Steven Pinker Replies

According to the defenders of the theory of group selection among the commentators, group-against-group competition is the only way to explain why humans have morality, empathy, culture, language, social norms, and locally useful adaptations such as digesting milk and breathing thin air. At the same time, the theory is mathematically equivalent to standard evolutionary theories based on gene selection or inclusive fitness, so the two theories make identical predictions and can never be empirically distinguished.

This muddle confirms the point of my essay: that the concept of group selection is making a shambles of the application of evolutionary biology to human psychology. It has led commentators to equate human goodness with prowess in tribal competition—as if the only reason we feel compassion for a lame puppy is that it helped our ancestors massacre the villagers across the river. It has prevented them from asking elementary questions, such as whether digesting milk or breathing efficiently could conceivably provide any benefit to the digester or breather as opposed to the group. It has led them to maintain both that group selection is empirically indistinguishable from gene selection and that it is the only theory that can explain of a raft of empirical phenomena, from punishing free riders in experimental games to obeying the categorical imperative when deciding whether to vote.

Group selection has become a scientific dust bunny, a hairy blob in which anything having to do with "groups" clings to anything having to do with "selection." The problem with scientific dust bunnies is not just that they sow confusion; as Tooby notes, the apparent plausibility of one restricted version of "group selection" often bleeds outwards to a motley collection of other, long-discredited versions. The problem is that it also obfuscates evolutionary theory by blurring genes, individuals, and groups as equivalent levels in a hierarchy of selectional units; Tooby and Dawkins remind us that this is not how natural selection, analyzed as a mechanistic process, really works. Most importantly, it has placed blinkers on psychological understanding by seducing many people into simply equating morality and culture with group selection, oblivious to alternatives that are theoretically deeper and empirically more realistic.

To be fair, several commentators (Henrich, Wilson, Gintis, Boyd & Mathews, Queller, and Richerson) extricate some hard kernels from the dust bunny, namely mathematical models of social evolution that have been called "group selection." They are right to complain that my essay mentioned these models only in passing. Later I'll explain why I think the models are unhelpful in understanding human psychology, but first I need to spend a bit more time picking apart the dust bunny. This will require an umpteenth exegesis of the term, which itself is a symptom of the mess it has put us in.^[1] To those like Everett who complain that "we need to get out of the dictionary and into the field," I offer a simple solution: stop using the term group selection as a loose synonym for the evolution of group living, group competition, group norms, group practices, social networks, culture, selflessness, kindness, empathy, altruism, morality, clannishness, tribalism, or coalitional aggression.

Is Group Selection Necessary to Explain Morality, Sociality, and Culture?

To read the commentaries, one would think that three facts about Homo sapiens constitute knock-down proof of the theory of group selection. The first is that people are not ruthless psychopaths but extend benefits to each other, such as holding open doors, helping the sick and injured, and voting in elections (Hochberg, Richerson, Gintis). The second is that humans belong to cultures which have different bodies of knowledge and skill and follow different norms and practices (Henrich, Richerson, Boyd & Mathews, Everett). The third is that people feel loyalty to their coalitions, such as tribes, teams, and nations (Hochberg, Haidt). Much of the rhetorical energy of their commentaries is directed at a misreading of my essay as a denial of these truisms: "Contrary to Pinker, humans are not solitary psychopaths; therefore group selection is true."

My point of course was not to deny that humans are social or moral or cultural but to show that group selection is unnecessary to explain these obvious facts. That is because living in groups and sharing its accumulated know-how can benefit individuals, so one can explain all the zoologically unusual traits of Homo sapiens using conventional gene-selection theories. My favorite of these theories, "the cognitive niche, " combines the old idea of man-the-toolmaker with the recent idea of reciprocal altruism.^_[2] Homo sapiens evolved neural adaptations to explore its physical environment, develop cognitive models of its causal structure, play out these models to invent tools and techniques which defeat the defenses of plants and animals, trade this knowledge with other humans, and cooperate with them to accomplish what no single human could do in solitude. Language, cognitive know-how, and social intelligence and emotions coevolved in the human lineage, each multiplying the benefit of the others, because knowledge is an ideal trade good, allowing a large benefit to be conferred to another at a small cost to oneself. As groups of cooperating, knowledge-sharing humans split up and the splinters lost touch with one another, differences in their skills, knowledge, and conventions of coordinated behavior (the cooperative equilibria that Boyd & Mathews rightly emphasize) accumulated separately within each group, giving rise to distinct cultures. None of this requires that individuals systematically incur a cost that exceeds the benefit they draw from group living, and none if it requires anything called "group selection."

Within gene-selectionist theories of human evolution , the mechanism of reciprocal altruism, augmented by reputation and commitment, provides a psychologically realistic and well-supported explanation of human social and moral life (see the summaries by Baumard, Krasnow & Delton, and Price). Social life provides ample opportunities for gains in trade (two agents are each better off if they exchange their surpluses and extend branches to rescue each other from drowning than if they let their surpluses rot and filed their nails while the other drowned), as long as each protects himself against one-sided exploitation by the other. That protection can be ensured by remembering individuals and their past behavior, or asking around the grapevine, and selectively cooperating with those who are likely to reciprocate. Quintessentially human mental faculties such as person recognition, person memory, sympathy for the needy, gratitude to benefactors, anger at cheaters, contrition at having cheated, forgiveness of transient and accidental cheaters, trust for the trustworthy, thirst for gossip about the virtue and perfidy of others, and a concern for one's own reputation, including guarantors of that reputation such as integrity and credibility, fall naturally out of this theory, which can also be extended from dyads to groups.^[3]

Gintis's recounting of this theory is seriously distorted. He conflates it with psychological egoism, the irrefutable doctrine that people, even at their most generous, invariably act in their interests, if only for the esteem and self-satisfaction. It's a common misunderstanding of the gene selectionist explanation of altruism which grows out of a confusion of the metaphorical selfishness of genes with the real selfishness of people (Nesse, Coyne, and Tooby also call attention to the confusion). Notwithstanding Gintis's out-of-context quotation by Dawkins, the Selfish Gene was an explanation of the evolution of altruism, including genuine, in-the-bone beneficence. As Robert Trivers explained in a seldom-remembered passage in his classic 1971 article introducing reciprocal altruism (happily rediscovered by Nesse and Baumard), as long as people can choose among many potential reciprocation partners they can shun not only the blatant fly-by-night exploiters but also the penny-pinchers and fair-weather friends, and can shop around for the most generous and reliable ones they can find.[^4] Since being chosen as a reciprocation partner is better than being shunned, this market sets up a corresponding pressure to stand up to such scrutiny, which can include actually being generous and trustworthy.^[5] Nesse provides the apt analogy of a reputable business that loses money in the short term on costly merchandise returns but gains it over the long term in customer loyalty.) At that point it becomes a philosophical question, in the worst sense, whether the agent is "truly" moral or has just evolved to be that way because of its "selfish" advantages. In the limit, where selfishness and deceit can never be concealed (and not coincidentally, under the religious construction of an omniscient deity), agents should be perfectly moral. Of course perfect assessment of one another's generosity and trustworthy is impossible, so agents may be tempted to compromise their principles for short-term profit—but then, real humans are not perfectly moral.

Also worth examining is Henrich's surprising claim that the "niceness" of successful reciprocation strategies (cooperating on the first move, or conferring a large benefit on a stranger if it can be done at a small cost to oneself) is an artifact of standard formulation of reciprocity models. What makes the claim surprising is that simulations of the evolution of altruism have shown that niceness is favored over a wide range of model specifications and parameter values, because the potential benefits of a lifetime of reciprocation by the beneficiary (or his friends and relatives) vastly outweigh the cost of a single small favor wasted on a nonreciprocator. (Krasnow & Delton's simulations demonstrate this clearly, notwithstanding Gintis's incomprehensible attack on them.) I fact-checked Henrich's assertion, and gets things backwards.^[6] His model with Hruschka built in the failure of niceness by brain-damaging their agents so that they could remember no more than a handful of other agents (three of them, in most of the models, and never more than eight). Not surprisingly, when these agents' paltry memory slots filled up, there was no advantage to being nice to a new partner, because they had no way of remembering him in the future. But of course humans' memory for people they meet is not crippled in this arbitrary way.

Do Formal Models of Group Selection Explain Human Altruism?

Let's turn from the morality/sociality/culture/group-selection dust bunny to the more formal models of group selection discussed by Queller and endorsed by Henrich, Wilson, Gintis, Boyd & Mathews, and Richerson.^[7] Wilson and Henrich suggests that criticisms of group selection are mathematically naïve, and that formalizing social evolution inevitably leads to the conclusion that group selection is a significant evolutionary force. Yet the old joke about the theoretical physicist called in to help a farmer increase milk yields ("First, assume a spherical cow") reminds us that mathematical models are only as good as the plausibility of their assumptions and their applicability to real problems. There are many ways of formalizing social evolution, as we see in the burgeoning industry of iterated prisoner's dilemma tournaments, and it's not enough to show that some of them contain a term that may be called "group selection." As Coyne notes, the question is whether these models are realistic and predictive.

The most obvious way to formalize group selection would be to treat groups themselves as replicators. As Dawkins, Dennett, and Gintis point out, the logic of natural selection does not restrict it to genes, and if groups fissioned into daughter groups that preserved the group-level traits of their progenitors and competed in a meta-population of groups, the theory of natural selection would apply to them, and group selection would be a perfectly coherent concept.

Henrich, Boyd, Richerson, and Gintis try to put as much distance as possible between their own views and this sense of group selection, and Wilson in particular has tried to bury it, insisting that for several decades now it has been a blunder to think of group selection in that old-fashioned way. But this lexical policing is problematic. For one thing, it is not true that no contemporary biologist uses "group selection" to refer to the evolutionary dynamics of group replication (see, for example, the models of Burton Simon) ^[8]. Second, the sense that Wilson wants to inter is exactly the one in which the word "selection," the analogy with gene selection, and the explanatory power of the theory of natural selection are used to full advantage. There would be far less confusion if this sense were retained (even if it turned out, as a matter of empirical fact, that real-world groups are lousy replicators and so group selection in this sense doesn't happen), and the others were given new labels.

Let's now turn to models of this new theory of group selection. According to Queller and the members of the Group group, these models aren't really different from models of gene selection (aka inclusive fitness). The two are synonymous languages, mathematically equivalent formalisms, alternative accounting methods, or intertransformable coordinate systems. Now, this position creates a tension in the commentaries that I mentioned at the outset: group selection can't simultaneously be a notational variant of gene selection which offers no contrasting empirical predictions, and a theory that uniquely predicts the existence of phenomena such as morality and generosity. But let's concede for now that a notational variant can prove its empirical usefulness by uniquely spotlighting certain phenomena as distinct variables in the model and showing that these phenomena are potent forces in nature.

In most models of the new group selection, a group is defined as any subset of interacting individuals, that is, as organisms which interact with one another more intensely than they interact with organisms selected from the population at random.^[9] Two sisters who help each other, for example, or a pair of friends who trade favors, are dubbed "a group" It's not hard to see how such "groups" would be favored by natural selection over pairs of individuals who let each other drown or starve, nor how this way of describing the process could be translated into the language of inclusive fitness and vice-versa, as long as some mechanism kept subsets of mutual helpers together. If the mechanism is common descent and limited dispersal, group selection is more or less equivalent to kin selection; if it is a tendency of altruists to preferentially affiliate with each other, it is more or less equivalent to reciprocal altruism.^[10]

If the two theories really are equivalent, then any advantage of group selection (in this new sense) would have to come from the models' being more convenient, elegant, simple, transparent, explanatory, or mathematically tractable. Yet by stretching the meaning of "group" beyond its ordinary sense, that's just what they fail to be. According to the old song, "We belong to a mutual admiration society, my baby and me"—the whimsy hinging on the unnaturalness of referring to a pair of individuals as a "society." The same is true for "group." While mathematically speaking one can identify a "group" with any arbitrary set, in practice using a single construct for a pair of siblings, a person holding a door open for a stranger, a waitress and a customer, a married couple, a street gang, a traditional band or tribe, a nation, and an empire conceals the significant psychological differences among them.

