The Price of Altruism

by Oren Harman

  When it came to the evolution of man, it seemed obvious to George that group selection must have played a role. Early humans had lived in groups, and cultural inheritance could have gone a long way in preserving the kinds of genetic behavioral differences that would otherwise be swamped by migration.

  Huxley, Kropotkin, Allee, Wynne-Edwards, Emerson, Fisher, Wright, JBS, Maynard Smith—he read them all and more. Every one was occupied with the question of the unit of selection and every one seemed to have an answer. Still, was it conceivable that each held a portion of the truth, that none was entirely right but none entirely wrong, either? Could the same kinds of economic mechanisms Hardin argued were necessary to square the individual and common good exist, biologically, in nature? Could it be, in other words, as Darwin had noticed when contemplating the ants, that selection worked on different levels simultaneously?

  At the end of April he said his good-byes to Alice, arranged for Miss McCartney’s eviction, closed up the old apartment, and got on a plane. Back at UCL, Cedric Smith was pushing him to complete a grant proposal to the Science Research Council. Classical theory, CABS wrote in his report, assumed that each individual possesses a “fitness” independent of the “fitnesses” of others, and that by analyzing the situation mathematically, the course of the evolution of a population can be predicted. This was a good approximation of reality in some situations, as when an inherited disease shortens life, but where the interaction between individuals plays an important role in determining their fate it simply wasn’t good enough. Sexual selection, parental care, formation of families and communities: All were situations in which conflict and cooperation were paramount. Interaction, not singularity, was the name of the game, and few besides Fisher and Hamilton had ever played it. “Dr. Price has come to the subject comparatively recently,” Smith wrote almost apologetically, adding, “I have however been greatly impressed by his ability.”49

  For George, meanwhile, little had changed. “I continue to have the plan of limiting my life span to about 50 years,” he wrote to Tatiana.50 Soon after, the news came that the SRC had awarded him a three-year grant effective July 1, 1969. Walking home from UCL to the flat on Little Titchfield, he eyed the spire at All Souls. The only passion that remained in him was to crack the mystery of the evolution of family—that and exacting revenge on Ferguson.

  A few days later Alice died peacefully in New York.

  Back in February, before leaving for New York, George heard a talk delivered at the Royal Society of Medicine by a psychiatrist from Maudsley Hospital named John Price. “The other Price” as he became known, had been interested, too, in ritualized animal combat: If there was no physical pain or incapacity, what made animals yield to the winner? Price’s idea was that ritual yielding is subserved by mental incapacity and mental pain, and that human depression and anxiety—both painful and incapacitating—might have evolved from it. The notion was attractive to a clinical psychiatrist: It suggested a wealth of animal models for the study of human neurosis as well as a battery of new ideas for prophylaxis and treatment.51

  Most important to George, though, was the gamelike logic behind the claim. Why, in fact, should there be any variation in yielding behavior? In terms of the group the answer was obvious: The greater the variation of yielding tendency in the population, the greater the chances that any two contestants are unevenly matched, and the shorter the duration of battles. If selection between groups had been important in evolution, groups with shorter spats would have surely done better than those with costly, protracted encounters.

  But variation in yielding behavior could also favor the individual, as “the other Price” explained:

  The disadvantage of being a yielder is counterbalanced by the likely mortality when two non-yielders meet each other. Thus it is advantageous to be a yielder when everyone else is a non-yielder, and to be a non-yielder when everyone else is a yielder. This dependence of the advantage of one’s phenotype on the phenotypes of the rest of the population…tends towards the maintenance of variation in the population.52

  It was precisely the same notion George had come up with in his paper on antlers, and it hinged on the logic of games. Hardin had shown that there was often conflict between individual interest and the common good, and von Neumann that you couldn’t maximize simultaneously for two variables. And yet even here, where the individual interest and common good seemed to correspond, George saw no reason why natural selection couldn’t be working on both at the same time. An “outsider” untrained in evolutionary theory, he really had no reason to choose sides: Why limit the scope of Darwin’s theory? The only relevant question, it seemed to him, was not whether selection was working on the individual or the group, but how, in each and every case, to tell which force was stronger.

