Inclusive fitness, models, and religious evolution

Connor Wood

One of the great things about studying religion is that it’s a huge evolutionary puzzle. If you’re the type who likes puzzles, you could sign up right now for a career in the evolutionary study of religion and probably never be bored again for the rest of your life. The riddles abound: Why have we (apparently) evolved the capacity for profound religious experiences? Is there an evolutionary function for spirits, gods, or religious rituals? Many researchers argue that religion is a functionless byproduct of other evolutionary developments, while others claim that religion is a useful adaptation that helps human groups survive. Funnily enough, one recent paper sheds light on this debate despite not mentioning religion at all.

The paper, entitled “Limits of Inclusive Fitness,” appeared in the Proceedings of the National Academy of Sciences (PNAS) earlier this fall. Its authors – Ben Allen, Martin Nowak, and E.O. Wilson – are evolutionary scientists at Emmanuel College and Harvard. (Full disclosure: Ben is also a friend of mine.) The paper is a mathematical critique of one of the most influential theoretical contributions to 20th-century evolutionary biology: inclusive fitness, or the claim that an organisms’ total “fitness” is an additive combination of its own success and its close relatives’ success at the game of life.

A quick background: Inclusive fitness theory was developed by a number of evolutionary scientists in the mid-20th century to answer the question of how costly altruism could possibly evolve in a Darwinian world such as our own. How could Darwinian theory explain the fact that animals (and humans) often apparently sacrifice their own genetic well-being to benefit their peers? Shouldn’t evolutionary pressures force everyone to look out for their own interests first? The most common example of this conundrum is in the colonies of eusocial insects such as ants, where sterile workers sacrifice their entire reproductive careers to serve the queen and her offspring. But costly altruism also famously occurs in humans, who often sacrifice their lives for one another or otherwise put the needs of their group above their own genetic interests.

The person most often connected with inclusive fitness is W.D. Hamilton (1936-2000), a British evolutionary biologist. In 1964, Hamilton published an early version of a formula that attempted to explain such apparently confusing social phenomena. Known as Hamilton’s Rule, the formula models altruistic sacrifice as actually beneficial to the organism doing the sacrificing – as long as the beneficiary is a fairly close relative. The expression goes like this:

rB > C

Let me sum up: The “r” means the “coefficient of relatedness,” or the degree to which two individuals are related to each other. You and a clone of yourself would have an “r” of 1, since your genes would be fully identical. But your coefficient of relatedness with a non-twin sibling is roughly .5, because about half of any genes in your genome are from the same parent. The coefficient of relatedness drops off fast; your “r” with a first cousin is only .125, for example.

The “B” refers to benefit – specifically, the benefit to the relative who’s receiving an altruistic boon. For example, let’s say that you rush into a fiery house to save your sibling, and this boosts your sibling’s chances of having children from zero – dead people don’t have children – to, say, 60%. This means that the “B” value would be pretty high!

In contrast, the “C” refers to the “cost” to the organism who gives the help. Let’s say that you get burned while saving your sibling. This reduces your probable lifetime fitness, because it’s harder to get a mate and good jobs with a scarred face. The “C” in Hamilton’s Rule quantifies this damage to your lifetime fitness.

Hamilton postulated that the only time organisms could evolve costly, sacrificial behavior would be when the partners are closely related enough, and the increase in beneficiaries’ fitness is great enough, to override the costs of the sacrifice to the benefactors. For example, Hamilton’s Rule predicts that ants will be closely related enough to their queen that their giving up reproduction will actually pay off genetically in the long term – since the little ant eggs they’re raising carry enough of their own genes to make up for their sacrifice.

Hamilton’s Rule and inclusive fitness theory were the uncontested champions in the struggle to explain costly altruism in biology for nearly a half-century. But in the past decade, a number of scientists have been challenging inclusive fitness, often – but not always – on mathematical grounds. This is where the PNAS paper comes in.

To (over)simplify things, Ben, Nowak, and Wilson argue that inclusive fitness theory depends too heavily on linear regression models to explain how genetic relatedness affects evolution. Regression analysis is a statistical technique that identifies the different factors influencing an observed outcome. For example, let’s say that students from the west side of your town are 30% more likely to attend college than kids from the east side. A regression analysis finds that several different factors combine to predict this difference: for example, the west side is wealthier than the east side; there are more native English speakers on the west side; and college recruiters visit west side high schools more often than east side schools. Together, these factors add up to explain the higher college attendance rate on the west side.