More generally, a mathematical model that submerges the psychological forces that keep different "groups" together (such as genetic relatedness or mutual sensitivity to altruism), and requires theorists to dig them out from under the equations, is hardly a perspicuous way to analyze sociality (as Coyne points out). In gene-selectionist theories, the theoretical constructs that power the models turn out to be fantastically psychologically important, including sensitivity to kinship, scrutiny of individuals, moralistic emotions elicited by benefits conferred or withheld, and the psychological differentiation of relationships into discrete models corresponding to mutualism, kinship, reciprocity, and dominance.^[11] Price, Krasnow & Delton, and Baumard point to a growing body of evidence that these emotions and aptitudes drive the empirical phenomena most often adduced in support of group selection, namely generosity and punishment in experimental games.

Now let's turn to a specific way in which the variables highlighted in the new group selection models might be connected to human affairs. Once a "group" is defined as a subset of interacting individuals, the variance in the fitness of individuals can be partitioned into two statistical components: how fit the individual is with respect to his groupmates, and how fit his group is with respect to other groups. This too may be a mathematical truism, another way of describing the fact that one of the ways in which I can help myself may be to help a group I am part of. Examples include huddling for warmth, mobbing a predator, and Tooby's example of pooling resources to get higher expected returns in a risky investment. In such cases one can separate the benefits that accrue to the entire group (including me) and whatever benefits or costs are assumed by me but no one else in the group. According to group selectionists, this formulation perspicuously explains the evolution of altruism because it includes cases in which a benefactor to the group suffers in comparison to his groupmates; the payoff trickling down from the group's benefit can exceed the cost he pays within the group.

Mathematical biologists such as Alan Grafen, Stuart West, Ashleigh Griffin, and Andy Gardner have criticized this formulation because it obfuscates the fact that individuals are still maximizing their genetic fitness: "The fundamental point is that the spread of a gene is determined by its 'fitness relative to others in the breeding population, and not to others with which it happens to interact.' … Natural selection selects for a gene if it causes a behavior that leads to that gene increasing in frequency in the population, not some other arbitrarily defined scale such as social partners." ^[12] Their fears about people getting confused are not groundless: some group selectionists allude to these models as explanations of group-benefiting acts that systematically decrease the actor's overall fitness, such as voluntary suicide in combat. (Remember Haidt and E. O. Wilson's likening of humans to social insects, whose soldiers and workers sacrifice themselves for the colony; since any human version of such martyrdom cannot be explained by the queen-and-drone reproductive system of social insects, the more general between/within-group accounting system is supposed to do the trick.) Yet as Tooby notes (echoing Grafen, West, et al.), one point of agreement between group- and gene-selectionists is (or ought to be) that apart from kin selection, "traits cannot evolve that cause an average net decrease in individual fitness."

So are there partitionings of fitness that involve costly self-sacrifice but that do add up to a net individual benefit, thanks to the positive effects spread over the group? Group selectionists often allude to the following theory. First, a major selective force in human evolutionary history has been competition between coalitions, often in the form of lethal combat (this is made explicit, for example, in several papers by Gintis's collaborator Samuel Bowles ^[13] ). The outcome of this combat can affect most or all of its members: they may all die, or they may all obtain additional resources. This causes their fitnesses to covary, and thus it makes the between-group variance component substantially greater than zero. Second, victory in such combat is largely determined by which coalition has the greater degree of individual self-sacrifice among its members, such as charging headlong into battle, shielding comrades from enemy projectiles, rationing one's own consumption to feed and equip the warriors, backing down in internecine squabbles that would weaken the coalition, and risking one's safety to punish cowards and shirkers. The within-group fitness of these self-sacrificing individuals would be lower than their comrades who selfishly look out for themselves, yet if battles were frequent and decisive enough, the jackpot of victory or the catastrophe of defeat in the competition between groups level could have outweighed the disadvantage of self-sacrifice in the competition within groups. Therefore humans were selected to make sacrifices on behalf of their coalitions, and models of group selection that distinguish within- and between-group fitnesses can transparently explain the evolution of human altruism, including generosity, empathy, sharing, bravery, and civic-mindedness. Gintis echoes the summary offered by the Wilsons, Haidt, and others: "Groups with many altruists will expand at the expense of groups with few or no altruists."

Thus we have an explanation of why the world is divided into the empires of the Amish and the !Kung, whose barn-raising and food-sharing allowed them to overpower rival groups weakened by internal selfishness. By the same token the model explains why those selfish groups, with their harem-holding despots, ruthless warlords, conscript armies, and exploited slaves and serfs have been so rare and short-lived in human history. It readily explains why warrior societies are distinguished by their charity, compassion toward the weak, and equality of women. And it helps us understand why empathy and kindness blossom in those societies that suppress individualism and promote elevation of their nation based on a commitment to an organic community united by ancestry and culture—a dictionary definition of "fascism," from the Italian word for "bundle."

The point is that even if the selfless-coalition sense of group selection is mathematically sound, it is a feedlot of spherical cows. In the real world, success in group-against-group competition is not primarily determined by differences in within-group empathy and kindness but by differences in ideology, technology, military strategy, and organization coerced by brutal discipline, none of which need be costly to the individuals who implement them. Conversely, many forms of real-world altruism provide no advantage in group-against-group competition–particularly compassion for the weak and needy, who are drags on military effectiveness and should be the first to be thrown overboard. Richerson cites the studies of empathy by Daniel Batson, but the typical targets of empathy in his studies are needy individuals such as a student with AIDs or a puppy with a broken leg; Batson himself proposed that empathy evolved from maternal care, the quintessential example of kin selection. ^[14] So the models of "groups" as self-sacrificing aggressive coalitions explain the evolution of altruism only by building it in, through the stipulation that within-group altruism is the primary cause of success in between-group competition.

One could, of course, switch models at this point and say that military effectiveness and other contributors to group success arose not by biological group selection but by cultural group selection (see, for example, Boyd & Mathews). This brings us to the third bone of contention.

Does Cultural Group Selection Add Anything to Conventional History?

Henrich, Richerson, and Boyd & Mathews object to my skepticism that cultural change is well explained with an analogy to natural selection. But they equate skepticism about this theory of cultural evolution with a denial that cultures exist, cohere, function effectively, and foster helpful norms and innovations among their members (including, sometimes, genetic changes such as lactose digestion in adult livestock herders). But reminding us of the fact that cultures change in adaptive ways does not prove that an analogy to natural selection is a good explanation of those changes. It's not that I object to mathematical models inspired by biology (including the groundbreaking work of Richerson and Boyd), particularly those that use the tools of epidemiology to explain how innovations can spread among individuals and change a culture. ^[15] My point is that natural selection in particular is a poor model of cultural change. Dennett nicely captures one of the reasons: "To call a phenomenon group selection which seldom if ever involves the differential replication (as contrasted with survival or growth) of groups seems to me to be gratuitously misleading."

The other reason applies as well to Dennett's own defense of the theory of memetics. ^[16] Human cultural innovations are quite different from the blind mutation and recombination that supply the raw material for bona fide natural selection. I'm happy to concede that this is a difference in degree. Sure, natural selection does not require blind mutation; it can add value to intelligently designed or directed variants. And sure, the brainchild of a single innovator generally must be tinkered with and combined with other innovator's brainchildren through social networks before it is of any use.

But what a difference in degree! Genetic mutations and recombinations are strictly typographical, twiddling the As, Cs, Ts, and Gs with no foreknowledge of their effects on the organism's interactions with the world. As Dennett, Dawkins, and Tooby emphasize, natural selection is an astonishing theory because it explains how biological design can emerge from this mindless background. But the trillion-synapse human brain, even when it is daydreaming, tinkering, riffing, brainstorming, or jamming, does not blindly substitute one phoneme or word or ingredient for note for another and live or die by the consequences. Genetic mutations are monkeys at a typewriter; human innovations are not. Millions of years of natural selection for know-how have equipped the brain to develop mental models of the world that prune the tree of potential innovations to an infinitesimal sliver of the space of typographical possibilities. To reduce the function of the human brain to a mutation generator, even a "nonrandom" one, is to bring back B. F. Skinner's empty organism and to miss out on the outsize adaptations that make human cognition, cultural learning, and sociality so unusual in the animal kingdom. It's not that analogues of natural selection have nothing to add to our understanding of cultural change. But unlike the case of genetic evolution, where selection assumes the full burden of generating adaptation from the vast space of genetic possibilities, most of the work done in exploring the space of logically possible ideas must be attributed to the organization of the brain.

When it comes to trying to explain cultural change in practice, I can speak from some experience, because I have devoted substantial chunks of my career to explaining two examples: diachronic change in language, and historical declines of violence. ^[17] No one could be more sympathetic to the application of evolutionary biology to human affairs than I am, and I have made use of many of its tools. But group selection and memetics have been unhelpful, and even evolutionary psychology in its totality can take us only so far. That is because human cultural change is driven by ideas. In the case of language, they are the lexical and grammatical analyses by which listeners make sense of the speech of others; in the case of violence, they are ideologies by which people justify their collective actions, such as religions, Marxism, nationalism, utilitarianism, enlightenment humanism, romantic militarism, and many others. If you reduce these ideas to simple tokens that are spread by contagion or multiply at different rates, and don't considering how their content affects the beliefs and desires of human protagonists, you will end up with a seriously incomplete understanding of cultural change. Nothing about this is mystical or unscientific when you remember that the human brain evolved as a mechanism that generates ideas and that our species occupies a niche in which this capacity is effusively put to use.

References

Batson, C. D., Ahmad, N., Lishmer, D. A., & Tsang, J.-A. (2002). Empathy and altruism. In C. R. Snyder & S. J. Lopez (Eds.), Handbook of Positive Psychology (pp. 485-498). Oxford: Oxford University Press.

Bowles, S. (2008). Conflict: Altruism's midwife. Nature, 456, 326-327.

Bowles, S. (2009). Did warfare among ancestral hunter-gatherers affect the evolution of human social behaviors? Science, 324, 1293-1298.

Boyd, R., & Richerson, P. (1985). Culture and the evolutionary process. Chicago: University of Chicago Press.

Daly, M. (1982). Some caveats about cultural transmission models. Human Ecology, 10, 401-408.

Darmuth, J. H., I. L. (1988). Alternative formulations of multilevel selection. Philosophy and Biology, 3, 407-430.

Fiske, A. P. (1992). The four elementary forms of sociality: Framework for a unified theory of social relations. Psychological Review, 99, 689-723.

Frank, R. H. (1988). Passions within reason: The strategic role of the emotions. New York: Norton.

Grafen, A. (1984). Natural selection, kin selection, and group selection. In D. Krebs & N. B. Davies (Eds.), Behavioral ecology: An evolutionary approach (pp. 62-84). Oxford: Blackwell.

Hruschka, D., & Henrich, J. (2006). Friendship, cliquishness, and the emergence of cooperation. Journal of Theoretical Biology, 239(1), 1-15.

Maynard Smith, J., & Warren, N. (1988). Models of cultural and genetic change. In J. M. Smith (Ed.), Games, sex, and evolution. New York: Harvester-Wheatsheaf.

McCullough, M. E. (2008). Beyond revenge : the evolution of the forgiveness instinct (1st ed.). San Francisco, CA: Jossey-Bass.

Okasha, S. (2009). Evolution and the levels of selection. New York: Oxford University Press.

Pinker, S. (1997). How the mind works. New York: Norton.

Pinker, S. (1999/2011). Words and rules: The ingredients of language. New York: HarperCollins.

Pinker, S. (2010). The cognitive niche: Coevolution of intelligence, sociality, and language. Proceedings of the National Academy of Science, 107, 8993-8999.

Pinker, S. (2011). The better angels of our nature: Why violence has declined. New York: Viking.

Queller, D. C. (1992). Quantitative genetics, inclusive fitness, and group selection. The American Naturalist, 139(3), 540-558.