  Once more he turned to his equation. Covariance was a most simple relationship, a mathematical tautology. To make it even better, George now saw, he’d need to include a further component: something called “transmission bias.” If a trait moved from a “mother” population to a derived “daughter” population or, in evolutionary terms, from generation to generation, it was important to know not only if it helped to increase fitness but also what the chances were for it being inherited. If the trait was genetic, for instance, it would be important to know whether it had undergone any kind of mutation or, equally important, whether the gene somehow biased the system so as to pass itself on more frequently than would be expected. And so, to get the equation just right, George added a second term

  where E(wz) is a measure of the extent to which the trait, z, will be passed on faithfully. Usually parents pass on their genes to children at random, meaning that genes don’t affect which particular sperm will fertilize the egg. Since it almost always equaled zero, the transmission term could therefore usually be safely ignored. But now his equation was formally complete: Given a trait, z, its fitness, and the likelihood of its transmission, it could tell you precisely how it would evolve from one generation to the next.

  The equation partitioned trait change in evolution into selection (Cov) and transmission (E). This was valuable enough,53 except that the new term did much more. With a few simple substitutions it showed that selection could work at two levels simultaneously; in fact it could even partition them to see how much each contributed to the overall change. Instead of defining the two terms of the equation as the selection and transmission terms, corresponding to the individual and the genes in the sperm and egg respectively, they could be bumped up one notch in the rung and redefined as relating to the individual and the group. The key was for the far right-hand side part of the equation to be treated just like the left-hand side of the equation, only at a lower level.54

  This was enormously useful. Imagine a group of people who only know how to be altruistic to one another, and have never heard of selfishness. Imagine a second group whose members never heard of altruism but rather only know that each has to be out for himself. Clearly the first group would function better as a society, for it would revel in cooperation, whereas in the second everyone would be poking one another’s eyes out. But what would happen if the groups were not entirely pure? An altruist would be easy fodder in the selfish group, whereas an egoist would quickly hoodwink all the members of the altruistic group. And so while the altruistic group is fitter than the selfish group, selfish individuals are fitter than altruists within each group. The question was, If two such groups live side by side, which trait will evolution select—altruism or egoism? Which is stronger—the interest of the group or the interest of the individual? The covariance equation could tell you the answer.

  Without entirely meaning to, George had written an equation that had the power to do what generations since Darwin had failed to: watch natural selection work in all its glory at different levels at the very same time. In fact, since it was infinitely expansible, all the levels of life could be included: gene, cell, individual, family, group, species, even lineage. Like a giant eye perusing creation, sel
ection could see everything; all one needed to do was to choose which two levels to compare. Finally, after all the evolutionists who came before George, not to mention Hardin’s “Tragedy of the Commons,” the equation could specify the exact conditions under which the good of the group would upstage the good of the individual. Crucially, contra Hamilton’s kin-selection model, it needn’t write off goodness as merely apparent: When selection worked more strongly between groups than within them, a genuine altruism could evolve.

  Staring at his own creation, unbelieving, George thought of it as a “miracle.”55

  Hamilton had arrived back from Brazil with Christine, Romilda, and Godofredo in January 1969, and was invited in May to give a keynote address at the “Man and Beast” meeting at the Smithsonian Institution in Washington, DC. The conference had been organized with great hype and expense to impress on politicians and the public that biology and evolution were relevant to man, but the bushy-browed loner brought with him more than they had bargained for. Xenophobia, Hamilton stated—even relishing cruelty to others—was selected for in the evolution of man, since altruistic groups must expand at the expense of other groups and to do this they need to fight them. Reliance on the instincts of a supposed “noble savage” was no answer to the prisoner’s dilemma or to Malthus: What produces altruism and kindness at one level only serves to produce hatred and violence at the other. Alas, group selection doesn’t exorcise the harsher aspects of natural selection; actually it leads to fascism.56

