is difficult to show this. I think it is possible...just difficult.

I've been wondering about that. Wouldn't the maths, as presented by the papers, being equivalent suggest that this is infact the case? That one is simply a generalisation of the other? Kinda like macro vs micro evolution if we consider macro to be simply accumulated micro, group selection would be accumulated gene selection and therefor the maths would have to be equivalent? I would think that you can express macro evolution in the mathematics of micro-evolution should you be so inclined, and I would argue that you can express group selection in the maths of gene selection. I'd bet you can't do it the other way around.

monesy wrote:
Squawk wrote:I think true altruism can only be considered a misfiring of genes that have evolved for reciprocal altruism and co-operation because on the face of it true altruism would always be deleterious, a situation that must be the case of selection really does occur at the gene level.

As mentioned, from an evolutionary point of view altruism is certainly not always deleterious. To many organisms, it is an evolutionary stable strategy. In fact, evolutionary theory predicts altruism in species that are highly social, highly related, or both.

That's what I am arguing against. I don't think highly social species displaying co-operation and reciprocal altruism can be said to be truely altruistic. I think they can only be considered pseudo-altruistic, though pseudo-altruism can be a very close approximation to true altruism.

The two examples I gave above, with the twins in different populations, would give reasonable, if highly simplified, mechanisms by which this pseudo-altruism can arise. Let's go one step further and consider an interaction in population 2 in which the two individuals making the choice are always of slightly different altruistic tendencies.

In this case the arising of an ESS is the only possibility. With differential survival based on altruism true altruism can never be selected for. Consider a population in which creatures act altruistically 99.9% of the time. They could, by simply chance, end up with a population that exhibits 100% truely altruistic behaviour. However there can be no pressure to get there. If any situation arises at any time such as those we posited earlier an individual that chooses not to act altruistically will be favoured. This individual may still act altruistically 99.9% of the time, but on that rare occasion when it does not act altruistically it gains a slight advantage over the individual that does act altruistically.

It is this final step that I cannot see a selection pressure for. I can't conceive how that final tiny step to 100% altruism can ever be selected for. I can see how it can arise by chance, through some variation of genetic drift, but it would seem that it would always be selected against.

And it is that point, as much as anything else, that leads me to conclude that group selection doesn't apply. If a selection pressure can exist on the group as a whole then that final step to 100% altruism could be selected for. I contend it can never get there, and thus that selection is always at the gene level.

natselrox · **#15** by **natselrox** » Mar 07, 2010 12:46 pm

susu.exp · **#16** by **susu.exp** » Mar 07, 2010 1:35 pm

Squawk wrote:I've been wondering about that. Wouldn't the maths, as presented by the papers, being equivalent suggest that this is infact the case? That one is simply a generalisation of the other?

Well, both kin selection and modern group selection can be shown to be equivalent mathematically using the price equation.
COVI(wi/w,gi)+EI(wi/w)*COVJ(wij/wi,gij)=COV(wij/w,gij)
Groups are indexed i belonging to a set of groups I and individuals within groups are indexed j. The left side divides selection into a within group and a between group term, the right side doesn´t and you can recover kin selection theory from it. By writing both sides out, you can easily confirm they are identical. Neither is a special case of the other.

Squawk wrote:Kinda like macro vs micro evolution if we consider macro to be simply accumulated micro, group selection would be accumulated gene selection and therefor the maths would have to be equivalent?

Bad analogy. While group and kin selection are the same thing, this does not apply to micro and macroevolution. Microevolution describes the change of allele frequencies in populations of organisms, macroevolution describes the change of allele frequencies in populations of species. These are not equivalent and to a great deal independent (with few exceptions - the species level frequency of an allele is 0 when it´s 0 for all populations of the species in question for instance). There´s a problem with terminology, because there are currently two senses in which hierarchial-selection is used. One use is for group selection theories as in the paper. The other one is for the various levels of evolution (macroevolution, microevolution and somatic evolution). The former is eqivalent to a single inclusive fitness model, the later is not.

Squawk wrote:I would think that you can express macro evolution in the mathematics of micro-evolution should you be so inclined, and I would argue that you can express group selection in the maths of gene selection. I'd bet you can't do it the other way around.

The former: no. The later: yes.

Squawk wrote:Fitness: The probability of a given individual passing on genes to future generations (ie, having kids).

A better definition for the fitness of an individual it the expected number of offspring. I.e. the sum n*p(n) where n is an offspring number and p(n) the probability of having n offspring.
In an infinite population, there are infinitely many clonal copies of each genome and their mean number of offspring is equal to the fitness.

Squawk · **#17** by **Squawk** » Mar 07, 2010 1:54 pm

Thanks suso, I'm gonna take some time to digest that.

DanDare · **#18** by **DanDare** » Mar 07, 2010 2:45 pm

Squawk wrote:I can see how this applies in a general sense in that you can be altruistic or not towards another species, but if we are to consider selection pressures as they may apply in selecting for or against altruism I concluded that it would only matter intra-species, and if we are to discuss group selection then also intra-group. I recognise that this is a restriction, but it seems to apply in that sense.