Simon, B. (2010). A dynamical model of two-level selection. Evolutionary ecology research, 12, 555-588.

Simon, B. (under review). Towards a general theory of group selection. Retrieved from http://www.math.ucdenver.edu/˜bsimon/Simonetal2012b.pdf

Sperber, D. (1985). Anthropology and psychology: Towards an epidemiology of representations. Man, 20, 73-89.

Symons, D. (1992). On the use and misuse of Darwinism in the study of human behavior. In J. H. Barkow, L. Cosmides & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture. New York: Oxford University Press.

Tooby, J., & DeVore, I. (1987). The reconstruction of hominid evolution through strategic modeling. In W. G. Kinzey (Ed.), The evolution of human behavior: Primate models. Albany, N.Y.: SUNY Press.

Trivers, R. L. (1971). The evolution of reciprocal altruism. Quarterly Review of Biology, 46, 35-57.

West, S. A., Griffin, A. S., & Gardner, A. (2007). Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. Journal of Evolutionary Biology, 20, 415-432.

West, S. A., Griffin, A. S., & Gardner, A. (2008). Social semantics: How useful has group selection been? Journal of Evolutionary Biology, 21, 374-385.

Wilson, D. S. (1977). Structured demes and the evolution of group-advantageous traits. American Naturalist, 111, 157-185.

Wilson, M., & Daly, M. (1997). Relationship-specific social psychological adaptations. In G. R. Bock & G. Cardew (Eds.), Characterizing human psychological adaptations. New York: Wiley.

[1] For others, see (Darmuth, 1988; Grafen, 1984; Okasha, 2009; West, Griffin, & Gardner, 2007, 2008).

[2](Pinker, 2010; Tooby & DeVore, 1987).

[3]See (McCullough, 2008), for a review.

[4](Trivers, 1971).

[5]See (Frank, 1988), for a particularly illuminating development of this point.

[6](Hruschka & Henrich, 2006).

[7](Grafen, 1984; Okasha, 2009; Queller, 1992; West et al., 2007, 2008; Wilson, 1977).

(Simon, 2010; Simon, under review)

[9](See, for example, Queller, 1992, p. 548).

[10]See (West et al., 2007, 2008).

[11](Fiske, 1992; McCullough, 2008; Trivers, 1971; Wilson & Daly, 1997).

[12](West et al., 2007), pp. 421, 422; (Grafen, 1984).

[13](Bowles, 2008, 2009).

[14](Batson, Ahmad, Lishmer, & Tsang, 2002).

[15](Boyd & Richerson, 1985; Sperber, 1985).

[16](Daly, 1982; Maynard Smith & Warren, 1988; Sperber, 1985; Symons, 1992); see also (Pinker, 1997), chap. 3.

[17](Pinker, 1999/2011; Pinker, 2011).

Founder of field of Evolutionary Psychology; Co-Director, UC Santa Barbara's Center for Evolutionary Psychology

Genic Selection and Adaptationism: Are We Moving Forward or Back?

It is a welcome relief that Steve Pinker has applied his usual acuity to the ever-vexatious debate on group selection. The first and deepest problem with this debate is that the term group selection does not have any single fixed meaning, but has been used over the last half century to convey a huge and tangled thicket of different and conflicting meanings. The great majority of these are seriously defective as a way of describing reality. Various entries in this forum provide good examples of how this terminological black hole still is a vector for spreading confusion: it invites us down garden paths into conceptual tarpits; it leads us to talk past each other; it gives the appearance of life to long dormant and justly discredited ideas, by conflating them with fundamentally different but potentially viable hypotheses; and it even leads us to confuse ourselves about what exactly we are referring to. This multiplicity and indeterminacy of meanings is reason enough to imprison the term in adamantine chains and pitch it into the Marianas Trench, limiting its future mischief to unwary coelacanths and unusually credulous archaebacteria. Banning the term (sadly, a utopian fantasy) might force us all to adopt terminology that clearly distinguishes distinct theories, and distinct (hypothetical or actual) phenomena. This would allow these issues to be cleanly sorted out and settled within our lifetimes, rather than postponing this to the next millennium. Other meaning-chameleons that sow similar confusion are moral, altruistic, and especially selfish. For example, using the definition of selfishness and altruism that biologists use, a loving and self-sacrificing mother is acting selfishly, while a drug addicted mother who starves her children to give all her money to her dealer is an altruist (i.e., she is lowering her own fitness in a way that increases a nonrelative's).

Those new to the group selection debate may not know how truly problematic thinking was about these issues before George Williams and a new generation of evolutionary scientists ushered in the genic selection revolution [1]. For those (still) not exposed to this revolution or working before it, group selection was mostly just a terminological placeholder for a ubiquitous, promiscuous, fuzzy evolutionary teleology that permeated both scientific and popular thinking. This luxuriant, pervasive, unanchored teleology choked off productive evolutionary reasoning: In the biological world, any good effect on anything else was supposedly explained by Darwinism's benign collectivism: Plants produce oxygen for animals to breathe; predators target the sick and the old to keep the unfit from reproducing, maintaining the balance of nature. The Nobel Laureate Konrad Lorenz, in his much lauded book On Aggression taught social scientists, animal behaviorists, and the educated public that dominance hierarchies and ritualized aggression evolved for the good of the population or species—for example, it was good for everyone if the unfit ceded resources to the fit and dominant; inferiors collaboratively joined in their eugenic self-removal because their adaptations were altruistically designed to submit nonviolently to their betters. This collaborative weeding out of the inferior was just one of an endless series of phenomena supposedly explained by a Darwinism that was held to operate by retaining whatever traits contributed to the "survival" of the group, population, species, or balance of nature (for a modern example, think Gaia hypothesis). The zoologist V.C. Wynne-Edwards famously argued that group selection widely selected for adaptations that sacrificed the individual's reproduction so that the local group would not exceed the carrying capacity of the environment. Psychology and anthropology were similar train-wrecks—a typical case being the prominent anthropologist Marvin Harris' early version of cultural group selection, in which he argued that warfare and male-supremacy were a culturally group selected form of population control.

By the laborious application of logic, conceptual clarity, mathematical modeling, and empirical tests, this jungle of misconceptions was largely cleared away, and evolutionary biology was rebooted on the far more rigorous and productive foundation of adaptationism coupled to genic selectionism. This generational act of intellectual reclamation was the most important advance in evolutionary biology since Darwin. As Darwin pointed out, "[t]o kill an erroris as good a service as, and sometimes even better than, the establishing of a new truth or fact." This rebooting not only cleared away a huge tangle of errors but in so doing, unleashed a flood of theoretical and empirical advances that are refashioning our understanding of the nonhuman and human worlds. The core of the genic selection revolution is the insistence that selectionist arguments posit a clear pathway of material causal steps that produces feedback from the effects of genes to their subsequent proliferation. That is, theories must specify an explicit chain of causation that leads from a given set of genes, to their recurrent (somatic or extended) phenotypic effects (the adaptation), to the adaptation's interactions with the world; these must have the net consequence of multiplying the frequency of replicas of those genes in subsequent generations.

Agreements and disagreements: Now that the revolution is over, what are we arguing about?

A variety of researchers have now introduced of a small and diverse set of new models of hypothetical evolutionary (e.g., genetic) or dynamical (e.g., cultural) processes which they unnecessarily (in my view) insist on labeling as group selection. (Usually, these turn out to be consistent with and traceable to the genic selectionism revolution, although many of these models depend on numerous restrictive assumptions which may or may not allow them to be applied to the real world.) Readers might detect a note of testiness here and there in the responses to these proposals. The reason why is that those who have worked long and hard to clear out the jungle, and to lay a rigorous foundation for modern evolutionary biology see this progress jeopardized by the rhetorical conflation of faulty or implausible (and dormant) ideas with these new models. The problematic but infectious strains of group selection thinking and fitness teleology are, unfortunately, highly attractive to our evolved intuitions, so small missteps are all it takes to release them from their proper home in biological containment facilities and into the general intellectual population.

Nevertheless, I strongly agree with the new group selectionists on one point: One should never be dogmatic, but should examine all hypotheses carefully without prejudice or partisanship. Indeed, the biological world is so rich and multifarious that it would be foolhardy to say that any particular logical possibility has never been realized anywhere in the seven (or so) kingdoms of life. For example, Paul Ewald's elegant work on parasite virulence and its attenuation invites a group selection analysis; i.e., when encounters with susceptible new hosts are rare, those hosts populated with more deadly strains die before infecting new hosts, ratcheting down average virulence across the parasite meta-population. (However, given asexual reproduction in many parasites, one could just as easily consider the strain inside a host a single biological individual, and not a group.)

Like the British and the Americans, evolutionary behaviorists are separated by a common language. So, it may be useful to see if apparent disagreements (where people are talking past each other) can be resolved into points of agreement, so that we can better understand where any remaining disagreements lie. I am guessing most or all participants:

(1) accept and value the genic selectionism revolution;

(2) think that for mammals and similar organisms, the great majority of the traits of organisms are the product of selection to promote inclusive fitness (for simplicity I'll call this individual selection), without needing to invoke processes operating at higher levels;

(3) agree that multilevel selection theory is not actually a theory, but one of several mutually consistent ways of representing theories; Dave Queller calls these alternatives different languages, and Joe Henrich felicitously likens them to different coordinate systems; multilevel selection models that invoke group selection are equivalent to one particular family of kin selection models;

(4) agree that traits can be beneficial to the individual, but costly to the group; beneficial to the individual by virtue of being beneficial to the group; costly to the individual, but beneficial to the group, etc.

(5) agree that humans could hypothetically be designed so that they are far more beneficial to the groups they inhabit than they actually are, but evolution has not made them so;

(6) agree that traits can evolve in which a functional phenotype is expressed across individuals or by a group; there are innumerable examples, including hive architecture, waggle dancing, group hunting, or hunter-gatherer sharing as risk pooling;

(7) agree that traits cannot evolve that cause an average net decrease in individual fitness (other than by kin selection).

Agreement on point 7 reflects the extinction of pre-revolutionary patterns of thought, in which individuals supposedly sacrificed themselves for all kinds of implausible ends, from maintaining maximum sustainable species biomass to implementing group eugenics programs to lowering the extinction probability of the species.

Let's pause now to consider some issues where people seem to be talking at cross-purposes. Those who prefer to phrase things in terms of group selection often seem to believe that those who don't deny the existence of functional group phenotypes. They seem also to equate the existence of functional group phenotypes as evidence for group as opposed to individual selection. In contrast, those who prefer to represent things in terms of individual selection are perfectly happy to recognize the existence of functional phenotypes composed of traits distributed across multiple individuals. However, they see no reason to interpret this as necessarily individually costly, and therefore necessarily falling into the category of things that require group selection explanations. Also, there is interpretive choice in whether one considers the functional phenotype as "really" existing at the individual level (which it must to initiate the feedback pathway), or as "really" existing at the group level (which it must to complete the feedback pathway—eventually, multiplying gene frequencies). Collaborative hunters must have adaptations in individual brains that cause them to hunt in groups, and must also hunt in groups (a group-level functional phenotype) to reap the advantage. Individual selectionists consider this a case of individual selection, because individuals have hunting adaptations, and these increase individual fitness. Group selectionists consider it a case of group selection because fitness increases come to the groups that manifest the functional group phenotype, and not to those that don't. As long as the phenotypic adaptation is correctly described at both the individual and the group level (necessary work that is frequently not done), then the approaches are different in name only.

It is important to recognize that group functional selection pathways do not violate intuitive ideas of self-interested behavior. The individual acts in a way that (on average) makes her better off, with increasing group welfare just being a means through which individual welfare (approximately, inclusive fitness) is increased. Imagine, for example, a lottery, in which there are disproportionate increases in the expected value of lottery tickets the more expensive they are. A dollar ticket has, say, an expected return on one dollar, while a four dollar ticket returns sixteen dollars. If individuals each have one dollar, then those who pool their money to buy four dollar tickets will do better. Nothing about this involves acting against self-interest or receiving a lower payoff. Indeed, since multilevel group selection simply partitions what once was considered individual fitness into two types (or more) of what should also be considered individual fitness, it would be simpler, less confusing, and less retrograde to simply describe it that way.

