Reciprocity, Reputation and Nepotism
SUPERCOOPERATORS: Altruism, Evolution, and Why We Need Each Other To Succeed. Martin A. Nowak with Roger Highfield. xxii + 330 pp. Free Press, 2011. $27.
SuperCooperators is Martin A. Nowak’s latest book reviewing his theoretical work on the evolution of cooperation. Written with Roger Highfield, the editor of New Scientist magazine, it is an ambitious attempt to present Nowak’s highly mathematical research to a general audience without using equations. In fact, the authors bend over backward to allay the fears of the math-phobic, to the point of always writing out a key variable’s name, sigma, rather than using the scary Greek letter σ. The result is a skillful courtroom brief that masterfully advocates a particular viewpoint on cooperation. However, like good lawyers, the authors don’t mention the weaknesses in their arguments, leaving those for others to discern. Thus the picture they paint is far from comprehensive.
Nowak and Highfield argue that cooperation drives the evolution of many features of biological complexity, from the earliest replicating strands of RNA to everything that makes humans a unique species. Cooperation is defined in the broadest possible way, so that a reaction that is catalyzed by one RNA strand and promotes the assembly of another strand is considered to be a cooperative endeavor. Probably most scientists do not view chemical reactions as examples of cooperation, but I suppose a good central theme can be forgiven some exuberant exaggeration.
After an introductory chapter, the book puts forth the proposition that cooperation can only evolve and overcome selfishness through one of five mechanisms: direct reciprocity (I’ll scratch your back if you scratch mine); indirect reciprocity (I’ll scratch your back, you scratch someone else’s, and eventually someone will scratch mine); spatial structure (back-scratching tends to bring back-scratchers together); group selection (groups that back-scratch do better than groups that don’t); and kin selection (if you’re a relative, I’ll scratch your back even if you won’t scratch mine). The authors argue quite persuasively that humans are the top “SuperCooperators,” because all five mechanisms are evident in our behaviors and societies.
None of the individual mechanisms for the evolution of cooperation is particularly controversial as a theoretical concept (although it is not clear to me why spatial structure cannot be subsumed into the other four). What is controversial is the relative importance the authors attach to each of the five mechanisms for influencing the evolution of cooperation.
Nowak argues repeatedly that a particular format for direct reciprocity, the Prisoner’s Dilemma, is the key to understanding the evolution of cooperation. A Prisoner’s Dilemma is defined as a situation with a specific set of payoffs for cooperation and defection. Pairs of cooperating individuals do better (each getting a higher reward, R) than pairs of noncooperators (who each get a lower reward, P), but this does not automatically lead to cooperative behavior. The dilemma arises from the fact that the strategy that has the highest payoff of all (reward T) is for a player to defect (not cooperate) in the presence of a cooperator (who then gets the lowest reward of all, the sucker’s payoff, S). A large portion of SuperCooperators focuses on how cheating to gain short-term advantages can be overcome to get to a stable state of cooperative, altruistic behavior yielding long-term benefits. Nowak tends to rank kin selection as a less important process than such reciprocity. Given that many evolutionary biologists believe the opposite—that kin selection is the far more important mechanism—this book will indeed ruffle some feathers.
I expect that the most ruffled will be those studying cooperation in nonhuman species. Nowak’s background is chiefly mathematical, and his ventures into nature are strikingly anthropocentric. This book attempts to chart the entirety of the evolution of life from algal mats to humans. Yet the authors cite barely a dozen original studies on the actual behavior of nonhuman animals in natural settings, while citing several hundred that model cooperation mathematically! The implication is that if the math works, then that is how evolution must also work. Why bother testing the conclusions against reality?
Nowak’s preference for math over biology is most evident in the fact that he considers Prisoner’s Dilemma models to be more useful and to have greater merit than kin-selection models. The dynamics of the Prisoner’s Dilemma game are a very interesting puzzle for mathematicians. It almost seems that every evolutionary game theorist, to prove his or her worth, is required to solve a variant of this game. This has resulted in innumerable mathematical solutions incorporating many scenarios and assumptions.
Solutions to the Prisoner’s Dilemma often invoke variations of the tit-for-tat strategy (“what you just did to me, I’ll do back to you”). However, as I routinely advise new behavioral ecology graduate students, “If you can find any natural system in which the payoff matrix of the Prisoner’s Dilemma is unequivocally present, you will be famous forever, no matter what the animals do.” There simply is no conclusive evidence that a Prisoner’s Dilemma applies anywhere in nature apart from human interactions. Indeed, in support of this supposedly ubiquitous model of cooperation, Nowak cites only a handful of mostly anecdotal observations of situations in which animals might be doing tit-for-tat. Only one study (by Manfred Milinski) is experimental: A stickleback inspects a potential predator, and angled mirrors give the fish a reflection that appears to be a companion who is either cooperating (keeping pace) or defecting (retreating). In the presence of a “cooperative” reflection, the fish usually comes closer to the predator than when paired with a “defector” reflection. In his summation, however, Nowak fails to mention the numerous criticisms of the study by those who have questioned whether the Prisoner’s Dilemma payoff matrix actually holds for the fish, or who have proposed that a “safety in numbers” explanation better explains the results. Nor is there any mention in the book of Kevin C. Clements and David W. Stephens’s ingenious 1995 cooperation experiments with blue jays. When placed in an iterated Prisoner’s Dilemma, these birds never exhibited stable cooperation. Overall, game theory might better predict nature by focusing less on the Prisoner’s Dilemma and more on games with payoff matrices in which cooperation benefits players both immediately and in the future, such as Snowdrift or Mutualism (which are briefly mentioned in the book).
