July 9, 2012 | 25
Prominent scientists are in a bitter struggle over the origins of kindness. But the root of this conflict may be the most ironic part of all.
What would it take for you to give your life to save another? The answer of course is two siblings or eight cousins, that is, if you’re thinking like a geneticist. This famous quip, attributed to the British biologist J.B.S. Haldane, is based on the premise that you share on average 50% of your genes with a brother or sister and 12.5% with a cousin. For altruism to be worth the cost it should ensure that you break even, genetically speaking.
This basic idea was later formalized by the evolutionary theorist William Hamilton as “inclusive fitness theory” that extended Darwin’s definition of fitness–the total number of offspring produced–to also include the offspring of close relatives. Hamilton’s model has been highly influential, particularly for Oxford evolutionary biologist Richard Dawkins who spent considerable time discussing its implications in his 1976 book The Selfish Gene. But in the last few years an academic turf war has developed pitting the supporters of inclusive fitness theory (better known as kin selection) against a handful of upstarts advocating what is known as group selection, the idea that evolutionary pressures act not only on individual organisms but also at the level of the social group.
The latest row was sparked by the publication of Edward O. Wilson’s new book, The Social Conquest of Earth, which followed up on his 2010 paper in the journal Nature written with theoretical biologists Martin Nowak and Corina Tarniţă. In both cases Wilson opposes kin selection theory in favor of the group selection model. For a revered scientist like Wilson–a Harvard biologist, recipient of the Crafoord Prize (the Nobel of the biosciences) and two-time Pulitzer prizewinner–to adopt a marginal and widely disputed concept has received a lot of attention and caused other prominent scientists to step forward and defend the mainstream point of view.
For example, writing at The Prospect magazine, in what The Guardian newspaper called “a searingly critical review,” Dawkins argued that the proposal in Wilson’s book was based on “erroneous and downright perverse misunderstandings of evolutionary theory.” Joining him at the website Edge was Harvard psychologist Steven Pinker who wrote that group selection was a “false allure” and “a loose metaphor, more like the struggle among kinds of tires or telephones.” Likewise, University of Chicago biologist Jerry Coyne dismissed group selection on his blog as “a fuzzy and nebulous concept” and one that merely “has an innate appeal to those with a penchant for the religious and the spiritual.” It should go without saying that online commenters were considerably less kind (a notable exception being at Edge, where scholars were invited to comment independently).
Taken together, along with the 137 scientists who signed a letter to Nature in 2010 supporting kin selection, this would seem to be the coup de grâce effectively sending group selection the way of the dodo. But, at the same time, it seems odd that so many prominent scientists would feel the need to forcefully defend what they all maintain is an irrefutable, textbook understanding of evolutionary biology. After all, science advances based on empirical evidence, not rhetoric (“An ounce of algebra is worth a ton of verbal argument,” as Haldane noted). Shouldn’t it therefore be an easy task to simply examine the evidence for group selection and leave it at that? Yes, it should in theory. But this is where things get complicated.
“The root of the problem is the existence of several different frameworks for modelling the evolution of social behaviour,” says Samir Okasha, philosopher of science at Bristol University in England, writing in the October 7, 2010 edition of Nature. “The relationships between these frameworks are sometimes ambiguous, and biologists disagree about which is most fundamental and which most useful empirically.”
In other words, the evolution of altruism is under dispute because of how it it is being measured. Therefore, in order to understand what the conflict is fundamentally about it is necessary to go back to the theoretical basis for these different evolutionary concepts. This, of course, leads us to honeybees.
The Sting of Altruism
When I was a boy, long before my professional interests in evolutionary anthropology and the history of science, my friends and I would frequently take hikes in the canyon just beneath the small mountain town in northern California where I grew up. At the base of the incline, bordered in the summer by golden fields of blooming Centaurea solstitialis flowers, was a little heart-shaped lake that you could see from my house. But to reach the lake and swim meant crossing through several football fields of these flowers, known as yellow star-thistles, whose stems were ringed with inch long spines that felt like needles on our hairless legs.