Still, there are two sets of cases worth distinguishing where group-functionality is driving a selective advantage. In one case, within-group selection selects against the group-functional adaptation in the individual; in the other case, selection pressures on the adaptation are aligned between the individual level and the group level (or at least within-level selection doesn't act against it). The evolution of group functional traits will be far easier and hence more common, other things being equal, when selection is parallel or at least not opposed between the two levels. Yet, curiously, the new group selectionists tend to focus on more problematic cases where within-group selection (hypothetically) puts individuals at a relative disadvantage, recalling the old group selectionists' emphasis on sacrificing for the group. It may tell us something about the strength of group selection in humans that clearcut cases of genetic traits selected against within the group, but favored at the group level have not yet been established.

Multilevel selection theory, heterarchic pathway feedback theory, and competition for phenotypic expression

A major problem I have with multilevel selection theory is its imposition of a hierarchy of nested class inclusion on a biological world that seems to have a very different structure. I favor an alternative that might as well be called heterarchic pathway feedback theory. In general, selection will favor any alleles that establish a positive feedback causal pathway between their effects and their subsequent frequencies. In general, we are better off modeling, detecting and inventorying these different pathways, rather than attempting to integrate them into a single unitary theory of fitness. These pathways need not be, and often will not be aligned, mutually consistent, or representable as operating at different levels in a hierarchy, but instead will often be cross-cutting and heterarchical. For example, Leda Cosmides and I published a paper introducing a general theory of intragenomic conflict [2], describing how individuals were actually better conceptualized as colonies akin to social insect hives, because the fitnesses of different subsets of the genome were maximized in mutually inconsistent ways: Mitochondria, flagella, the Y chromosome, the paternal genome, the maternal genome—all had different fitness interests, and therefore selected for adaptations that were sometimes designed to disrupt each other (see, e.g, David Haig's work on conflict within the fetus between the maternal and paternal genome). To take one out of dozens of examples, we found evidence supporting the prediction that across a broad range of species, intragenomic conflict disrupted designs in flowering plants in which both male and female functions were combined in the same individual. One set of adaptations or the other were kept from phenotypic expression. Similarly, different social strategies may select for different and potentially conflicting adaptations both within individuals and in group functional phenotypes. Adaptations for reciprocation may conflict with, e.g., adaptations for kin-directed altruism or adaptations for coalitions; or adaptations for kin-directed altruism may conflict with aposematic coloration. Group-functional adaptations may particularly suffer from such disruption, because the total number of possible alternative partitionings into different groups is high, and different partitionings may select for different and mutually inconsistent phenotypic functions, with perhaps the strongest selection pressures winning out. (Indeed, advocates of multilevel selection in different analyses routinely posit different group structures that do not map on to each other; e.g., modeling kin selection as group selection; modeling reciprocation as group selection. In fact, the same individual may be modeled as being in hundreds of dyadic reciprocation groups over the course of a lifetime.) The existence of heterarchic feedback pathways may lead to competition among different adaptations for expression in the phenotype—one of many issues not theoretically well-explored, or well-captured by multilevel selection theory.

Effects on scientific practice

Some aspects of the new group selection research program strike me, at least, as perhaps contributing to the erosion of good scientific practice. Results that straightforwardly falsify theories are presented as if they supported them. [3] The claimed "failures" of individual selection explanations for various empirical findings are extremely weak and polemical (e.g., Henrich's claim that evolutionary psychological approaches predict cross-cultural uniformity in measured behavior is simply baffling, since adaptations like the language acquisition device or reciprocation systems take local inputs as parameters, and generate variable output. [4] Moreover, the eagerness to leap to conclusions has warped good experimental design. For example, experimental designs testing for the existence of "altruistic punishment" are full of what psychologists would see as clear experimental artefacts. There is a strong response bias built into the experiment (subjects can only punish or do nothing), and such a design, even subject error counts as altruistic punishment. Indeed, the extravagant explanatory claims made for many of these models (about altruism, punishment, morality, fairness, etc. in humans) have been accompanied by a growing neglect of testing for or even bothering to characterize the associated neurocomputational adaptations which constitute their phenotypic reality.

Obviously researchers should be free to choose to work with any coherent representational system they like, and I continue to hope that the creative people working in this area may discover important new insights about humans using multilevel selection theory or cultural group selection theory.

________

_{Cosmides, L. and Tooby J. (1981) Cytoplasmic inheritance and intragenomic conflict. Journal of Theoretical Biology. 89: 83-129.}

_{Delton, A. W., Krasnow, M. M., Cosmides, L., & Tooby, J. (2010). Evolution of Fairness: Rereading the data. Science, 329(5990), 389.}

_{Tooby, J. & Cosmides, L. (2010). Groups in mind: Coalitional psychology and the roots of war and morality. In Høgh-Olesen, Henrik (Ed.), Human morality and sociality: Evolutionary and comparative perspectives.(pp.191-234) Palgrave Macmillan.}

_{[1] Best articulated in George Williams 1966 classic, Adaptation and Natural Selection, and in Richard Dawkins' 1982 The Extended Phenotype}

_{[2] Cosmides & Tooby, 1981}

_{[3] See discussion in Delton, Krasnow, Cosmides & Tooby, 2010}

_{[4 See, Delton et al, 2010}

Co-Director of LSE's Centre for Philosophy of Natural and Social Science; Author, The Ant and the Peacock: Altruism and Sexual Selection from Darwin to Today

When You Hear The Word 'Culture', Don't Reach For Your Metaphor

The eminent philosopher sat frozen in horror, forkful of lunch poised between plate and mouth. What enormity had caused the shock? The conversation had turned to cultural evolution. And I had suggested that there is no such thing. There's culture; there's history; there's change; there's progress; there's technological innovation; there's growth of knowledge; there's social learning; and there's lots more. But there's no cultural evolution.

Steven Pinker's criticism of group selection is an invaluable exposé. And it is doubly welcome because it also, no less forcefully, skewers the equivalent failings of cultural evolution. Pinker's accounts of how natural selection works, and why group selection could not work that way, apply (as he briefly notes), mutatis mutandis, to cultural evolution.

So how should we understand culture—its richness, inventiveness and multifarious wonders; its growth, diversification and proliferation; its uniqueness to our species; its universalities, deep in time and across all human groups today; its quirky, distinctive, idiosyncractic manifestations; its cumulative and progressive aspects? Pinker has answered this, too—and without the slightest need for any groupishness. All that is needed is a discerning evolutionary understanding of human nature and some rigorous, detailed work on history.

On culture and human nature, for perspectives that are of special relevance, see Pinker's paper on our "cognitive niche"[i] which, not least, deals with the problem of how natural selection came to endow our species with adaptations that enable us to perform cognitive feats so advanced that they would have been surplus to the adaptive requirements of our ancestors. (As Wallace, Darwin's co-discover of natural selection, put the problem: "Natural selection could only have endowed savage man with a brain a few degrees superior to that of an ape, whereas he actually possessesone very little inferior to that of a philosopher"[ii]. By the way, I'm happy to report that Darwin disagreed with Wallace on both the purported impotence of natural selection and the purported superiority of philosophers.)

On culture and history, Martin Daly and Margo Wilson's classic study of homicide[iii] pioneered the way to go; and Pinker's work on "the decline of violence in history and its causes" [iv] can be seen as a systematic demonstration of how right these pioneers were.

So, pace the eminent philosopher, when you hear the word 'culture', don't reach for your metaphor. I wasn't talking heresy—just "what we call history"[v].

____________

[i] Pinker, S. (2010) The cognitive niche: Coevolution of intelligence, sociality, and language. Proceedings of the National Academy of Sciences of the United States of America, 107 (supplement 2), 8993—9

[ii] Wallace, A. R. (1891) Natural Selection and Tropical Nature: Essays on descriptive and theoretical biology (Macmillan, London, p. 202)

[iii] Daly, M. and Wilson, M. (1988) Homicide (Aldine de Gruyter, New York)

[iv] Pinker, S. (2011) The Better Angels of our Nature: The decline of violence in world history and its causes (Viking Penguin, New York)

 [v ] Pinker, S. (2010) The cognitive niche: Coevolution of intelligence, sociality, and language. Proceedings of the National Academy of Sciences of the United States of America, 107 (supplement 2), p. 8996

Evolutionary Biologist; Emeritus Professor of the Public Understanding of Science, Oxford; Author, The Greatest Show on Earth, The Magic of Reality
 

"Group Selection" Is A Cumbersome, Time-Wasting Distraction

Steven Pinker is right that "group selection" confuses more than it clarifies. 'Groupishness' is not evidence for group selection, nor, for heaven's sake, is the tendency for individuals to benefit from living in groups! Pinker's fifth paragraph nails most of the confusions so well that I could do no better than quote it verbatim, so I won't. Nor shall I talk about ecological replacement, nor  'species selection' nor 'clade selection' [1], all of which have contributed to the irritating muddle of ideas that have been, at one time or another, subsumed under "group selection". And I shall not talk about humans. The issue at hand has been obscured enough without dragging in the one species that flouts almost every rule. What I shall do is sort out the argument on units of natural selection. The key is to abandon the word 'fitness', replacing it with the concepts of replicators, vehicles and the extended phenotype. What follows will necessarily be extremely abbreviated. The full argument is in The Extended Phenotype.

What is really going on in natural selection is this. As the generations go by, entities defined as successful take some action to become more numerous in a pool of such entities, while rival entities defined as unsuccessful become less numerous. Entities that take action to become numerous consist of information, instantiated in physical objects which they outlive. They are 'replicators'. The archetypal replicator is DNA information, which temporarily resides in short-lived DNA molecules. Successful DNA information can survive, by duplication, through a hundred million years, while the lifespan of a physical DNA molecule, or of an individual animal, is never more than a few years. Other replicators are computer viruses (which temporarily reside in computer hardware) and memes (which temporarily reside in brains). Among replicators I shall talk only of DNA information—the 'codices' of George Williams [2]—which we can think of as 'genes' in his 1966 sense.

Successful genetic replicators, then, take action to ensure their survival. What kind of action do they take? Phenotypic action.  A phenotype is usually the end product of a cascade of caused events. Commonly, the first member of the cascade is a protein chain, whose sequence and hence 3-D folding structure is determined by the DNA information. The second step in the cascade is then typically enzymatic catalysis by that protein of a biochemical process in a cell. This may be the phenotypic effect of interest, that is to say the one that sorts successful replicators from unsuccessful ones. In other cases the cascade continues, perhaps through biochemically mediated events in cell membrane dynamics, then perhaps through consequences in embryonic morphogenesis, then perhaps through consequences in brain wiring, then perhaps through consequences in adult behaviour and so on. The key point is that, however complicated and intricate such cascades may be, and however richly the different genes may share phenotypic cascades with each other, replicators qualifying as successful will—on average, over many instantiations and many generations—tend to find themselves associated with good phenotypes, while replicators defined as unsuccessful will find themselves statistically associated with bad phenotypes. 'Good' and 'bad' phenotypes are defined as good and bad at passing on the replicators responsible for them.

So far I have said nothing about individual organisms (or groups). It incidentally happens to be the case that, at least on our planet, phenotypic effects of genes are mostly bundled in large clusters which we call organisms. The organism is an agent, a large, well-integrated unit of action, whose sensory, motor and computational parts collude, often in highly complex ways, towards one end—reproduction (strictly gene survival).

An organism is a coherent bundle of phenotypic effects, integrally combined with each other in such a way that, with few exceptions such as segregation distorters, the survival of each gene is bound up with the survival of every other one. The primary quality a gene needs, in order to be successful, is the capacity to cooperate with the other genes with which it is statistically likely to share a body. And in sexually reproducing eukaryotes this means the other genes in the gene pool of the species. The reason genes cooperate with each other in building bodies is that all share the same exit route from the present body (a long succession of 'present' bodies) into the next body—the next generation. All the genes in an individual lion share the same exit route to the future —the sperms or eggs of their shared body. All stand to gain from their particular lion's success in hunting, courting, seeing off rivals, parental care. A lion is a vehicle for its genes, constructed by a cooperative of genes as a shared ark to carry them forward to the future, house them and protect them until it is ready to pass them on.

