Monday, June 8, 2015

The Meaning of Life, Part II

In re: one of my recent reviews, my faithful correspondent Sean asked: 
For those of us not in the know, what are the models of group evolution you mention?
The answer turned out to be too long for a reply, so here it is. DISCLAIMER: I am a complete amateur in this area, so take this synopsis for what it's worth.

Conventional evolution--the "Darwinian Synthesis"--notes that the gene, rather than the individual, is the unit of selection. That is, evolution operates by killing off some genes and replicating others. The organism is simply the vehicle.

However, biologists have long argued about the evolution of certain traits that don't seem to favor the reproductive success--and, hence, the genes--of the individuals who have them. The classic conundrum is altruism. If you jump into a raging river to save a drowning person, you are much more likely to die than if you watch dispassionately from the shore. Yet altruism is a human universal, and has been observed in other species as well. Why have not those genes been eliminated?

One simple answer is group selection. The naive version goes like this: Yes, your genes will perish. But your tribe is stronger because you are willing to do this. Your tribe will cooperate more, save each others' lives, and therefore outcompete other, less-altruistic tribes. This model imagines natural selection operating at the levels of groups: better-adapted groups thrive, worse-adapted groups don't.

This is intuitive. The problem is that the math doesn't seem to support it. The benefit to the group has to be implausibly huge, and the risk tiny, for these genes to survive. The reason is that you're directly competing with people within your group, and compared to them, you, Mr. Altruist, are a total loser. As in, you are dead, rendering your further reproductive success highly doubtful

A somewhat different explanation, which came to prominence in the 1960s and 1970s, is called kin selection, or inclusive selection. Kin selection says, basically, that altruism is disguised selfishness. Your genes are willing to risk themselves to the extent that the victim in the river may have some of those same genes. Thus, most people are very altruistic towards their children; somewhat altruistic towards near relatives; slightly altruistic to distant relatives; and very little altruistic towards complete strangers.

Kin selection is now the orthodoxy among biologists. It has some math to back it up, and some examples that fit the observed data. Edward Wilson himself was originally a prominent supporter.

More recently he's broken with the Church, and has been appropriately vilified therefor. His current argument is that there is a more subtle form of group selection that does meet the mathematical tests, while the math behind kin selection is flawed. 

Boiling it down, my understanding of his position is that, just as you can model an individual as a carrier of heteregeneous genes, you can model a group the same way. The unit of selection is still the gene--but if you're talking about a gene that's dispersed through the group, it may make sense to talk about the group collectively "having" some genetic trait.

In a social species, for example, a gene that promotes a feeling of social cohesion may be an individual reproductive advantage; individuals that have it will cooperate to outcompete individuals who don't. At the point where the "cooperation" gene is dispersed through the group, you can sensibly model natural selection at higher levels than the individual.

Obviously, you should not rely on a dilettante engineer for your biology homework. I don't know enough about the subject to have an opinion one way or the other, although it does seem to me that both Wilson and his opponents--especially Richard Dawkins--are behaving somewhat childishly. Here is what seems to me to be an in-depth "middle-ground" position, if you're curious.

10 comments:

  1. Fascinating. There must be a meme out there, because I’ve been reading, watching, and listening to a lot of stuff that reminds me of this.

    One of your recommended books, Freakonomics, also has a podcast (http://tinyurl.com/oa253a8) and radio show that I listen to a bit. The most recent one is about homo-economicus, a fictional human that acts purely as an economist would model us to be. It hit home for me, largely because humans often really don’t (and I mean REALLY don’t) act rationally. Homo-economicus would have us calculate with perfect precision the opportunity cost for choosing one mate over all potential future mates, for example, and therefore most likely result in never committing to a monogamous relationship. But, as you know, we don’t act that way. People do things for perfectly valid reasons (note, I did not say rational reasons, though they are valid and comprehensible), but they are not always strictly in our objective best interests. So you can “model an individual as a carrier of heterogeneous genes” but you can’t get him to always act in his own best interest.

    So why does a person “jump into a raging river to save a drowning person?” Maybe it has something to do with feeling invulnerable, like teenagers do. Part of it, I think, has to do with synchronization. As you mention, we are altruistic toward people we feel kinship toward (which is the great premise behind the Forever War, and the Forever Peace, for that matter – both really good books I should re-read). There really is something in our wiring that makes us this way. But we don’t always need the close relationship. We have mirror neurons that fire when one person acts and when another observes the act. We cringe when we see someone hurt themselves, whether we know them or not.