As if that weren’t enough, the owner of the land thought he’d take advantage of this invasive weed on his property by establishing beehives in the nearby forest shade. My friends and I used to joke that you could only tell the difference between getting stung by a bee or stuck with a spine because a bee sting left a blue welt afterwards. Once, as we were wading through these star-thistles on a hot summer day, a bee stung my nextdoor neighbor. As he stood there screaming we all turned to watch in horror as the bee remained stuck to his arm, buzzing incessantly. After what seemed like minutes (but was probably only about ten seconds) the insect broke free only to dive bomb to the ground dead, her viscera yanked from her body on this suicide mission to protect the colony.
What I didn’t know at the time was that the altruism displayed by the eusocial insects in the Hymenoptera (bees, wasps, and ants) was an enormous problem for Charles Darwin. As he wrote in On the Origin of Species, an act of self-sacrifice that only served to benefit another was a “special difficulty” that could potentially undermine his theory of natural selection.
The central premise of Darwin’s theory was that all characteristics of a species–whether physical, like the elaborate antlers of an Irish Elk, or behavioral, like the formation of a V-shaped flock in migratory geese–were traits that had evolved through successive, slight modifications passed down over many generations. Because these modifications would only be passed on if they were beneficial, any trait that brought harm to their possessor would ultimately be discarded. “Natural selection acts solely by and for the good of each,” Darwin wrote. Therefore, any characteristic that violated this premise “would be absolutely fatal to my theory.” With the eusocial Hymenoptera, not only do individuals sacrifice themselves for the group, the vast majority of colony members have given up reproduction altogether.
Darwin’s solution to the problem is what today would be called multilevel selection in which certain traits are selected because they are advantageous at the individual level while others are advantageous at the family or group level. In both cases, the trait is selected because it allowed more offspring to be born who carried that particular trait. As Darwin wrote:
[W]e can perhaps understand how it is that the use of the sting should so often cause the insect’s own death: for if on the whole the power of stinging be useful to the community, it will fulfill all the requirements of natural selection, though it may cause the death of some few members.
It would take more than a century for Darwin’s “special difficulty” of altruism to be tested empirically and, when it was, the great naturalist was considered to be only half right.
While Darwin was a gentleman scholar, a man whose social life was concerned with pedigree and the ties between aristocratic families, William Hamilton was a rugged individualist. This highly influential British biologist, whom Richard Dawkins has called “a good candidate for the title of the most distinguished Darwinian since Darwin,” was just as comfortable hacking his way through the jungles of Brazil with a machete as he was calculating sex-ratios among fruit flies. At public lectures Hamilton was fond of telling the story about how he once jumped overboard to plug a hole with his finger while on a field trip down the Amazon, dryly noting that the dangers of piranhas were much overrated.
In 1964 Hamilton published two papers titled “The Genetical Evolution of Social Behavior” that revolutionized the field of evolutionary biology by suggesting a “gene’s eye” view of the world. By looking at the interests of genes, not just individuals, Hamilton argued that the altruism of honeybees could be explained by calculating the likelihood that an individual’s genes would be shared by close relatives. His equation was beautiful in its simplicity. By multiplying the genetic relatedness of the individual being helped (r) by the reproductive benefit received (B) the equation could predict whether or not it would be worth the reproductive cost (C) incurred as a result, or rB > C. Just like J.B.S. Haldane’s anecdote in the introduction, the evolution of altruism was ultimately a numbers game. If you wanted to maximize your genetic stock portfolio, nepotism paid the highest dividends.
However, the real breakthrough was when this equation was applied to the unique reproductive system of the eusocial Hymenoptera, what biologists refer to as haplodiploidy. In honeybees, for example, male drones receive all of their genes from their mother (haploid) while female workers receive half of their genes from their mother and half from their father (diploid). If queens mate with only one male it means that, on average, they will share 50% of their genes with daughters but each daughter will share 75% of these genes with their sisters (all of their father’s genes and half of their mother’s). Female honeybees are therefore genetically closer to their sisters than they would be to their own offspring. Rather than breed themselves it was in their genetic interest to be sterile and, in Hamilton’s words, use their mother as a “sister-producing machine.”
This haplodiploidy breeding system, when combined with Hamilton’s equation, also meant that a female worker bee would be predicted to accept enormous costs, even death, to promote the interests of her sisters. After all, what use is your own genetic endowment when you’ve got hundreds, or even thousands, of sisters back at the hive that each share three-quarters of your hereditary sum? The moral math is unambiguous and that group of boys approaching through the star-thistles could mean trouble.