If a parasite, such as a bacterium, happens to use the same exit route as the lion's own genes, namely the lion's gametes, the 'interests' of the parasite genes will overlap with the host's genes and may even be identical. Unlike a bacterium which is passed on by, say, coughing, a gametically transferred bacterium will 'want' the same host qualities as the host genes and will consequently be benign. The bacterial genes stand to gain from the host's being good at parental care, good at courting a mate, good at hunting, good at surviving. Eventually, they become symbionts so intimate that we overlook their separate origin. This is what happened with mitochondria. We can view all the genes in the gene pool of a sexually reproducing species as symbiotic with each other, in the same sense as mitochondria are symbiotic. All the genes in the lion gene pool, including those of mitochondria and of other gametically transmitted bacteria, are cooperating symbionts, because their normal pattern of existence is to share a long succession of vehicles from which they have only one exit route—a shared exit route—the gametes of the vehicle.

Phenotypic effects happen to be bundled together in large, coherently integrated vehicles. On other planets it may be different. But animals and plants, at least, have bundling as one of their most salient features. An organism is a colossal enterprise in replicator cooperation: a gigantic lumbering robot as I have called it. We cannot ignore the organism, with its sense organs and limbs, its brain and its coordinated, unitary, purposeful behaviour. It is the primary vehicle, the vehicle that we first notice. The organismal vehicle is so prominent that most evolutionary theorists concentrate on it and forget that it didn't have to exist. One could image a version of Darwinian natural selection in which there are no vehicles, only replicators, but not the other way around. When life first began, there were presumably no vehicles, only replicators. But now the individual organism thrusts itself forward as so prominent and undeniable a vehicle that many biologists formulate their theories in vehicle language. Instead of talking about replicator survival, they talk about individual 'fitness'. Hamilton realised that this strictly has to be refined to inclusive fitness, which may be informally defined (with Bill Hamilton's informal approval) as "that quantity which an organism will appear to be maximising when what is really being maximised is gene survival".

The fact that is possible for biological researchers to work with 'fitness' is testimony to the coherence with which, as it happens, replicators on this planet bundled themselves into vehicles called organisms. When field biologists look at individual lions chasing individual antelopes, or meerkats warning each other of predators, they can make sense of what is going on by treating each individual organism as a fitness-maximising machine: an agent striving to maximise the survival of the replicators that it contains. But we should never forget that this is only a convenient (in my view not always very convenient) approximation to what is really going on, which is that replicators (genes in the Williams sense) are striving to maximise their own survival. 'Striving' should not be taken in the personal sense, of course. Rather, those replicators that survive in the world are the ones that have, in the past, succeeded in maximising their survival.

Not all phenotypic effects are bundled into individual bodies. There are some extended phenotypic effects in which the cascade of genetic action reaches outside the body. A beaver dam is not a part of a beaver's body like its tail, but it is clearly an adaptive phenotype, which has entered into natural selection in the same kind of way as a tail. 'Good' tails help beavers to survive by speeding their swimming. They thereby propagate genes 'for' making good tails. 'Good' dams help beavers to survive by providing a lake for their safety, thereby helping to propagate genes 'for' making good dams. The dam is a phenotypic effect of genes, just as a tail is. Both phenotypes are the product of a long cascade of causes and effects. In the case of the tail, the cascade stops at the body wall. In the case of the dam, the cascade reaches out beyond the body wall—via genetic influences on building behaviour—to an inanimate object built of logs, sticks and mud. The key point is that differences in dams reflect differences in genes (this must be so if the dam is a Darwinian adaptation that has evolved by natural selection): the vital covariance that defines a phenotypic effect for our Darwinian purposes. Nobody, as far as I know, has studied the genetics or the evolution of beaver dams, but they could certainly be studied in the same way as with beaver tails or beaver behaviour. Indeed, dams are 'frozen behaviour'.

A caddis house, which is made by one individual, is straightforwardly a part of the vehicle, its outer casing. The same cooperative of genes that built the caddis legs and eyes also built the house. The only difference is that the cascade of effects reaches out beyond the body, via the medium of (genetically influenced) building behaviour. The beaver dam is complicated by the fact that more than one individual beaver built it. The consortium of genes that built the dam is larger than the consortium that built any one tail.

A termite mound is an even more complicated case of a shared phenotype, shared by genes in tens of thousands of individual termites. These individuals, as it happens, are sterile, but no matter. Copies of them are passed on via queens and kings, whose replicating genes are protected by the mound, with all its many chambers and air-conditioning conduits. The principle is the same. Genes cooperate with each other to build shared phenotypes: queens, kings, workers and mounds.

Now, in this vision of life, which I regard as definitive, where can we find room for 'group selection'?

Is a group a replicator? No. We do not have a 'group pool', a metapopulation in which some groups are more successful than others at making replicas of themselves, replicas that persist through geological time.

OK then, is a group a vehicle? This is the only hope for group selection, and I think it is a pretty forlorn one. In order to qualify as a vehicle, all the genes in a group would need to share the same exit route to the next group in the generational sequence. There would have to be something equivalent to 'group gametes' or 'group propagules', and all the genes in the group would have to share the same expectation of leaving the present group via the same propagule.

Let's go straight to what might seem to be the most promising example, a beehive. I suppose that, at a stretch, you could argue that the hive has some of the properties of a coherently integrated vehicle. Not surprisingly therefore, it has something that you could identify as a 'propagule', namely a swarm, which leaves the old hive and flies off  to found a new one. One hive can throw off half a dozen swarms in a good summer, so there is something sort of equivalent to group reproduction (not replication, reproduction—what individual organisms do). The old metaphor of the 'superorganism' has some applicability to a beehive, and the worker bees are partially comparable to the cells of an individual lion.

But is it helpful to call it group selection? Not for my money it isn't. It 's no more helpful than to say that a lion is a product of 'group selection' because a lion is a 'group' of cells. The comb and even the swarm might be regarded as an extended phenotype of bee genes. As ever, the right way to handle the evolution of beehives is to think about genes surviving by virtue of their phenotypic effects on the world. Successful genes pass to the future via queens and drones; and the workers, who constitute or construct the greater part of the phenotypic apparatus of the hive, do so under the influence of copies of those reproductive genes. But why is it helpful to talk of a group phenotype? It isn't. The agents of the hive, the agents who build the comb, the agents who fly out in search of food and in search of new hollow trees for swarms to inhabit, are still individual organisms, worker bodies whose 'vehicular' credentials cry out for recognition. Workers have eyes and legs, jaws and antennae, instincts and stings, which work together within each organism. All those vehicular phenotypes are best seen as Hamilton saw them: phenotypic expressions of genes working for copies of themselves in the reproductive elements of the hive, the queens and drones. The gene's-eye-view—of replicators and extended phenotypes—can handle everything about a social insect colony with the same elegance as it handles all other animals.

'Group selection', even in the rare cases where it is not actually wrong, is a cumbersome, time-wasting, distracting impediment to what would otherwise be a clear and straightforward understanding of what is going on in natural selection.

_____

_{Dawkins, R (1982) The Extended Phenotype.  Oxford University Press.}

_{Williams, G.C. (1966) Adaptation and Natural Selection. Princeton University Press.}

_{Williams, G.C. (1992) Natural Selection: domains, levels and challenges.}

_{[1] Williams, 1992.}

_{[2] Ibid.}

Professor of Psychiatry, UM Medical School; Professor of Psychology, University of Michigan; Co-author, Why We Get Sick

Cooperation Has Many Explanations, But Social Selection Has Been Neglected

In 1994 I published an essay titled "Why is group selection such a problem?" After nearly two decades, little has changed. Essays like Steven Pinker's attempt to set things straight, but despite many such clear arguments, the debate goes on. This is very curious. Why hasn't this been resolved?

I think we need to turn to social psychology for some of the explanation. Human minds tend to emphasize one cause. So, we find really smart people arguing as if kin selection is unimportant just because there are other routes to selection. Others just advocate for the importance of one cause in preference to others. Also, moral issues grab our minds and we engage them forcefully as if our reputations were on the line. It is easy to see selfish gene proponents as selfish, and group selection advocates as generous, even though if group selection explains anything about human behavior, tendencies to genocide are near the top of the list. These social factors will continue to keep the problem of the evolution of cooperation problematic. But progress is possible.

I have been preoccupied with the problem of the evolutionary origins of morality and other profoundly prosocial tendencies partly because I am the grandson of missionaries, but also because in my psychiatric practice I keep hearing from people who lie awake nights wondering if they might have, perhaps, slightly offended someone. And I hear from those wracked by guilt for abandoning a sick partner, or for a single episode of infidelity. Such phenomena need explanation. I doubt that group selection has much to offer for such traits. Sure it is possible, there just are not examples of traits shaped by group selection that result in individuals sacrificing fitness for the sake of group fitness. Nonetheless, morality and guilt and prosocial traits exist. How is this possible?

I spent years working on commitment theory, finally concluding that it might contribute but was not of major importance. Then I finally went back to Mary Jane West-Eberhard's seminal work on social selection and found what I was looking for. She notes that just as a trait that makes one a preferred mate can be elaborated by sexual selection, a trait that makes one a preferred social partner can also give big fitness advantages. Indeed, we humans seem to care enormously about what others think about it. Yes, these can be interpreted as reputation effects in an indirect reciprocity model, but stepping back from identification with the actor gives a new perspective. The selection forces are exerted by the choices of others as they decide who to be friend with and who to let into their group. They want others who have resources and who will share them. Generously. So, most humans are very careful to not only appear generous and honest, but to actually be remarkably generous and honest. I have argued this may result in runaway selection that shapes quite extreme and expensive traits, but I am not so sure it is true runaway like sexual selection. It may be more like the costs companies pay in advertising and generous return policies to get and keep customers.

There are many recent publications on social selection. The more nasty side of social selection is emphasized well by Christopher Boehm in his new book Moral Origins. Wade and others provide the mathematical underpinnings. A review is in this month's Proceedings of the Royal Society B by Bruce E. Lyon and Robert Montgomerie," Sexual selection is a form of social selection."

I don't think social selection is the one explanation for prosociality; lots of selection forces together shape social tendencies. Also, once social selection gets things going, cultural group selection can modify and elaborate in conjunction with more social selection. But social selection is distinct from other explanations and neglected. It explains much of what group selection is trying to explain. Perhaps it can help to resolve this controversy.

Nesse RM. Social selection and the origins of culture. In: Schaller M, Heine SJ, Norenzayan A, Yamagishi T, Kameda T, editors. Evolution, culture, and the human mind. Philadelphia, PA: Psychology Press. p. 137-50, 2010.

Nesse RM: Runaway Social Selection for Displays of Partner Value and Altruism,Biological Theory 2 (2): 143-155, 2007.

Anthropologist and Professor of Psychology and Economics. Canada Research Chair in Culture, Cognition and Coevolution. University of British Columbia.

Too Late: Models of Cultural Evolution and Group Selection Have Already Proved Useful

Apprehending the place of "group selection" in evolutionary thinking requires understanding the use of formal mathematical models. Many scientific disciplines, ranging from engineering to ecology, develop mathematical models to study, analyze and understand complex dynamic processes. This is an important component of research in most scientific disciplines because—unaided and untutored—our minds are ill-equipped to think clearly about such processes. For many purposes, these models act as mental prostheses that make us rigorously formulate our ideas, help us understand key features of even simple systems, and develop predictions and insights, often including non-intuitive predictions one would not have made before studying the model. Building a useful model hinges on the specific modeling approach, assumptions and choices made during construction.