Similarly, evolutionary biologists have long been in the habit of using linear regression models to separate organisms’ genetic fitness into components. For example, 50% of an organism’s total genetic fitness might be due to the offspring it raises itself, and 25% due to the offspring its siblings help it raise, and 12% to the offspring its cousins help it raise. The authors of the PNAS paper argue, roughly, that this is an unhelpful way of modeling genetic fitness. To quote Ben’s blog post on the paper:

At this point you may be asking “Wait, does it really make sense to divide offspring into those  produced on one’s own versus those produced by help from others?” This is exactly the problem! Aside from the obvious point that no one reproduces without help in sexual species, nature is full of synergistic and nonlinear interactions, so that making clean divisions like this is impossible in most situations. Thus the idea of inclusive fitness theory only works in simplified toy models of reality.

So, Ben and the other authors claim, the inclusive fitness model is just not close enough to reality to be useful. The second major problem with the mathematics behind inclusive fitness theory, they argue, is that such mathematics can – and does! – lead researchers to make false attributions of causality when only correlations are evident. The paper offers some elucidating hypothetical examples, including the “hanger-on:” an organism that has a gene which drives it to find high-fitness individuals and hang out with them. The hanger-on doesn’t get any extra fitness benefit from palling around with its high-fitness peers, but an observer looking at the long-term population data would always find that individuals with the “hanger-on” gene were found closely associated with individuals who had lots of offspring. An inclusive fitness analysis would conclude that the “hanger-on” gene was actually a helping gene – one that inspired individuals to sacrifice their own well-being in order to help nearby partners raise their own offspring. Of course, this would be exactly the wrong explanation: the hanger-on doesn’t cause the high-fitness partner to have offspring. The high-fitness parter causes the hanger-on to glom on.

These and related arguments lead the authors of the PNAS paper to conclude that the right way to investigate the evolution of altruism is using old-fashioned Darwinian techniques based on natural selection models. Inclusive fitness, the authors argue, requires too many assumptions to work, doesn’t make good predictions or offer explanations, and is too general to be applied to any real circumstances.

Now, what does all of this have to do with religion? Surprisingly, a lot. The main opponents of inclusive fitness theory have tended to be proponents of various sorts of group selection – a broad term that generally refers to the ability of Darwinian processes to operate not only at the level of the gene, but also at the collective. For example, let’s say two tribes occupy the same valley, but one of them has a preponderance of genes that encourage costly cooperation within the group. Over generations, the tribe whose members cooperate better with one another slowly take over and push out the other tribe. The costly-cooperation gene hurts the individuals who carry it within their own groups (since they’re always sacrificing their own well-being to help their fellows), but helps the group as a whole (since the group functions better if it has lots of members who are willing to sacrifice for one another). The result? Over time, the costly-cooperation gene comes to dominate the valley, even though it imposes fitness costs on the individuals who have it.

Several writers, including Martin Nowak and David Sloan Wilson, have suggested that group selection might be a better explanation for the evolution of morality than inclusive fitness. This might seem plausible to an educated outsider, but there’s a problem: many of these writers also use group selection to argue for the evolutionary utility of religion. In his 2002 book Darwin’s Cathedral, for example, D.S. Wilson argued that religion overwhelmingly helps groups survive, and that religious practices and beliefs – as well as the costly altruism they encourage – have evolved by group selection.

On the other side of the equation, the proponents of inclusive fitness theory have tended to be among the most vociferous critics of religion writing today. Steven Pinker, Richard Dawkins, Jerry Coyne, and Daniel Dennett are all on record as being strongly pro-inclusive fitness and very much opposed to group selection theories. There are both personal and professional commitments on the line here; to pick an example, Hamilton’s memorial service in 2000 was organized by Richard Dawkins.

Now, I’m not claiming that Richard Dawkins’s personal friendship with and admiration for W.D. Hamilton means that Dawkins’s support for inclusive fitness is automatically suspect. I happen to personally side with the religious adaptationists – those who argue that religious phenomena are dedicated adaptations – but I’m not convinced that group selection is the way to get there. And there is evidence that inclusive fitness theory works very well in some situations; for example, in explaining why red squirrels adopt some orphaned babies but not others (hint: they don’t adopt babies whose “r” coefficient doesn’t meet the predictions of Hamilton’s Rule).

The authors of the PNAS paper, including my buddy Ben, wouldn’t contest the squirrel finding; they freely admit that sometimes inclusive fitness theory works. Instead, they argue that it’s simply not a universally generalizable rule. They have some very good points, and I think they’re probably right. But what I’m most interested in is how the debates around group selection illustrate how scientific and ideological commitments entwine and influence one another, no matter what the textbooks say. It is not coincidence that the ranks of the anti-group selectionists are a who’s who of the New Atheists. It’s also not coincidence that the group selection advocates tend to argue for religion’s adaptive value.