But there was one scientist in 1960s England who was deeply troubled by Hamilton’s equation. George R. Price, a scraggly young chemist and divorcé with a brooding penchant for existential angst, had traveled from New York to London to create a new life for himself. When he encountered Hamilton’s papers in the Journal of Theoretical Biology he decided that altruism was just the question on which he would stake his claim. Perhaps it was because he had abandoned his two daughters, or perhaps it was because he held anti-war views in a world teetering between rival superpowers, but Price had serious concerns about a model for altruism that was based on nothing more than selfish nepotism. He became determined to prove Hamilton wrong.
Thus began one of the most profoundly strange relationships in scientific history (see the wonderful book The Price of Altruism by Oren Harman for more on this fascinating story). By teaching himself the mathematics necessary for his task Price found inspiration in the concept of covariance in order to show how a trait would evolve from one generation to the next. He designed an equation that calculated how a specific trait would covary with its fitness (i.e. the number of copies passed on). While this was little more than an accounting method, what it meant was that altruistic genes could therefore be tracked without relying on relatedness. In essence, Price revealed that Hamilton’s equation was incomplete.
However, Price didn’t stop there. By then calculating what the likelihood was that a trait would be passed on, the Price equation could predict how a trait would evolve over time and could also track multiple levels of selection (from gene to individual and from individual to group). A year after Price had begun working he finally had his answer and wrote to Hamilton about the covariance equation he’d derived. Hamilton called him the very next day.
As Hamilton recalled in his memoir Narrow Roads of Gene Land (p. 172-73), he listened as a squeaky voice on the other end of the line asked him, “Have you seen how my formula works for group selection?”
A Group By Any Other Name
In his book Naturalist, first published in 1994, E.O. Wilson describes himself as a social conservative in both politics and morality. “I cherish traditional institutions, the more venerable and ritual-laden the better,” he says. While Wilson may hold firm to the bedrock of traditional values in his personal life, in his latest scientific work he has taken on the role of a bomb-thrower.
Wilson was once one of the central proponents of kin selection theory, even standing side by side with Hamilton in front of a hostile academic crowd to defend the idea. But now he has abandoned this model and adopted what many perceive to be its mirror opposite. As he writes in The Social Conquest of Earth:
“Inclusive fitness is a special mathematical approach with so many limitations as to make it inoperable. It is not a general evolutionary theory as widely believed, and it characterizes neither the dynamics of evolution nor the distribution of gene frequencies” (p. 180).
However, one important reason for his change is the very reason he adopted kin selection in the first place: the haplodiploidy reproductive system of the eusocial Hymenoptera. At the time when Hamilton applied his equation to these insects the evidence seemed clear that queens would only mate with a single male and it was this that allowed such a high level of relatedness between sisters. However, if you increased the number of matings all bets were off. Sisters might then be no more related to each other than they would be to their own offspring and sterility wouldn’t make evolutionary sense.
As Wilson points out, there have been many studies demonstrating that multiple matings occur in queens among the eusocial Hymenoptera. For example, one study published in Proceedings of the Royal Society by Arnaud Estoup and colleagues in 1994 found that in five different colonies of honeybees (Apis mellifera) there were between 7 and 20 fathers, making female workers related to their sisters by only about 30%. Likewise, another study published in Proceedings in 1999 led by Jacobus Boomsma found that queens in the leaf-cutter ant (Acromyrmex octospinosus) mated with between 4 and 10 males, giving a genetic relationship between sisters of about 33%. Another study published in the journal Evolution by Michael Goodisman and colleagues found that queens in the eastern yellow jacket (Vespula maculifrons) mate with between 3 and 8 males. Further research has suggested that genetic diversity within a colony is important for resisting disease. However, not all biologists think these cases ultimately challenge kin selection in the eusocial Hymenoptera.
Wilson also points out that multiple other eusocial species have been discovered since Hamilton’s study that do not have a haplodiploid reproductive system, including termites (previously thought to be explained under the kin selection model). Others include a species of platypodid ambrosia beetle, several lines of synalpheid sponge-dwelling shrimp, as well as two species of bathyergid mole rat.