What Steven Pinker wants banished from our science is a modeling tool that has proved useful for breaking down and analyzing different components of a selective process. Natural selection is, at the same time, blindingly simple and incredibly subtle. Studying genetic evolutionary processes formally often involves specifying and integrating several different contributors to understand the total effect of natural selection, not to mention a whole host of other evolutionary forces such as drift and mutation. When the situation under investigation involves something like group extinctions, due to warfare or environmental shocks, or biased migration due to economic success, multi-level selection accounting can help isolate and analyze the impacts of different components of selection.

As with other kinds of dynamic processes, models of genetic evolution can be built many different ways. The issue of "group selection" revolves around the choice of an accounting system: how does one wish to track fitness or changes in gene frequencies. In many cases (though not all), the exact same process can be represented and developed using quite different evolutionary accounting systems [1-3]. These accounting systems include (1) individual fitness, (2) inclusive fitness, and (3) multi-level or "group" selection.

This last accounting system involves first tallying up all the effects of natural selection within groups on genes ("within group selection"), and then putting that together with the relative contributions of each group to the overall gene pool ("between-group selection"). In such an accounting system, we sometimes find that the average effect of everything going on within groups is opposed by the differences in the relative contributions of groups with different compositions. Or, in a situation that Pinker skips, the net effect of what's going on within groups is zero (a "stable equilibria"), and all the action comes from the relative contribution of each group to the evolutionary change. In verbal descriptions, when the between-group component of natural selection influences the evolutionary process, this is "group selection". This is what "group selection" has meant since 1972 [4].

It's certainly true that often one can figure out how to use any of these three accounting systems to solve a simple problem, and they give the same answer (about equilibrium states). However, it is NOT true that all three are equally easy to apply to any given problem. It is also not true that all methods generate the same kinds of insights or understandings about the evolutionary dynamics or equilibrium states. Which accounting system is best entirely depends on the problem and the assumptions one is willing to make in obtaining an answer.

A useful analogy might be the problem that an aerospace engineer faces when trying to model the trajectory of a satellite. A critical first step in solving such a problem is to select a coordinate system and a place to anchor that coordinate system in space (the origin). Among others, one can pick a spherical coordinate system (two angles and a distance) and anchor it to, say, the center of the earth; or, one can pick a Cartesian coordinate system (x, y, z orthogonal dimensions) and anchor it to a passing meteor. It is completely possible to calculate the orbit of a satellite with any number of different coordinate systems including these two, but picking the first system will allow you to easily solve the problem (analytically, using some solid assumptions) while building your intuitions about the movements of earth's satellites. The second approach will be really hard, and provide you with no new intuitions. So, these are "equivalent" in some sense, but they are not equally useful for any particular problem. And, so it is with evolutionary accounting systems.

Pinker writes, "There's no need to complicate the theory of natural selection with a new 'level of selection'…" This makes little sense. Why would we remove an analytical tool from our toolbox? Following from his argument, Pinker presumably doesn't think we should study and label "sexual selection," since it's really just another component of natural selection. Why break natural selection down into the sexual and non-sexual components? The answer is, of course, because breaking down more complicated processes into components has proved analytically useful, and driven much empirical work. Below, I will show that explicitly considering and modeling inter-group processes has done the same.

Rejecting group selection models is like banning spherical coordinates because you prefer to do your verbal reasoning in Cartesian coordinates.

Now, you'll notice I've not mentioned "replicators" or "vehicles." There are several reasons for this, but I'll mention just one here that proceeds from the above argument. Theorists working at the forefront of improving evolutionary theory neither routinely use, nor need, these metaphors. In fact, many genetic modeling approaches do not even assume discrete traits (replicators), such as adaptive dynamics or quantitative genetics. Quantitative genetic models, which track continuous phenotypes (not genes), are some of the most predictive models we have, and consequently are used in practical applications (e.g., animal breeding). Since its phenotypes, and not genes, that are the true target of selection, then for some purposes it makes good sense to track phenotype. Just like group selection models, these approaches break the total effect of natural selection down into parts linked to different variance components. Steve would presumably argue that—in the interest of theoretical purity—we should toss "quantitative genetic" models onto the junk heap with "group selection" models because such models unnecessarily complicate matters by ignoring the digital nature of DNA code.

Cultural Evolution

This brings us to cultural evolution, which Pinker dismisses as a bad metaphor. Let's first consider what cultural evolutionists actually do, keeping in mind my points about the purposes of mathematical modeling. From its beginning, cultural evolutionists have sought to formalize human learning processes, and then ask what happens in the long-run if we have a population of human learners who are interacting with each other. The psychological ingredients of these learning models come from (1) theoretical genetic evolutionary models focused on the kinds of psychological processes that will be favored by natural selection under different conditions, and (2) empirical evidence on how people actually learn. By combining psychology with social interaction, these models are designed to improve our understanding of otherwise complex historical processes.

To be clear, these models do NOT require any assumptions about replicators, discrete traits, longevity, fidelity or fecundity. In fact, going back to Boyd and Richerson's Culture and the Evolutionary Process, we find that 19 of the 38 different models presented involve continuous traits with arbitrary amounts of error—nothing ever replicates exactly in these models [5]. Much work since then has underlined this point [6, 7]. Now, for a gene-like system, fidelity, fecundity and longevity are required. But, it turns out that there are other ways to turn the trick of creating a stable inheritance system without those characteristics. Perhaps Pinker disagrees with this work, but he does not appear to be aware of it.

To underline this point, consider Pinker's suggestion that if somehow the variation on which a selective process acts is non-random, it is not an "evolutionary" process. It turns out that some of the earliest cultural evolutionary models included precisely this, individual (insight-driven) learning with an arbitrary amount of random noise along with social learning [5, 8-12]. Rather than just asserting it, these models allow theorists to study how varying amounts of random variation vs. individual insights influence the cultural evolutionary process. Furthermore, such models allow us to consider what happens when our intuitions or judgments from experience tend to be wrong, and when this can be overcome by our instinct to just copy more successful people [6, 13]. Supporting this, empirical work suggests that humans possess complex cultural repertoires that exist because of our tendency to learn from more successful or prestigious people, and despite our individual learning abilities, not because of them [13, 14].

So, in what way is cultural evolution a metaphorical extension of genetic evolution? It's not. It's standard-issue science that involves the construction of a class of simple models as a means to glean insights into complex processes. This is why so many mathematical evolutionary biologists are now building cultural evolutionary and culture-gene coevolutionary models [15-20].

Above I pointed out that the proper use of mathematical models, whether they are aimed at genetic or cultural evolutionary processes, is often to make verbal theorizing more rigorous, generate novel insights, formulate non-intuitive predictions, and drive empirical research. On these counts, formal models of cultural evolution, and gene-culture coevolution more broadly, have been highly successful [21-23]. These models are driving thriving research programs in primatology [24, 25], evolutionary biology [19, 26], animal social learning [23], genetics [27, 28], anthropology [29-31], archeology [32, 33], paleoanthropology [34], religion [35, 36]and psychology [37, 38].

Within a large class of models designed to illuminate various aspects of cultural evolution ranging from technological change to ethnicity [10, 39], a small subset have examined the impact of inter-group competition, as one element in the evolutionary process. Here are three examples that illustrate how the explicit consideration of inter-group competition and interaction has already driven important empirical work.

Example 1: Sam Bowles has used multi-level selection (MLS) to examine how social norms that level fitness differences within groups can create the conditions for the spread of altruism via inter-group competition [40]. To be clear, the model tracks an allele frequency, but the MLS formulation naturally exposes the inter-group dynamics in a convenient and intuitive manner. Bowles's formulation tells you exactly what kind of empirical data you need to test the model: (1) what the minimum ratio of genetic variances needs to be (between vs. within groups), (2) how much fitness differences within groups have to be leveled by social norms, and (3) how much lethal inter-group violence has to (or had to) occur. This model directed Bowles to the key pieces of data he needed to obtain and analyze. Of course, Bowles might be wrong about the importance of inter-group competition, but now we know just how to show it.

Pinker might point out that it should be possible to entirely reformulate Bowles's model using another fitness accounting system. Of course, it may be possible! It's also possible to calculate the position of that earth orbiting satellite using the coordinate system attached to the passing meteor. Good luck.

Example 2: For decades anthropologists have been trying to explain cooperation among foragers. They find evidence for both kinship and reciprocity-based mechanisms, but there still seems to be a fair amount of unexplained cooperation. The problem with much of this work is that it was based on single-group theorizing. When one studies a single population model, one is only studying within-group processes. Working among Hadza hunter-gatherers in Tanzania, Coren Apicella and her collaborators mapped the social networks and willingness to cooperate of people from 17 foraging bands/camps [41]. Unlike their predecessors, these researchers had studied the cultural evolutionary models of cooperation, and understood the potential importance of population structure and networks. Consequently, they calculated the within-band and between-band variation in cooperativeness. According to the models of group-selection from the 1970s, on which Pinker is leaning for his dismissal of group selection, this team found something that should not exist. As Figure 1 shows, the variation between-camps was much larger than the variation within camps. That can't be right? According to those old models, the between-group variation is supposed to be MUCH smaller than the within-group variation. Not larger. This suggests there's something wrong with those old models when applied to humans (well, at least to foragers like the Hadza). This also means that the summary dismissal of the importance of between-group processes (cultural and genetic) in the 1970's caused most evolutionary researchers to focus tightly on only what happens within one-band—because they believed, the between group differences couldn't matter [42].

Example 3: Peter Turchin is combining cultural evolutionary theory and tools from ecology to better model historical processes. His formal models, which include and consider inter-group competition, provide general theories that (1) can be applied to different times and places in history, (2) tell him precisely what kind of data will test his theories, and (3) make predictions, about—for example, when and where empires will emerge [43-46]. Empirical efforts to test these predictions have already borne fruit. These efforts are going far beyond what standard history supplies. Moreover, this body of theory is currently driving the assembly of a vast historical database, which will allow even greater quantitative testing of cultural evolutionary theories. Thus, Pinker's claim that carefully specifying and analyzing the causal processes underlying historical change (which include inter-group competition) and testing these ideas across times and places, "nothing to conventional history," is like saying genetic evolutionary theory adds nothing to butterfly collecting.

The Mismatch Hypothesis

Pinker argues in favor of an explanation for large-scale human cooperation known as the "Big Mistake" [47]or "Mismatch" Hypothesis. This hypothesis has been one of a set of hypotheses about human sociality that have been pursued intensely for many years. While a consistent favorite among those who eschew any role for a dynamic interrelationship between genes and culture, the Mismatch Hypothesis has not fared well in empirical tests against alternative hypotheses [47-54]. Moreover, it emerges from dubious theoretical foundations [52, 55, 56]. The problems with this idea are too numerous to present here, so I will summarize three categories of problems.

The first problem is that the Mismatch Hypothesis does not meet any of the five major challenges of human cooperation [56, 57]. It does not explain why (1) the scale and intensity of human cooperation and sociality have expanded in the last 10,000 years, (2) humans are different from other primates, who also live in kin-groups of foragers with lots of repeated interaction, (3) cooperation and sociality vary so dramatically across modern human societies, right down to the behavior in controlled psychological tasks, (4) cooperation and sociality vary across domains within a society, independent of their costs and benefits, and (5) the same incentive mechanisms related to reputation and punishment also operate on non-cooperative behavior, such as ritual performance or food taboos (this is impossible in reciprocity models).

Second, the Mismatch Hypothesis was tested by performing Ultimatum Games across a diverse set of human societies, as well as with American participants. Unlike the Westerners so commonly used in behavioral experiments, many of the participants from these societies really do live in small, face-to-face groups with limited anonymity. If the prosociality in both fairness and willingness to punish in such games was the result of a "misfiring" reciprocity psychology, as Pinker proposes, then we'd expect either (1) no variation among societies, or (2) variation such that those populations who actually experienced a non-anonymous, face-to-face, life would be more prosocial and more willing to build their reputations by punishing low offers. Instead, what this collaboration of evolutionary psychologists, economists, and evolutionary anthropologists actually found contradicted this prediction. People from societies with larger populations, more market integration and more anonymous roles were more prosocial and ready to punish unfairness, not less. In fact, people from the smallest-scale societies sometimes showed no willingness to punish. These results have been replicated and extended, both with a new sample and to include two additional experiments [48, 49, 58, 59].