    So “why do we still have these genes?” You would suppose that humans, the most rational animal (loaded statement there, I know) would have figured this out and stopped the behavior. Or, one could just as easily argue that we are above our baser instincts and should therefore be more altruistic, because to act in pure self-interest is… well our baser instinct. It could be, however, that our baser instinct is indeed to be selfless (or at least to work with others). There’s a really interesting TED talk by Frans de Waal (http://tinyurl.com/p9skfq3) in which he concludes that we find empathy, consolation, pro-social tendencies, reciprocity, and a sense of fairness in primates (not homo-economicus, and not sapiens either). Even other animals like elephants, too. So, morality (selflessness, pro-social tendencies) comes from animals. Why do animals have the genes? I’ll let you know when I write my best-selling book on the subject :)

    ReplyDelete
    Replies
    1. There's no doubt that empathy, consolation, etc. exist, in humans and others. Same for "feeling invulnerable" and "mirroring". The question--presuming that these traits are genetic--is why? Shouldn't the sociopath, who's concerned purely with his own self-interest, out-compete the rest of us? None of these "positive" traits has any obvious positive effect on fitness, in the Darwinian sense, and some of them should be actively bad. Why haven't they been bred out?

      Yet there they are. Clearly these genes are adaptive, odd as it may seem. This is, I think, the flip side of your Freakonomics case. That book has two themes:
      1. People respond to incentives, as they perceive them
      2. People's perceptions are not always economically rational, and can even be quite wacky.

      That's because we have genes for behaviors that are adaptive, even if they're not "logical". The emerging field of behavioral economics, as practiced by folks like Dan Ariely, is concerned with these behaviors.

      Suppose we play a game. We flip a coin. If it's heads, I give you $20. Most people are happy to take this bargain, for obvious reasons.

      Now reframe it. I give you $20 up front, but then we flip a coin. If it's tails, you have to give me the $20 back.

      These are economically identical: either way, you have a 50/50 shot at gaining $20. But people do not like the second version! That's our genes talking. In the wild, it's maladaptive to let people take stuff from you once you've gotten hold of it.

      Delete
    2. I think it's to gain trust. I see this a lot where I work in Japan. Self-sacrifice is a way to show that you are worthy within the group. The bigger the sacrifice the greater the trust you gain. It's a social currency. One example was a major project that had failed. But the manager in charge of the project had worked an extreme amount of overtime. He was not demoted, or even significantly reprimanded. The problem was not with the amount of work that went into the project, however. It was poor decision making. Even if he had worked an additional year of overtime, he still would have made the bad decision. Yet he was "forgiven" because he had built up the bank of trust.

      If you combine this with a sense that you are invulnerable, you get the risky behavior that doesn't seem to make sense on the surface.

      Delete
    3. Sure, but the question still remains: how the hell did those genes survive?

      Delete
    4. Think of it like this: You are a cave man. You have a mate and offspring. But you need to go out and kill a wooly mammoth for the good of the tribe. Your mate and offspring stay at the cave. You want the rest of the tribe to protect your mate and offspring for you when you're not around. You need to have some really good relationships to do that. If you don't have trust in your social bank then they'll be left out to dry. You won't want to leave their side. So no mammoth burgers for dinner. The tribe loses.

      Delete
    5. Which is exactly the "naive" group-selection argument. Your genes are bad for you, but they're good for your tribe; therefore, your tribe is better able to compete with other tribes.

      The trouble is that it doesn't seem to work like that. Your tribe may be competing with other tribes, but you personally are competing for reproductive success with other members within your tribe. While you're out hunting mammoth and maybe getting killed, they're goofing off in safety and maybe trying to seduce your mate. All you've done is give them an excellent reason to take advantage of you, you poor trusting sod, you. Your cooperative genes are a massive disadvantage if even ONE of your fellow tribespersons doesn't share them.

      Basically, the math shows that--except in some very extreme situations--the pressure of within-group competition is much, much, much greater than the pressure of between-group competition.

      Here's another well-known model, by the geneticist John Maynard Smith.

      Suppose you have a bunch of haystacks. Each haystack is colonized by mice. Each mouse is either "docile" or "aggressive," with a high but not perfect degree of heritability, and mice may move between haystacks.

      The naive group selection argument is that most of the haystacks should end up full of docile mice. Aggressive mice fight each other; docile mice cooperate. So a haystack can hold many more docile mice.

      The trouble is, when you introduce a single aggressive mouse into a docile haystack, he rules. He totally outcompetes all the docile mice in that haystack. Within a few generations, the docile gene is wiped out. You end up, in this model, with haystacks of aggressive mice.

      Now, of course, this is a simplified version of reality. It's supported by some equations, though, and has been modeled on computers, so it's not an idea that can be simply refuted. Kin-selection and Wilson's "multi-level selection" are both ways to explain the fact that the observed data don't fit the model.

      Delete
    6. I don't think that it necessarily obviates my point. One could argue that we still need some poor trusting fool to go out and kill the mammoth. If we didn't then we wouldn't have enough to eat. And we wouldn't have someone's mate to steal. Trust is the fundamental glue that keeps groups together, and we are by our nature trusting. The ones that betray that trust are hated, but they exist. That doesn't mean that they can necessarily dominate permanently. The advantages of the group far outweigh living alone. So if we need trust to form groups then we keep the gene.