And finally, it’s not a failure of science that researchers’ ideological and theoretical commitments are sometimes aligned; people have to get their convictions from somewhere. If Daniel Dennett, say, believes that group selection is bunk because he’s afraid that group selection will prove that religion is adaptive rather than a worthless – or harmful! – byproduct, I think that’s great. Research and debate will eventually prove whether he’s right or wrong, regardless of where his commitments come from. This is one of science’s most admirable, and valuable, traits: hypotheses and predictions can come from anywhere, and testing and observation will eventually separate the wheat from the chaff. Ben’s paper in PNAS is just one more fine example of how this winnowing process works.

  • http://triangulations.wordpress.com/ Sabio Lantz

    @ Connor,

    Don’t know if you read Chris Campbell at “Genealogy of Religion“. I can’t find anyone more exciting to explore the edges of what religion means and how it think about it anthropologically and genetically.

    Anyway, Chris is not sympathetic to “group selection theory”, see his applause to Pinker, here. Chris’ specialty is hunter-gather societies and the theory of religion. You may enjoy him.

    Chris is not a theist, but he is sympathetic to some things that get put under the term “religion”. You may find he stretches your categories.

    Concerning this post, you said,

    The second major problem with the mathematics behind inclusive fitness theory, they argue, is that such mathematics can – and does! – lead researchers to make false attributions of causality when only correlations are evident.

    Important correction: It is not the mathematics, but the overgeneralizing or misapplication of the math. The mathematics does not lead the researchers.

    You also said,

    So, Ben and the other authors claim, the inclusive fitness model is just not close enough to reality to be useful.

    But later you said,

    The authors of the PNAS paper, including my buddy Ben, wouldn’t contest the squirrel finding; they freely admit that sometimes inclusive fitness theory works.

    So, it isn’t useful or it works [sometimes]. Boy is that confusing. Point is, the math works fine, it does not mislead, correct application is important to make something consistently useful.

    In the end, you sound like you disagree with group theory too but you want to support your friend’s paper. I agree when you said:

    And finally, it’s not a failure of science that researchers’ ideological and theoretical commitments are sometimes aligned; people have to get their convictions from somewhere.

    .

  • Pofarmer

    Would human mental facuties tend to short out Inclusiveness theory that works well in lesser species? We can obviously reason ourselves into and out of thing, and we show empathy, which is something no other species does, to my knowledge, or would empathy itself be an evolutionary developement?

  • Ben Allen

    A very interesting perspective on the controversy!

    It is definitely noteworthy—and as you say, probably not
    coincidental—that inclusive fitness champions overlap so much with New
    Athiests, whereas group selection champions tend to be friendlier
    toward religion. I think a lot of this is due to the influence of
    Dawkins himself. But it’s also true that inclusive fitness has a
    calculating, “rational” feel to it—you just add up amounts of help times gene-sharing fractions and that tells you what you should do. Group selection, on the other hand is more “feel-good-y”, in that it involves individuals working together for a collective good.

    Of course, these are both just mathematical frameworks with no intrinsic implications for ethics or values. But it’s inevitable that humans attach meanings to these equations—especially since the underlying question goes to the root of human morality.

    I am also perhaps not as rosy as you are about the ability of
    scientific debate to overcome ideological commitments. My go-to example
    on this is eugenics or “race science”. It wasn’t advances in science
    that led to the discrediting of these theories; it was advances in our
    societal conscience.

  • Ben Allen

    To clarify, my position is the following:

    As a quantitative law, inclusive fitness theory holds only under idealized conditions that are almost never met in the real world. Attempts to generalize the theory beyond these conditions lead to nebulous quantities devoid of biological meaning.

    As a heuristic, inclusive fitness reasoning can sometimes lead to useful insights. But these insights must be tested with rigorous mathematics and/or empirical observation.

  • http://triangulations.wordpress.com/ Sabio Lantz

    And do tell, Ben, why should we take your word on that generalization? Would someone who saw inclusive fitness theory as highly usable under non-idealized conditions agree with you?

    First, Generalizing ANY theory incorrectly (as related to tested group — sampling size and method etc) makes that generalization “lead to nebulous quantities devoid of biological meaning.”

    This sounds more like rhetoric than any attempt at being mathematically concrete and careful. And I am just a layman.