This large body of ethnographic and experimental work stands in stark contrast to the empirical work Pinker offers to support his view, being limited to American undergraduates who are known not only to be psychologically unusual in many important dimensions, but also to be particularly unusual in behavioral games, in stark contrast to the several foraging groups we tested [48, 49, 59]. Theories of human nature cannot seriously be tested using only WEIRD people [60].

Finally, reciprocity theory does not say what Pinker thinks it says. For example, the finding that successful strategies are generally "nice," meaning they always cooperate on round 1, was shown to be an artifact of the standard formulation of reciprocity models [55]. If, instead, individuals can only maintain a limited number of relationships, the result disappears.

Pinker is also under the mistaken impression that "reputation" is a solution to cooperative dilemmas and an alternative to cultural group selection. Models that examine how reputational systems might solve the problem of large-scale cooperation show that, in fact reputation, can sustain a wide range of behaviors, which may or may not be cooperative [56, 61]. This fits with the fact that human societies vary immensely in what goes into a "good reputation." Honorable or required behaviors in one place can be horrible in another. Moreover, reputations specify how much and when one should cooperate, but this varies from place to place. For example, tipping at least 10% at a restaurant is nearly obligatory in the U.S., while it would be considered strange in Australia and Japan (no tipping). The ability of reputation to sustain almost anything creates what economists have long recognized as an "equilibrium selection problem." Thus, even if we agree that much human cooperation is in response to reputational concerns, this merely opens up the question of why people, and populations, vary so much regarding what goes into a reputation and how it is weighted. Cultural evolutionary hypotheses, which include inter-group competition, provide theories regarding how reputational systems might evolve (culturally) in ways that create group benefits [54]. Cultural group selection and reputation systems are in fact complementary in addressing the problem of cooperation, not alternatives.

The False Allure of One-Group Thinking

Pinker may be right that for non-specialists an evolutionary approach that includes "group selection" may have an allure that arises from its emotional impact. However, there is also a false allure and intuitive appeal that comes from the "individual as strategist in a single group" heuristic used by many evolutionary psychologists (made explicit by Daly and Wilson [62]). Thinking only about one group means you are considering only one component of natural selection, the within group component. To see how this thinking can steer one wrong, consider this situation: an organism faces a single binary choice once in her lifetime, she can either (a) consume half of a valuable resource herself (eating her fill) and allow a randomly selected member of her social group (size N) to consume the other half, or (b) consume half the resource and leave the rest to rot. In a world with only one social group, natural selection will favor only picking choice b (half the resource gets wasted every time). This occurs because an animal that picks choice a will, on-average, benefit all others in his group, leaving herself at a relative fitness disadvantage to those who pick b (and only benefit themselves). This favored strategy is pretty selfish, since even the classical selfish rational actor would be merely indifferent between a and b, perhaps picking a or b at random (this is because she gets the same payoff herself either way). So, a group of selfish agents will be better off than a group of agents built by within-group natural selection. Now, if you create a world with multiple social groups, natural selection will often favor some picking of choice a, and everyone will do better. This can be true even when all the groups are randomly remixed every generation such that genealogical relatedness is irrelevant.

The consequence of this overemphasis on single-groups is that zero-sum thinking has come to pervade much of the verbal reasoning in evolutionary psychology. You can see this in the way Pinker prefers hypotheses based on manipulation, even suggesting that kinship systems and forms of social organization are primarily products of individuals' efforts at a grand deception or conspiracy. Single-group thinking also causes researchers to theorize in terms of strategically navigating a particular reputational system, rather than stepping back and asking why different groups have different reputational systems with different aggregate consequences. Deception and manipulation are obviously part of the human condition, but there is much more to the story, including inter-group competition and much cultural evolution. I recommend that Pinker and those he cites try modeling their ideas. I predict that they will learn that deception and manipulation, as evolutionary strategies, have serious limitations (they are frequency dependent), especially in a world with multiple social groups, since they will cause natural selection to favor avoidance, group dissolution, and social disengagement.
 

References

1.      McElreath, R. and R. Boyd, Modeling the Evolution of Social Behavior2007, Princeton Princeton University Press.

2.      Doebeli, M. and C. Hauert, Limits of Hamilton's rule. Journal of Evolutionary Biology, 2006. 19(5): p. 1386-1388.

3.      van Veelen, M., et al., Group selection and inclusive fitness are not equivalent; the Price equation vs. models and statistics. Journal of Theoretical Biology, 2012. 299: p. 64-80.

4.      Price, G., Extensions of Covariance Selection Mathematics. Annals of Human Genetics, 1972. 35: p. 485-490.

5.      Boyd, R. and P.J. Richerson, Culture and the Evolutionary Process1985, Chicago, IL: University of Chicago Press. 331.

6.      Henrich, J., R. Boyd, and P.J. Richerson, Five misunderstandings about cultural evolution. Human Nature-an Interdisciplinary Biosocial Perspective, 2008. 19(2): p. 119-137.

7.      Henrich, J. and R. Boyd, On Modeling Cultural Evolution: Why replicators are not necessary for cultural evolution. Journal of Cognition and Culture, 2002. 2(2): p. 87-112.

8.      Boyd, R. and P.J. Richerson, An evolutionary model of social learning: the effects of spatial and temporal variation, in Social Learning: Psychological and Biological Perspectives, T.R. Zentall and B.G. Galef, Editors. 1988, Lawrence Erlbaum Associates: Hillsdale, NJ. p. 29-48.

9.      Henrich, J. and R. Boyd, The evolution of conformist transmission and the emergence of between-group differences. Evolution and Human Behavior, 1998. 19: p. 215-242.

10.    Henrich, J., Demography and Cultural Evolution: Why adaptive cultural processes produced maladaptive losses in Tasmania. American Antiquity, 2004. 69(2): p. 197-214.

11.    Henrich, J., Cultural transmission and the diffusion of innovations: Adoption dynamics indicate that biased cultural transmission is the predominate force in behavioral change and much of sociocultural evolution. American Anthropologist, 2001. 103: p. 992-1013.

12.    van Schaik, C.P. and G.R. Pradhan, A model for tool-use traditions in primates: implications for the coevolution of culture and cognition. Journal of Human Evolution, 2003. 44: p. 645-664.

13.    Boyd, R., P.J. Richerson, and J. Henrich, The cultural niche: Why social learning is essential for human adaptation. Proceedings of the National Academy of Sciences of the United States of America, 2011. 108: p. 10918-10925.

14.    Henrich, J. and N. Henrich, The Evolution of Cultural Adaptations: Fijian taboos during pregnancy and lactation protect against marine toxins. Proceedings of the Royal Society B-Biological Sciences, 2010. 366: p. 1139-1148.

15.    Wakano, J.Y., K. Aoki, and M.W. Feldman, Evolution of social learning: a mathematical analysis. Theoretical Population Biology, 2004. 66(3): p. 249-258.

16.    Enquist, M., S. Ghirlanda, and K. Eriksson, Modelling the evolution and diversity of cumulative culture. Philosophical Transactions of the Royal Society B: Biological Sciences, 2011. 366(1563): p. 412-423.

17.    Laland, K.N., et al., How copying affects the amount, evenness and persistence of cultural knowledge: insights from the social learning strategies tournament. Philosophical Transactions of the Royal Society B-Biological Sciences, 2011. 366(1567): p. 1118-1128.

18.    Lehmann, L. and M.W. Feldman, The co-evolution of culturally inherited altruistic helping and cultural transmission under random group formation. Theoretical Population Biology, 2008. 73(4): p. 506-516.

19.    Sigmund, K., et al., Social learning promotes institutions for governing the commons. Nature, 2010. 466(7308): p. 861-863.

20.    Enquist, M., et al., One cultural parent makes no culture. Animal Behaviour, 2010. 79(6): p. 1353-1362.

21.    Mesoudi, A., Cultural evolution : how Darwinian theory can explain human culture and synthesize the social sciences2011, Chicago, Ill.: University of Chicago Press ; Bristol : University Presses Marketing [distributor]. 1 v.

22.    Mesoudi, A., A. Whiten, and K.N. Laland, Towards a unified science of cultural evolution. Behavioral and Brain Sciences, 2006. 29(4): p. 329-+.

23.    Rendell, L., et al., Cognitive culture: theoretical and empirical insights into social learning strategies. Trends in Cognitive Sciences, 2011. 15(2): p. 68-76.

24.    Whiten, A., V. Horner, and F.B.M. de Waal, Conformity to cultural norms of tool use in chimpanzees. Nature, 2005. 437(7059): p. 737-740.

25.    Dean, L.G., et al., Identification of the Social and Cognitive Processes Underlying Human Cumulative Culture. Science, 2012. 335(6072): p. 1114-1118.

26.    Lehmann, L., M.W. Feldman, and K.R. Foster, Cultural transmission can inhibit the evolution of altruistic helping. American Naturalist, 2008. 172(1): p. 12-24.

27.    Laland, K.N., J. Odling-Smee, and S. Myles, How culture shaped the human genome: bringing genetics and the human sciences together. Nature Reviews Genetics, 2010. 11(2): p. 137-148.

28.    Richerson, P.J., R. Boyd, and J. Henrich, Gene-culture coevolution in the age of genomics. Proceedings of the National Academy of Sciences, 2010. 107(Supplement 2): p. 8985-8992.

29.    Currie, T.E. and R. Mace, Political complexity predicts the spread of ethnolinguistic groups. Proceedings of the National Academy of Sciences, 2009. 106(18): p. 7339-7344.

30.    Tehrani, J.J. and M. Collard, On the relationship between interindividual cultural transmission and population-level cultural diversity: a case study of weaving in Iranian tribal populations. Evolution and Human Behavior, 2009. 30(4): p. 286-300.

31.    Hill, K., M. Barton, and A.M. Hurtado, The Emergence of Human Uniqueness: Characters Underlying Behavioral Modernity. Evolutionary Anthropology, 2009. 18(5): p. 187-200.

32.    Collard, M., S.J. Shennan, and J.J. Tehrani, Branching, blending, and the evolution of cultural similarities and differences among human populations. Evolution and Human Behavior, in press.

33.    Shennan, S., Genes, Memes, and Human History: Darwinian Archaeology and Cultural Evolution2003, London: Thames & Hudson.

34.    Powell, A., S. Shennan, and M.G. Thomas, Late Pleistocene Demography and the Appearance of Modern Human Behavior. Science, 2009. 324(5932): p. 1298-1301.

35.    Norenzayan, A. and A.F. Shariff, The origin and evolution of religious prosociality. Science, 2008. 322(5898): p. 58-62.

36.    McKay, R., et al., Wrath of God: religious primes and punishment. Proceedings of the Royal Society B: Biological Sciences, 2011. 278(1713): p. 1858.

37.    Mesoudi, A., How Cultural Evolutionary Theory Can Inform Social Psychology and Vice Versa. Psychological Review, 2009. 116(4): p. 929-952.

38.    Chudek, M., et al., Prestige-biased cultural learning: bystander's differential attention to potential models influences children's learning. Evolution and Human Behavior, 2012. 33(1): p. 46-56.

39.    McElreath, R., R. Boyd, and P.J. Richerson, Shared norms and the evolution of ethnic markers. Current Anthropology, 2003. 44(1): p. 122-129.

40.    Bowles, S., Group competition, reproductive leveling, and the evolution of human altruism. Science, 2006. 314(5805): p. 1569-1572.

41.    Apicella, C.L., et al., Social networks and cooperation in hunter-gatherers. Nature, 2012. 481(7382): p. 497-U109.

42.    Henrich, J., Hunter-gatherer cooperation. Nature, 2012. 481(7382): p. 449-450.

43.    Turchin, P., A theory for formation of large empires. Journal of Global History, 2009. 4(2): p. 191-217.

44.    Turchin, P., Historical Dynamics: Why States Rise and Fall2003, Princeton: Princeton University Press.

45.    Turchin, P., War and peace and war : the life cycles of imperial nations2005, New York: Pearson Education distributor. viii, 405 p.