“The result was that the connection between haplodiploidy and eusociality fell below statistical significance,” Wilson writes. “Consequently the haplodiploid hypothesis has now been generally abandoned by researchers on social insects” (p. 170).
Wilson’s solution then is to also abandon kin selection as a category and explain eusocialty through multilevel selection like Darwin did. Since individuals are expected to adapt behavioral strategies that maximize their fitness, species that live in groups have done so because it was in each individuals reproductive interest. However, because there will always be a tension for group-living species between what can be called selfish behaviors and groupish behaviors, multilevel selection means that different traits will be subject to different selection pressures towards one pole or the other. For most species the result will end up being a compromise that benefits each individuals overall fitness in relation to their group. However, in special cases, natural selection will push a species towards one extreme or the other.
As E.O. Wilson explained in a paper co-authored with David Sloan Wilson (no relation) in 2007 for the Quarterly Review of Biology (pdf here), multilevel selection is best understood by using what are referred to as vectors in physics. The simplest way to visualize this is to imagine a toy boat on one side of a river that has two strings being pulled simultaneously by children at different points on the other side. If both children pull with the same amount of force the boat will end up directly between them. However, if one child is larger and pulls the string with more force, the boat will end up closer to one than the other.
Wilson therefore believes that abandoning Hamilton’s equation and embracing a multilevel approach like the Price equation would be the most reasonable, and parsimonious, way to move forward.
“If there is a general theory that works for everything (multilevel natural selection) and a theory that works only for some cases (kin selection), and in the few cases where the latter works it agrees with the general theory of multilevel selection, why not simply stay with the general theory everywhere?” (p. 175).
Ironically, most of Wilson’s critics tend to agree, they just use this same argument for retaining kin selection and rejecting group selection. The three most common critiques of Wilson’s position in the scientific literature are that: 1) Group selection models are functionally equivalent to kin selection (Lehmann, 2007; Marshall, 2011) 2) Kin selection models are precise while group selection is ambiguous (Mallet, 2010; West et al, 2011), and 3) Kin selection is supported by empirical studies while there is no evidence to support group selection (Bourke, 2011 as well as the recent online critiques by Dawkins, Pinker, and Coyne).
The first critique is conceded on both sides and could apply equally to both. So, then, what is the group selection model and is there any evidence to support it?
All For One, One For All
Imagine a simple system, like a petri dish of E. coli for example. Now suppose there was a simple genetic trait that offered benefits to other cells but brought a cost to the individual producing it, perhaps a gene that promoted antibiotic resistance for cells nearby. Theoretically, if a cell had this gene it would benefit its daughter cells and the trait would spread. But then, of course, there’s the problem of free riders. Even if this altruistic trait was extremely successful in the population, wouldn’t the emergence of even a single selfish cheater destroy everything? The cheater would benefit from the antibacterial resistance produced but wouldn’t give anything back, meaning they would end up leaving more daughter cells than the altruists. Continue this pattern a few more generations and wouldn’t the altruists be completely eliminated? Not necessarily.
A prediction of the Price equation is that altruistic groups will ultimately be more successful than selfish groups under certain conditions, even though selfish individuals could outcompete altruistic individuals. Imagine the same petri dish as before but this time cells randomly congregate into a large number of mixed groups containing both altruists and cheaters. Two things will now be expected to happen. Because altruists are at a disadvantage they will reproduce more slowly than the cheaters as a result of within-group competition. However, the benefit provided by the altruists means that those groups who accidentally have more altruists will grow faster and end up having more total cells than groups composed mostly of cheaters. Because the altruistic group as a whole ends up being more successful there will be an overall increase in the number of altruistic individuals in the total population. Bring all the cells together to form random groups again and eventually you could have groups, as well as an entire population, composed only of altruists.
At this point it should be pointed out that this model isn’t simply theoretical but represents the empirical results of a study published on January 9, 2009 by John Chuang, Olivier Rivoire, and Stanislas Leibler in the journal Science (and was successfully replicated earlier this year in Nature). Based on this example it should be clear that there could be no preference towards relatives (i.e. daughter cells) to explain the results because the altruism involved was broadcast indiscriminately. The evolution of the altruistic trait in this case was based on association, not kinship.