Other species may share the same gene, so inta species altruism could still foster that as long as you are selective about choice of species. Also very diferrent species could co-evolve mutually supportive genes. So the gene of species A benefits the fitness of species B and the gene in species B benefits the fitness of species A.

An interesting test would be contact with intelligent aliens. We could not possibly share any genes but it is possible our genes could cause us to help them and v v.

monesy · **#19** by **monesy** » Mar 08, 2010 2:39 am

Squawk wrote:
monesy wrote:
Squawk wrote:
Altruism: Any act by organism X that serves to reduce it's fitness whilst increasing the fitness of organism Y, where X and Y are of the same species group and/or species.

Yes, except it is not a requirement that organism X and Y are part of the same species, or even the same group.

I can see how this applies in a general sense in that you can be altruistic or not towards another species, but if we are to consider selection pressures as they may apply in selecting for or against altruism I concluded that it would only matter intra-species, and if we are to discuss group selection then also intra-group. I recognise that this is a restriction, but it seems to apply in that sense.

That restriction is unneeded. Not only is inter-species altruism observed in nature, but under the framework of reciprocal altruism, a selective advantage can be conferred in scenarios involving altruism between different species. The advantage of such behavior is easy to see when one (1) takes a gene-centered view of evolution, and (2) observes that there can be a very high degree of conservation of genetic similarity between members of different species.

Squawk wrote:
monesy wrote:
Squawk wrote:
I think that, by definition, true altruism must always be deleterious. I cannot conceive of any selection pressure that can select in for true altruism.

Scenario: We both have gene X. We are in a situation where there is a high probability both parish unless one of us sacrifices our fitness.

In this scenario, NON-altruistism deleterious to gene X. Dawkins provides many examples of how and why altruism is selected for in The Selfish Gene, which introduces the idea that selection is best understood when we view it at the level of the gene.

I see that as a counter example, but let me expound upon it properly and see if it still holds water. I apologise if I get terminology wrong, please correct any.

I want to refer to pseudo-altruism to mean an act that appears to be altruistic but might not be, depending on where we consider selection to occur. For example, a mother sacrifices herself to save her 5 kids. I would refer to this as pseudo-altruism in that she gave her progeny the best chance of survival, although in everyday speak I would simply refer to it as altruism. I wish to do this here to see if I can find a selection pressure that would turn pseudo-altruism into true altruism...

I am unsure as to why you are using “pseudo” as a prefix here because you are simply describing is altruism. Is it because you are noticing that what appear as altruistic behavior at the level of the organism or group, is actually, paradoxically enough, “selfish” at the level of the gene? If so, drop the “pseudo” tag and give yourself a pat on the back for making the connection. Of course it is selfish--that is precisely why it is selected for! What is most interesting is that this gene-level ‘selfishness’ manifests itself behaviorally as ‘unselfishness’ at the level of the individual or group. Nonetheless, this is not pseudo-altruism; it is altruism.

Squawk wrote:
monesy wrote:
Squawk wrote:In all cases I can conceive of group and kin selection it seems that both are actually generalisations of gene selection, rather than being mechanisms in and of themselves.

I could not agree more! However, it is difficult to show this. I think it is possible...just difficult.

I've been wondering about that. Wouldn't the maths, as presented by the papers, being equivalent suggest that this is infact the case? That one is simply a generalisation of the other? Kinda like macro vs micro evolution if we consider macro to be simply accumulated micro, group selection would be accumulated gene selection and therefore the maths would have to be equivalent? I would think that you can express macro evolution in the mathematics of micro-evolution should you be so inclined, and I would argue that you can express group selection in the maths of gene selection. I'd bet you can't do it the other way around.

.

The equivalence indicates that kin selection and multilevel (i.e., “group”) selection are not really distinct types of altruism. This, however, does not tell us what level is the “correct” level to view selection. As stated in the paper of Wild et al., “the presence of group selection does not invalidate the idea that the individual is an adaptive unit, and it does not validate the idea that the group is an adaptive unit.” Of course, this doesn’t stop me from ripping into group selection a little; I’ll just take a different angle of attack...

Let’s again look at Hamilton’s rule and D.S. Wilson’s expanded version of Hamilton’s rule.

Hamilton’s Rule is rb > c, where b represents the benefit to the recipient of altruism, c the cost to the altruist, and r their degree of relatedness.

D.S. Wilson’s expansion of Hamilton’s Rule to account for multilevel selection is (r(bk) + (be)) > c, where (bk) = b in the original equation, and (be) is the benefit to the group.

Notice first that D.S. Wilson’s expansion of Hamilton’s rule is indeed distinctive. Notice also that (be) is completely independent of relatedness. This means that the group benefit has nothing to do with, and is unaffected by, inclusive fitness. This leads us into a couple of big problems with the multilevel selection model.