  • http://triangulations.wordpress.com/ Sabio Lantz

    And BTW, Ben, is this you:

    http://www.emmanuel.edu/academics/our-faculty/benjamin-allen.html

    Just want to be clear that is you at a Catholic College with its strong Catholic purposes. Just saying, as long as you are dancing on the edge of the genetic fallacy with the “New Atheists”.

  • Carl Veller

    Two points (and I suppose I should note that I’m a student working in Martin Nowak’s program):

    First, the squirrel study emphatically does not show that “inclusive fitness theory works well in some situations”. The sample size is too small (five adoptions, if I recall) and the data not finely calibrated enough (“adopt” or “not adopt”) to properly test the precise quantitative predictions of inclusive fitness theory. Sure, if we take the cost and benefit calculations at face value (and given the small sample size, they’ll be very imprecisely estimated), the results don’t contradict the predictions of inclusive fitness theory, but they also don’t contradict the predictions of a whole host of other theories that are similar in spirit but quantitatively different to inclusive fitness theory. A more conservative reading of the study (and basically every other study purporting to ‘prove’ Hamilton’s rule empirically) is that it’s evidence that increasing relatedness is associated with more altruistic behaviour. This is a property of inclusive fitness models, sure, but also of an infinite number of other models. That’s what Ben means by using inclusive fitness as a heuristic, I think. It captures a qualitative logic that we think is going on, so very broadly will be consistent with observed phenomena, but not necessarily precisely.

    Second, while I’m a big fan of all of Dawkins, Coyne, Dennett, and Pinker, none of them would claim to be particularly mathematically sophisticated, and none of them has ever attempted a response to the mathematical points made by Nowak and Tarnita. Dawkins, for example, offered the wonderfully circular “If you think, as Nowak et al. do, that ‘Hamilton’s rule almost never holds’, that simply means you haven’t been measuring B and C carefully enough”! While Gardner, West, etc, might agree with Dawkins on this, I would guess it’s for different, slightly less crude, regression-method reasons, which themselves suffer from a form of circularity, as the Allen, Nowak and Wilson paper makes clear. The value of this paper is that it explains in very clear terms and examples (which should be understandable to non-mathematical biologists like Coyne and Dawkins) what the problems with the current theory of inclusive fitness are. I’m very curious to see what the responses to the paper are.

  • http://triangulations.wordpress.com/ Sabio Lantz

    I saw this quote today on The Browser by a statistician today:

    “All models are wrong, but some are useful” — George Box (1919-2013)

  • connorwood

    Thanks for the corrections about the red squirrel study, Carl. When it comes to the debates between the group selectionists and their opponents, my read on things has been that what matters at the larger level is whether cooperation ultimately depends on degrees of relatedness, or whether cooperation can be stabilized through other means – specifically, intergroup competition or through having nonrelated individuals otherwise share group-level outcomes that depend on cooperative endeavor. I agree that the red squirrel study is too small to prove much (and if I’d realized so beforehand I wouldn’t have included it), but the point I was making is that a lot of cooperation in nature really does depend on degrees of relatedness – which you substantiate. So I think the correct interpretation of the squirrel study has a lot of bearing on which mathematical heuristic is the right one for modeling kin-based cooperation, but not as much on the question many people in my field are interested in: does cooperation evolve independently of relatedness, particularly in humans?

    And I completely agree about the lack of real mathematical responses to Nowak and Tarnita’s points. It does raise questions, in my mind, when a scientific community feels the need to sign a huge petition rather than simply challenge a new argument on its own merits. It’ll be interesting indeed to see what the reactions turn out to be!

  • GCBill

    I’m a bit confused how this:

    But what I’m most interested in is how the debates around group selection illustrate how scientific and ideological commitments entwine and influence one another, no matter what the textbooks say. It is not coincidence that the ranks of the anti-group selectionists are a who’s who of the New Atheists. It’s also not coincidence that the group selection advocates tend to argue for religion’s adaptive value.

    …manages to appear in the same article that notes:

    The second major problem with the mathematics behind inclusive fitness theory, they argue, is that such mathematics can – and does! – lead researchers to make false attributions of causality when only correlations are evident.

    You’re sure that these people have selected their explanations of altruism based on their worldviews, and not modified their worldviews in accordance with their theories of altruism? Recall, for instance, that (as far as I know) Dawkins wrote The Selfish Gene long before indulging in any public criticism of religion. If Dawkins had been convinced of a different theory that cast religious views in a more benevolent, adaptationist light, perhaps he wouldn’t be an antitheist? I’m willing to bet that the group selection folks (whose views I am admittedly not that familiar with) followed a similar logical process as the one I outline as well.