Coming Home to Roost
When William Hamilton received the covariance equations from George Price in the late 1960s he recognized the inherent limitation in his model of inclusive fitness (“everyone makes a mistake now and then,” Price told him). However, this was more than compensated for by the extension that Price’s equation now offered. As Hamilton wrote in his memoir (p. 173):
Some months before he died I was on the phone telling him enthusiastically that through a ‘group-level’ extension of his formula I now had a far better understanding of group selection and was possessed of a far better tool for all forms of selection acting at one level or at many than I had ever had before.
“I thought you would see that,” the squeaky laconic voice said, almost purring with approval for once.
Just after Christmas in 1974 George Price died after he jabbed nail scissors into his neck and snipped his carotid artery. By this point he had given up evolutionary biology and was living in an abandoned building as a squatter with the homeless alcoholics he had been trying to rehabilitate. A few months after attending his funeral Hamilton published his little known paper entitled “Innate Social Aptitudes of Man: An Approach from Evolutionary Genetics” in the edited collection Biosocial Anthropology where he used the group selection component of Price’s covariance equation. In this paper he proposed that inclusive fitness should now be understood as a multilevel selection approach in which there is a nested hierarchy extending from genes, to individuals, to kin, and finally to groups.
Because of the way it was first explained, the approach using inclusive fitness has often been identified with ‘kin selection’ and presented strictly as an alternative to ‘group selection’ as a way of establishing altruistic social behaviour by natural selection. But the foregoing discussion shows that kinship should be considered just one way of getting positive regression of genotype in the recipient, and that it is this positive regression that is vitally necessary for altruism. . . But it seems on the whole preferable to retain a more flexible use of terms; to use group selection where groups are clearly in evidence [whereas] ‘kin selection’ appeals most where pedigrees tend to be unbounded and interwoven, as is so often the case with humans.
In Hamilton’s suggestion, which is virtually identical to multilevel selection as David Sloan Wilson presents it today, group selection should be reserved only for those cases where the altruistic behavior cannot be explained by kin selection. In contrast, E.O. Wilson wants to do away with kin selection because the group selection equations reach the same result anyway. Ultimately, Wilson’s argument is about reducing the number of steps between the levels of gene and group. While it is possible that Wilson has a point for the eusocial Hymenoptera (though this remains to be seen), the utility that kin selection has for other species–particularly primates–remains strong, meaning that Wilson probably overreached. As for multilevel selection more generally, there seems to be no reason why Hamilton’s suggestion for a “gene’s eye” view of group selection shouldn’t be considered when the necessary conditions are met.
There have been multiple empirical studies that scientists argue support a group selection approach. Charles Goodnight and Lori Stevens reviewed many of these in The American Naturalist in 1997 and David Sloan Wilson has discussed others online and in his book Unto Others: The Evolution and Psychology of Unselfish Behavior co-authored with Elliot Sober. This evidence (as well as that from additional studies) should be scrutinized to determine whether it supports such claims or not, but it nevertheless represents the empirical results necessary for building a scientific theory.
In the final analysis, multilevel selection is little more than a rebranding of Hamilton’s inclusive fitness (albeit the “enhanced” 1975 version). So is that what this fight is really all about, the objection over a name change? On a superficial level, yes, but there are larger stakes involved. Consider the case of Pluto. When the International Astronomical Union demoted Pluto from its longheld status as a planet in 2006 it was met with outrage. Even something as simple as a name change must face the reality of textbooks that have to be rewritten, professional reputations that are invested in the status quo, funding opportunities that could be lost, as well as a theoretical shift that some scientists may be unwilling to make. As the physicist Max Planck once quipped, and Richard Dawkins has repeated, “Science advances one funeral at a time.”
But this is simply how the scientific process works. Boundaries are tested and concepts are challenged. If a theoretical framework remains internally consistent it will gain more adherents while disputes within other frameworks may cause them to dwindle and ultimately disappear. There is a rich fossil record of unsuccessful ideas–from phlogiston to Lysenkoism–that litter the field of our scientific past. If nothing else, the modern conflict over altruism will help us ensure that the best ideas are passed on to the next generation. It’s a good fight to have.