The Problem of Parsimony:
Entia non sunt multiplicanda praeter necessitatem (entities must not be multiplied beyond necessity). As discussed, an equivalence has been shown in the maths underlying kin and multilevel selection. This means that any selection occurring at the group level, if such selection even exists, is already accounted for by b in Hamilton’s rule. Thus, (be) is a superfluous term, and the D.S. Wilson’s expansion of Hamilton’s rule lacks parsimony. Moreover, while one can empirically measure b (and by extension, (bk)), one cannot empirically tease out (be) in the Hamilton expansion.

The Problem of Relatedness Independence:
Genetic inheritance is paramount to evolution by natural selection, and r is essentially a measure of genetic inheritance. We see by D.S. Wilson’s expansion of Hamilton’s rule that (be) is completely independent of r. So without r, how does one relate (be) to genetic inheritance? It goes without saying: groups, as a whole, do not reproduce. So how, then, is selection supposed to work on (be)?

In summary: group selection is superfluous, thus, non-parsimonious, it cannot empirically be shown to exist, and it appears to be invisible to mechanisms of selection. Wow.

Squawk wrote:
monesy wrote:
Squawk wrote:I think true altruism can only be considered a misfiring of genes that have evolved for reciprocal altruism and co-operation because on the face of it true altruism would always be deleterious, a situation that must be the case of selection really does occur at the gene level.

As mentioned, from an evolutionary point of view altruism is certainly not always deleterious. To many organisms, it is an evolutionary stable strategy. In fact, evolutionary theory predicts altruism in species that are highly social, highly related, or both.

That's what I am arguing against. I don't think highly social species displaying co-operation and reciprocal altruism can be said to be truely altruistic. I think they can only be considered pseudo-altruistic, though pseudo-altruism can be a very close approximation to true altruism.

The two examples I gave above, with the twins in different populations, would give reasonable, if highly simplified, mechanisms by which this pseudo-altruism can arise. Let's go one step further and consider an interaction in population 2 in which the two individuals making the choice are always of slightly different altruistic tendencies.

In this case the arising of an ESS is the only possibility. With differential survival based on altruism true altruism can never be selected for. Consider a population in which creatures act altruistically 99.9% of the time. They could, by simply chance, end up with a population that exhibits 100% truely altruistic behaviour. However there can be no pressure to get there. If any situation arises at any time such as those we posited earlier an individual that chooses not to act altruistically will be favoured. This individual may still act altruistically 99.9% of the time, but on that rare occasion when it does not act altruistically it gains a slight advantage over the individual that does act altruistically.

It is this final step that I cannot see a selection pressure for. I can't conceive how that final tiny step to 100% altruism can ever be selected for. I can see how it can arise by chance, through some variation of genetic drift, but it would seem that it would always be selected against.

And it is that point, as much as anything else, that leads me to conclude that group selection doesn't apply. If a selection pressure can exist on the group as a whole then that final step to 100% altruism could be selected for. I contend it can never get there, and thus that selection is always at the gene level.

Again, I am going to ignore the “pseudo” tag you are using, as per my assumption made at the beginning of the post as to why you are using it.

An ESS does not require that a population consists of 100% altruistic individuals, nor does it require that individuals act altruistically 100% of the time. Just because a population is does not (or can not) reach a hypothetical level of 100% altruistic behavior by 100% of individuals, 100% of the time does not constitute evidence against group selection at all. I think you might be misunderstanding group selection. Here is an excerpt from the Wade et al. paper which will probably clear things up a bit:

When seeking to maximize individual traits like ‘egg laying’ or ‘survival’ in hens, breeders find that selecting the most productive coops works better than selecting the most productive individuals, because it allows variation in social effects to contribute to the response. Indeed, the heritability of survival is 1.5- to 6-fold higher when indirect effects are considered.

Well, here is the definition of heritability. If one reads between the lines, it appears that an implicit argument is being made that indicates group selection is based upon (or perhaps more correctly stated: “influenced by”) non-genetic components. Again I must ask--how is a mechanism of selection supposed to see this? Ah...but I digress... (Just trying to get some final jabs in against group/multilevel selection.)

As for your suggestion that cheaters can prosper: yes, there certainly can be a selective advantage to cheating. But keep in mind that this potential advantage is trumped by an organism’s ability to recognize a cheater (or potential cheater, for that matter). Ergo, If you cheat, don’t get caught!

Anyways -- I hope I haven’t misunderstood your argument, and I hope this post helps you. (How very altruistic of me!)

Squawk · **#20** by **Squawk** » Mar 08, 2010 11:19 am

Thanks monsies. You are correct about my use of pseudo when referring to altruism. I'm unsure how best to phrase it though, I think altruism is one of those terms that is open to abuse or simple misunderstanding. To really refer to a given instance of altruism I suspect we would have to identify a given behaviour as altruistic at all levels of selection. As you point out, a behaviour can be altruistic at the group level, but selfish at the gene level. Is that action, then, altruistic?

I'd go with no, in the strictest sense, but then I'm hardly qualified :waah:

It's a quibble over semantics really, but I think it's an important quibble.

'Nature' Paper Refigures the Evolution of Altruism

'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Re: 'Nature' Paper Refigures the Evolution of Altruism

Who is online