    Of course I might be wrong, but the claim that these people’s views are “no coincidence” is hardly evidence of ideological contamination on its own, especially when one’s chosen explanation entails certain ideological commitments that could as easily flow from it as cause it.

  • connorwood

    >You’re sure that these people have selected their explanations of altruism based on their worldviews, and not modified their worldviews in accordance with their theories of altruism?

    >the claim that these people’s views are “no coincidence” is hardly evidence of ideological contamination

    I did not claim that anyone’s view was “contaminated,” nor did I specify a direction of causality. I pointed out, as Ben does in his comment on this post, that there are certain conceptual and affective affinities between certain stances toward religion and certain perspectives on the evolution of altruism. Those relationships are self-perpetuating feedback loops in which a single point of origin – a chicken before the egg – can’t be found. And so a careful re-reading of this post will show that you’re responding to a mistake that I didn’t make.

  • connorwood

    Lots of other species show empathy. Haven’t you ever seen a dog get sad when its owner is sick? Lab rats have been shown to become distressed when witnessing other rats in pain. And so forth. And yes, empathy is an evolutionary development.

  • http://youtube.com/user/BowmanFarm Brian Bowman

    Thanks for this post. I have Wilson’s Darwin’s Cathedral, and I like how it fits with Quinn’s observation on why Salvationist religions began.

  • Pofarmer

    O.K. But, I think it’s still a valid question. Would human interactions and reasoning short out a strictly mathematical model that works well in lower species?

  • connorwood

    You’re asking whether human biocultural evolution operates under different mathematical rules than those governing other types of evolution, or whether math just can’t be used to model human evolution? I think the answer to both is that we don’t know yet, because we don’t know yet which mathematical models are best for general evolutionary problems, especially those having to do with cooperation. My guess is that human evolution is going to be harder to model than that of other species, partly because of the complexity and agency you mention. We’re just pretty hard to predict, no matter what the evolutionary psychologists say. Still, I think that mathematical evolutionary models will be useful in understanding human evolution. They’ll just have to be more complex and better able to deal with cultural variation than models used for animal evolution. (My advisor does some computer modeling in this area, specifically in how cultures transmit attitudes and norms, and is achieving good results.)

  • Pofarmer

    Thank you.

  • amanimal

    My apologies Connor, I’d meant to post a comment to this and was side-tracked and never got back here. My current thinking lies along the lines of:

    ‘Evolutionary Perspectives on Religion: An Overview and Synthesis’, Smith/Arrow 2010
    http://www.evostudies.org/pdf/SmithVol2Iss2.pdf

    If you’re not already familiar you could probably get away with reading the abstract, last 2 paragraphs of the intro, then skip to ‘The Integrated Model’ on page 13

    … as laid out in the stair-case analogy or in:

    ‘The Evolution of Religion: How Cognitive By-Products, Adaptive Learning Heuristics, Ritual Displays, and Group Competition Generate Deep Commitments to Prosocial Religions’, Atran/Henrich 2010
    http://www.artisresearch.com/articles/Atran_The_Evolution_of_Religion.pdf

    … which you’re no doubt familiar with. I did read Bulbulia’s 2007 contribution to ‘The Evolution of Religion’ and have to say it was less than persuasive though it may well be my prejudice.

  • connorwood

    Thanks for the links! Bulbulia’s work since then has gotten more sophisticated and insightful, I think – check out his collaborations with Richard Sosis of UConn, in which they postulate a “charismatic ecology” of religious symbols to solve risky coordination problems.

    I’m looking forward to checking out the Smith/Arrow paper.

  • amanimal

    Thanks Connor, I definitely will!

  • amanimal

    Thanks Connor – took me back to several of your pieces, the Uffe Schjødt et al 2013 paper, Bulbulia/Frean 2010,and just watched the Bali water temple video – next up is ‘Signalling theory and the evolution of religious cooperation’, Bulbulia/Sosis 2011.

    I’m just an interested amatuer who likes reading somewhat over my head. I’ve a passing familiarity with costly signaling theory, but much of this is new to me, especially the automaticity aspect. Thanks again!

  • connorwood

    Sounds like you’ve got more familiarity with this literature than many professionals! Happy reading!

  • Y. A. Warren

    It seems that all the scientific study is based on procreation of species. What if the ultimate goal of humanity is to return all energy back to Universal Eternal Energy through making physical procreation obsolete as a way to create community?

  • Pingback: URL

  • Pingback: Great opportunities to make money online


CLOSE | X

HIDE | X