  The audience was shocked. At the aftertalk cocktail party a senator’s wife pinned Hamilton with her “chin and fierce eyes.” Could his theory help reduce crime in America’s inner cities? she asked, desperate to eke out a measure of solace. Hamilton himself was deeply troubled by the thought: If not in human nature, where else could trust and hope in a peaceful and creative future be placed? And yet only honesty about the manner, he determined, stood the faintest chance of taming “the beast within.”57

  It was a terrible irony to come to terms with: With biology at the helm, goodness could be bought only at the price of cruelty. Worst of all, this tragic truth would be most pronounced in man; in all other creatures group selection was such an ill-defined abstraction, Hamilton thought, that it could be safely omitted from the evolutionist’s tool-kit.58 The reason was that there was simply no plausible mechanism, like cultural inheritance, that would allow for it. Only in humans could the force of culture work to counteract selection at the genetic and individual levels; in the process it would allow selection to work at the level of the group. When it came to all other creatures, Hamilton remained a bitter enemy of the old notion of the greater good.

  Altruism entailing malevolence had been precisely what George had written to him about a year ago. Hamilton had forgotten about it. But having worked out more of his equation, George was now ready to write to him again. “I must be one of the world’s worst correspondents,” his letter began, a year of silence having lapsed since Hamilton wrote his sweaty reply from Mato Grosso. Then he went on to retell the entire story of trying to rederive Hamilton’s kin-selection math, coming up with the covariance equation, going into Cedric Smith’s office, getting the keys and honorarium at UCL. Clearly, alongside altruism, spite was a possibility: What kin selection will do is increase the frequency of a gene that causes animals to give greater benefit to near relatives, but this always meant that lesser relatives had to be relatively harmed. Hamilton thought that kin selection could cause individuals to act altruistically to the group as a whole, but this was an ideal miscalculation. It was an oversight he would obviously need to fix. “But I did want,” George wrote, “in view of your friendly correspondence, because I respected your work, and because everyone makes a mistake now and then—to publish in a way that would not embarrass you.” Perhaps the best way, then, would be simply for Hamilton to publish his own correction. In any event George would be glad to speak with him on the phone.59

  When Hamilton called the next day, he found the voice on the other end strange, rather guarded, “squeaky and condescending.” His covariance equation, George said, was “surprising to me too—quite a miracle,” really. But most surprising to Hamilton was what George had to say next, following a rather awkward silence: “Have you seen how my formula works for group selection?” “I told him, of course, no,” Hamilton later remembered, “and may have added something like: ‘So you actually believe in that do you?’”60

  After the phone conversation, each man returned to his own world. Hamilton was adjusting to family life and the two new adopted kids, as well as figuring out the outlines of a new theory of the “geometry of the selfish herd.” George, for his part, had one more proof to complete the equation, and was using the FORMAC computers at UCL to help with the algebraic expressions. “I think it will be considered a very important piece of work,” he wrote to his elderly aunt Ethel in Hiawatha, Michigan. “I wish mother might have lived to see it.”61

  October was the driest and warmest England had seen in centuries. Over the summer George had been working on a long paper, an ambitious attempt at a general theory of selection. Now, though, he had a change in publication plans. On November 11 he sent off a short letter to the editor of Nature, barely longer than a page and, reflecting his independent path, with not a single reference note. In it he derived his equation and explained how covariance could be applied to problems in genetic evolution, without dwelling on any in particular. It was a “preliminary communication,” he wrote to Annamarie, just a tool that would ultimately allow him to crack the mystery of family.62 A few months later he sent the longer paper to Science.