      Delete
    7. What's this "we" you're talking about? Genes don't know about "we". Genes either reproduce themselves successfully, or they don't. You're still conflating the success OF the group with success WITHIN the group. Genes operate at the latter level, and there's no reason that's what's good for an individual gene must be good for the group its owner happens to be part of.

      Consider two imaginary tribes.

      The Aldas are nice, cooperative, and trusting. When they go out hunting mammoths, everyone shares equally in the kill. Everyone helps one another, even when some cost or risk is involved.

      The Lecters are just the opposite. They're no less intelligent than the Aldas, but they're amoral and entirely self-interested. They're willing to cooperate to hunt a mammoth, but when they're done there's no sharing; you get what you can grab. Nobody helps out anyone unless there's some reciprocal advantage to it. Non-hunting mates starve unless they provide some useful service.

      Your argument says that the Aldas will be a larger, more successful tribe. They'll have more successful hunts, and they'll lose fewer members to accidents. In the real world, this is generally the case. Hunter-gatherer societies mostly behave like the Aldas, not like the Lecters.

      But this is hard to explain theoretically, given how natural selection operates. If even ONE Lecter joins an Alda tribe, his Lecter genes give him a significant advantage over the competition. They're out risking their lives for each other and cooperating; he receives all the benefits but takes none of the drawbacks. Some very simple math shows that it only takes a tiny small reproductive advantage for the Lecter gene to drive the Alda gene right out of the population.

      Even if there are no Lecters at all, your argument only works if 100% of the Aldas are at an equal level of cooperativeness vs. selfishness. Due to natural variabilities, this isn't true. Some Alda children will turn out just a little less cooperative, just a little more selfish, than others. These children have the same reproductive advantage: they help out a bit less, risk their own lives a bit less, and take a bit more than their fair share. This sets off a kind of genetic arms race, because among their children, the slightly-less-cooperative ones will have the same advantage. This repeats itself every generation.

      So the theoretical result is that Lecter-like genes should win out over Alda-like genes. The actual, observed result is quite different. The argument is over how to explain this discrepancy in terms of natural selection.

      Delete
    8. In prehistoric times the mammoth burgers were necessary for survival. The Lecters literally would not survive. I can't speak to the modeling, but without cooperation, we hoomans are pretty weak. When we compete with the sabertooths of the world, it is our only advantage.

      Without the ability to know the modeling, I don't think I can take this any further, but I suspect that until we as a species had more control over our environment, we needed the trust trait. After that, we had enough control over the environment that we could support the trusters along with the Lecters. The genes stay in the pool.

      Delete
    9. But in generation 2 of the Aldas, the Aldas who are just a tiny bit more Lecter-like have an advantage, and those genes prosper; there are more of their children in generation 3. And in generation 3, the grandchildren who are just a tiny bit more Lecter-like have an advantage ... and so on. Logically, this can only end in one way.

      It's far from the only example of a trait that's advantage to inividual success being bad for the species success. Look up the Irish Elk some time, for example. The short version of the story goes like this:

      1. A few male elk are born (quite randomly) with larger sets of antlers. They're better at fending off predators. The big-antler gene begins to dominate.

      2. A few female elk are born (again, quite randomly) with an attraction for elk guys with really big, manly antlers. Those females are more likely to mate with the male elks from 1., so that gene is preferred as well.

      3. A couple generations down the road, all the male elk have big antlers. Some of them (yet again, quite randomly), are especially massively endowed.

      4. The female gene that prefers larger antlers has also spread. So the females mate preferentially with the biggest of the big.

      5. At some point the antlers stop being useful and start being something of a burden. But the female elk, by this time, universally have the bigger-is-better gene when they're selecting their mates. So it doesn't matter if Medium Marvin Elk has antlers that are just the right size for fending off wolves. None of the girls will give him a second glance. They're all swooning over Ginormous George Elk and Humongous Harold Elk.

      6. Eventually this scenario results in elk with antlers that are comically large. The antlers grow so large that they're an actual disadvantage. Result: the Irish Elk is extinct.

      The more Lecter-like Aldas are like the slightly-larger-antlered elk. Evolution should inexorably lead to the full Lectrization of the tribe, just as all of the Irish Elk herds acquired titanic antlers.

      To say that "we needed the trust trait" is to mis-state how natural selection works. To your genes, there is no "we". If I do better than you, then more of my genes survive. If I can do better by disadvantaging you at the same time, so much the better.

      Your statements are all true, and they're all consistent with observation, but the result is still a puzzle with respect to how natural selection works. Hence the argument between the supporters of kinship selection and those, like Edward Wilson, of multi-level selection.

      Delete