Kin selection is often represented by evolutionary biologist W. D. Hamilton’s simple rule: Cooperate if the reproductive benefit (B) provided to an individual, prorated by its genetic relatedness (r), exceeds its reproductive cost (C) to the actor (rB–C>0). Because of all the hidden assumptions and complexity rolled into the seemingly simple variables B and C, the rule is a mathematical purist’s nightmare. Nowak may well be correct that the full calculation of inclusive fitness required to make the rule mathematically rigorous comes at the cost of losing its elegant simplicity. He has presented an alternative approach in “The Evolution of Eusociality,” an article he coauthored with Corina E. Tarnita and Edward O. Wilson, which was published in the August 26, 2010, issue of Nature. His suggested approach is a population-genetics model, which essentially predicts that cooperation can evolve when members of groups are reproductively more successful than solitary individuals. This must be true, of course, because cooperation could not evolve if it were not reproductively beneficial. But such a prediction is also profoundly trivial, because—unlike Hamilton’s rule—it offers scant insight into what exactly makes cooperation the more reproductively advantageous choice.
Although Hamilton’s rule occasionally fails to accurately predict behavior, it has yielded numerous major insights into the evolutionary dynamic between cooperation and conflict. Sometimes imperfect math leads to great biology. Nowak and Highfield acknowledge this, observing vaguely that “Hamilton’s rule has been a valuable heuristic” and has “inspired many field studies.” But they don’t specifically describe any of the rule’s successes in what must now be at least a thousand experiments and studies. Nor do they give a single example demonstrating that their alternative model produces a different, and more valid, prediction than do inclusive fitness calculations. In my view, they advocate replacing a model that has repeatedly illuminated how evolution works with one that often does no more than confirm the obvious.
Furthermore, kin selection lurks in the background in several of the book’s chapters, apparently unnoticed by the authors. In “Society of Cells,” cancer cells are proposed to be mutant cheaters in a cooperative game. Instead of restraining their reproduction for the common good of the body, cancerous cells defect and favor selfish growth. But a mutation for unrestrained growth also makes the cancer cell unrelated at that locus to the rest of the body. And selfish growth is exactly what Hamilton’s rule would predict for an unrelated parasite exploiting its host.
In the penultimate chapter, “Game, Set, and Match,” Nowak describes his “cosmic conclusion”: Cooperation evolves in the iterated Prisoner’s Dilemma game based on three variables: the average payoff for a cooperator (R+S); the average payoff for a defector (T+P); and the variable σ, which is the relative rate at which cooperators and defectors encounter one another. If σ is greater than one, then like individuals interact with one another more often than by chance—just as, in many species, kin are more likely to interact with one another than with random strangers. Therefore, cooperation is favored when (σR+S)>(T+σP). Rearranging gets σ(R–P)–(T–S)>0, or in other words, the equivalent of rB–C>0. Hamilton, like a phoenix, rises again.
Of the remaining chapters, some work better than others. “The Gift of Gab” is a fascinating account of how mathematical descriptions of language can reliably predict its cultural evolution. The authors put forth the interesting proposition that the evolution of human language and of the large brain required to handle its complexity were both driven by the advantages gained from being able to cooperate through indirect reciprocity. This chapter is marred only by a one-sided consideration of Noam Chomsky’s “universal grammar” hypothesis. Critics have very vigorously challenged this hypothesis, but Nowak and Highfield present only the supportive mathematical results and not the full debate.
In “Punish and Perish,” the authors argue that punishment rarely makes cooperation work more efficiently. They describe experiments with human subjects showing that groups meting out punishments do not succeed as well as ones that are supportive. However, humans appear to have a great need to punish social transgressions. Consider, for example, that subjects in psychology experiments have been easily induced to inflict harsh punishments with enthusiasm, and prisons in the United States are overflowing. Left unaddressed is the question of why models so poorly predict this distinctly human behavior.
Finally, I found the chapter on networks of human interactions that can transmit both information and disease (“How Many Friends Are Too Many?”) to be disappointingly superficial—particularly in comparison with Malcolm Gladwell’s incisive 2000 book, The Tipping Point, which is full of examples and practical suggestions. It is shocking that Gladwell’s book rates no mention here.
In reading SuperCooperators, I was often annoyed by the authors’ self-promoting tone, obliviousness to alternative views, ignorance of actual data and inability to distinguish a truism from an informative insight. Despite this, I still quite enjoyed the book, and I found it hard to put down once I started reading. Time and again I needed to really think through why I disagreed with the authors. And sometimes—with regard to the evolutionary uniqueness of humans, for instance—I came around to their point of view. A book that continually challenges you to think is more valuable than one that leaves you nodding absentmindedly in agreement. I am still an unrepentant kin selectionist, but now I have a much better idea of why. Whether you agree with Nowak or not, the book’s subtitle makes a valid point: “We need each other to succeed.” Anyone who wants to sharpen his or her thinking about why we cooperate ought to read this book.
Peter Nonacs is a professor in the Department of Ecology and Evolutionary Biology at the University of California, Los Angeles. He is an evolutionary behavioral ecologist who has done extensive theoretical and experimental work on the evolution of cooperation in social insects.