  The rejection arrived in February; “It is too hard to understand,” Nature’s editor, John Maddox, wrote. The letter did little to help George’s mood: He was working as a teaching assistant in a statistics course at the Galton to earn a little extra cash now, and still thinking about possible magazine articles to sell in the United States. Obviously the readers were just “stupid” after all, this was the miraculously simple equation that had earned him his room at UCL. The year 1970 had rolled in, and still no breakthrough.63

  The next month, as he walked home to Little Titchfield beneath the ever-present All Souls spire, George discovered in the mailbox that his troubles hadn’t ended there. Science had rejected the longer selection paper, too. It was too abstract, the note said, and besides that brimming with hubris, not obviously applicable to any particular scientific problem. George thought it “vicious…really breathing hatred toward me,” but the rejection, perhaps, wasn’t all that surprising. After all, in capital letters so that no one could miss it, he had stated that he was in search of a general “Mathematical Theory of Selection” analogous to what Claude Shannon had done with the theory of communication. It was quite a claim—especially coming from someone no one had heard of.64

  Back in America, Al Somit had become the chairman of the Political Science Department and vice president elect of the State University of New York in Buffalo. He was planning a session at the International Political Science Association meeting that summer in Munich on “Biology and Politics” and wrote to ask his old buddy if he’d like to attend. “Our interests seem to be converging,” he added amid their usual Chicago-day banter. George, for his part, complained that even in London he couldn’t escape the “non-Aryans”: his landlords, Basil and Howard Samuel; the kosher bakery across the street; the kosher restaurant at the corner; Sandy, the Yeshiva University fling who had visited in May; his instrument-making friend Ludwig Luft, who was lending him money; Harry Harris, the chairman of the department at Galton—all were Jews. Al shrugged it off as usual George.65

  But not all was as usual. George was feeling the world slowly closing in on him. He contemplated Munich. He contemplated life. Down and out, dejected, he offered to speak about morality being nothing but a “masquerade.”

  His search for a miracle, it was patently clear, had ended in glorious failure. His mother was dead. Julia hated him. Edison and he had lo
ng drifted apart, but so now were his quickly maturing daughters, living out in California and starting families of their own. He was alone in a foreign country and had no friends. His arm and shoulder were partly paralyzed. His forays into human evolution were incomplete, dependent on a selection mathematics that had been humiliatingly rejected, twice. Swinging London was as much at odds with his grim lifestyle as the search for the origins of family was with his tattered personal affairs. His equation might say otherwise, but if there was any goodness in this world, George Price hadn’t found it.

  Excerpts from George Price’s seminal paper in Nature

  “Coincidence” Conversion

  Munich never materialized, but in May a turn arrived as unexpected as it was welcome. “I am enchanted with your formula,” the usually lugubrious Hamilton had written in a hurried script, barely able to contain his excitement. “I really have a clearer picture of the selection process as a result. In its general form I can see how we might use your formula to investigate “‘group selection.’”1

  It was quite an about face for a man who had judged group selection as so “wooly” an abstraction that it could be safely omitted from the tool kit of the evolutionist. Altruistic behavior was disadvantageous within groups; to evolve it would require groups fighting among each other. Kin selection, therefore, could not be an alternative to group selection: Both existed in nature’s plan and needed to be related to each other. Accordingly Hamilton had taken George up on spite, and, in the process, devised a foolproof stratagem. It was a piece of trickery that made him especially proud: He would send his article on spite to Nature, wait for its acceptance, and then write to its editor, Maddox, to say that he would have to withdraw since it was entirely based on the covariance math of George Price, whose paper—in case Maddox forgot—had been rejected by him in February. Hamilton would, of course, be happy for “Selfishness and Spiteful Behaviour in an Evolutionary Model” to appear in Nature but not until the status of Dr. Price’s article was finally resolved. A masterstroke of subterfuge, a coup of cunning and guile, amazingly, it worked. George’s letter to the editor, Nature soon replied, would be published as an article, “Selection and Covariance.”2