46.    Turchin, P., Warfare and the Evolution of Social Complexity: A Multilevel-Selection Approach. Structure and Dynamics, 2010. 4(3).

47.    Boyd, R. and P. Richerson, Solving the Puzzle of Human Cooperation, in Evolution and Culture, S. Levinson, Editor 2002, MIT Press: Cambridge MA.

48.    Henrich, J., et al., Costly Punishment Across Human Societies. Science, 2006. 312: p. 1767-1770.

49.    Henrich, J., et al., Market, religion, community size and the evolution of fairness and punishment. Science, 2010. 327: p. 1480-1484.

50.    Henrich, J. and R. McElreath, Dual Inheritance Theory: The Evolution of Human Cultural Capacities and Cultural Evolution, in Oxford Handbook of Evolutionary Psychology, R. Dunbar and L. Barrett, Editors. 2007, Oxford University Press: Oxford. p. 555-570.

51.    Fehr, E. and F. Schneider, Eyes are on us, but nobody cares: are eye cues relevant for strong reciprocity? Proceedings of the Royal Society B-Biological Sciences, 2010. 277(1686): p. 1315-1323.

52.    Chudek, M., W. Zhao, and J. Henrich, Culture-Gene Coevolution, Large-Scale Cooperation and the Shaping of Human Social Psychology, in Signaling, Commitment, and Emotion, R. Joyce, K. Sterelny, and B. Calcott, Editors. forthcoming, MIT Press: Cambridge.

53.    Fehr, E., U. Fischbacher, and S. Gächter, Strong Reciprocity, Human Cooperation and the Enforcement of Social Norms. Human Nature, 2002. 13: p. 1-25.

54.    Mathew, S. and R. Boyd, Punishment sustains large-scale cooperation in prestate warfare. Proceedings of the National Academy of Sciences of the United States of America, 2011. 108(28): p. 11375-11380.

55.    Hruschka, D. and J. Henrich, Friendship, Cliquishness, and the Emergence of Cooperation. Journal of Theoretical Biology, 2006. 239(1): p. 1-15.

56.    Chudek, M. and J. Henrich, Culture-Gene Coevolution, Norm-Psychology, and the Emergence of Human Prosociality. Trends in Cognitive Sciences, 2010. 15(5): p. 218-226.

57.    Henrich, N. and J. Henrich, Why Humans Cooperate: A Cultural and Evolutionary Explanation 2007, Oxford: Oxford University Press.

58.    Henrich, J., et al., eds. Foundations of Human Sociality: Economic experiments and ethnographic evidence from fifteen small-scale societies. 2004, Oxford University Press: Oxford.

59.    Henrich, J., et al., "Economic man" in cross-cultural perspective: Behavioral experiments in 15 small-scale societies. Behavioral and Brain Sciences, 2005. 28(6): p. 795-855.

60.    Henrich, J., S.J. Heine, and A. Norenzayan, The Weirdest People in the Word? Behavior and Brain Sciences, 2010. 33(2/3): p. 1-23.

61.    Panchanathan, K. and R. Boyd, Indirect reciprocity can stabilize cooperation without the second-order free rider problem. Nature, 2004. 432: p. 499-502.

62.    Daly, M. and M. Wilson, Sex, Evolution and Behavior. 2nd ed1983, Belmont, CA: PWS Publishers.

Lecturer in Psychology & co-director of the Centre for Culture and Evolutionary Psychology, Brunel University

A Distinct Theory of Biological Group Selection Is Unnecessary (Although Multilevel Selection Occurs)

Steven Pinker's essay is a welcome corrective to the renewed popularity of biological group selection models. As he points out, much of this renewal has involved little more than semantic confusion [1, 2], which we must get past in order to address any underlying substantive issues. The key question should be: do biological group selection theories generate any accurate predictions that are not already generated by existing individual-level theories of cooperation? In the case of human cooperation, the answer seems to be "so far, no" [3].

Over the last half century, the most productive theories of human cooperation—in both behavioral biology and general social science—have predicted that people will cooperate in individually-adaptive (that is, inclusive fitness-enhancing) ways. These include the evolutionary theories of kin altruism, direct and indirect reciprocity, partner choice and positive assortment, and costly signaling [4, 5], as well as the influential works of Hardin and Olson on group cooperation [6, 7]. These theories have inspired decades of empirical research, much of it cross-cultural, which have produced the following general picture of human (non-kin) cooperation [3-5]. People cooperate, often in groups, in order to acquire individual-level fitness benefits—resources and social status—that they could not acquire via individual action. They are frequently deeply interested in the success of their group, because membership in a successful group can be the best means for an individual to acquire resources and status; as they cooperate, they are preoccupied with whether they are being sufficiently compensated with these rewards. When relatively high contributors to group success are allocated relatively high rewards, members compete to be the highest contributors. However when rewards are not tied to individual contribution, members are more likely to free ride, especially if they can do so with impunity. If they perceive that they are themselves being free-ridden, they attempt to curtail their own exploitation; for instance, they simply stop cooperating, or they experience punitive sentiment towards free riders in proportion to their own individual vulnerability to being free ridden. When people are able to develop a reputation based on their cooperative behavior, they cooperate more, and when cooperators are given a choice of social partners, they choose one another and ostracize free riders [3-5].

This doesn't exhaust the list of relevant findings, but should suffice to make the point. Many commonly-observed cooperative behaviors are uniquely predicted by individual-level theories, whereas few if any seem uniquely predicted by group-level theories. Some observed economic game behaviors—such as cooperation in one-shot, anonymous games—have been claimed as support for unique group selection predictions [8]. However, these claims have attracted a number of criticisms [9-13], not least because of the highly artificial nature of experimental economic laboratories. It is one thing to induce apparently individually-maladaptive behavior in a lab, but not at all clear that such behavior would be individually-maladaptive under ecologically valid conditions [3].

So the question becomes: if the theory of biological group selection is so necessary, then why does it seem so much less useful than individual-level theories for predicting how people actually behave? To doubt this utility is not to reject multilevel selection theory, which seems vital for explaining the major transitions in the evolution of life [14]. In fact, a multilevel perspective helps clarify why a distinct theory of biological group selection is unnecessary: just as genes create individuals as vehicles for replication, individuals form groups to further their own fitness goals (although this is not a perfect analogy; because the fates of genes and individuals are much more tightly correlated than the fates of individuals and groups, intragenomic conflict is much rarer than cases of within-group conflict like free riding). Groups serve the fitness interests of individuals, so it's not clear how they could transcend these interests. It is also unclear, therefore, why we would expect a biological group selection theory to make any predictions that are not already made by an individual-level theory.

A multilevel selection perspective can also be useful for thinking about human cultural evolution, although again, it's a mistake to see culture as something that transcends individual-level biological (cognitive) adaptations, because culture is ultimately caused by those adaptations. I agree with some other commentators here that it makes sense to speak of cultural group selection, because human cognitive adaptations generate innovative cultural traits that vary across societies, and some such traits could spread by virtue of being advantageous in the context of between-group competition [15]. However, by using the same terms to describe biological and cultural evolutionary processes, we increase the risk of the two processes being confused with one another. Therefore, when speaking of adaptation and group selection, it's essential to specify whether one is referring to the biological or the cultural kind.

One can accept multilevel selection theory, without accepting that there is a need for a biological group selection theory of cooperation that is distinct from the individual-level theory. On the contrary, multilevel selection theory helps clarify why the distinct group selection theory is superfluous: just as individuals serve genes, groups serve individuals.

___

_{1.  West, S. A., Griffin, A. S., & Gardner, A. (2007). Social semantics: altruism, cooperation, mutualism, strong reciprocity and group selection. Journal of Evolutionary Biology, 20, 415-432.}

_{2.  West, S. A., Griffin, A. S., & Gardner, A. (2008). Social semantics: How useful has group selection been? Journal of Evolutionary Biology, 21, 374-385.}

_{3.  Price, M. E. (2012). Group selection theories are now more sophisticated, but are they more predictive? Evolutionary Psychology, 10, 45-49.}

_{4.  Price, M. E. (2011). Cooperation as a classic problem in behavioural biology. In V. Swami (Ed.), Evolutionary Psychology: A Critical Introduction (pp. 73-106). Chichester, West Sussex: BPS Blackwell.}

_{5.  Price, M. E. & Johnson, D. D. P. (2011). The adaptationist theory of cooperation in groups: Evolutionary predictions for organizational cooperation. In G. Saad (Ed.), Evolutionary Psychology in the Business Sciences (pp. 95-134). Berlin: Springer.}

_{6.  Olson, M. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press.}

_{7.  Hardin, G. J. (1968). The tragedy of the commons. Science, 162, 1243-1248.}

_{8.  Bowles, S. & Gintis, H. (2011). A Cooperative Species: Human Reciprocity and its Evolution. Princeton University Press.}

_{9.  Trivers, R. (2004). Mutual benefits at all levels of life. Science, 304, 964-965.}

_{10.                Burnham, T. & Johnson, D. D. P. (2005). The evolutionary and biological logic of human cooperation. Analyse and Kritik, 27, 113-135.}

_{11.                Hagen, E. H., & Hammerstein, P. (2006) Game theory and human evolution: A critique of some recent interpretations of experimental games. Theoretical Population Biology, 69, 339-348.}

_{12.                Price, M. E. (2008). The resurrection of group selection as a theory of human cooperation. Social Justice Research, 21, 228-240.}

_{13.                Delton, A. W., Krasnow, M. M., Tooby, J., & Cosmides, L. (2011). The evolution of direct reciprocity under uncertainty can explain human generosity in one-shot encounters. Proceedings of the National Academy of Sciences USA, 108, 13335-13340.}

_{14.                Maynard Smith, J., & Szathmáry, E. (1995). The Major Transitions in Evolution. Oxford: Oxford University Press.
15.                Alexander, R. D. (1987). The Biology of Moral Systems. Hawthorne, NY: Aldine De Gruyter.}

SUNY distinguished professor of biology and anthropology, Binghamton University; Editor-in-Chief of Evolution: This View of Life

The Central Question of Group Selection

Steven Pinker's essay does not identify the central question that provides continuity for the group selection controversy, from Darwin to the present: How can traits that are "for the good of the group" evolve, when they are selectively disadvantageous within groups? The answer, according to group selection theory, is that groups of individuals who behave for the good of their group outcompete other groups, as surely as group-oriented individuals are outcompeted by more self-oriented individuals within groups.

This is the question and potential answer that Darwin posed in words and population geneticists such as Wright, Haldane, Fisher, and Maynard Smith posed in equations. In the 1960's, a consensus emerged that appeared to establish two points:

1) Selection among groups in a multi-group population is almost invariably weak, compared to selection among individuals within the groups.

2) Other mechanisms can explain the evolution of prosocial behaviors without invoking group selection.

Today we know that both of these conclusions were mistaken. Traits can evolve by between-group selection, despite being selectively disadvantageous within groups. And all evolutionary theories of social behavior that were developed as alternatives to group selection include the logic of multilevel selection within their own frameworks. In other words, they all assume that social behaviors are expressed in groups that are small compared to the total population, the behaviors variously labeled "altruistic" and "cooperative" are typically selectively disadvantageous within the groups and evolve only by virtue of the differential contribution of the groups to the total population. When it comes to answering the central question originally posed by Darwin, there is no alternative to group selection.

Pinker's essay is centered on a different question: Can a concept of group selection be formulated that is comparable to the concept of gene selection? For Pinker, this is the only way that the term "group selection" makes sense, but it is not the concept of group selection that began with Darwin, was refined by population geneticists, rejected in the 1960's, and revived today.

To understand why Pinker would pose such a question, it is necessary to describe the concept of gene selection. G.C. Williams wrote Adaptation and Natural Selection [1] in part to explain population genetics theory to a wider audience of biologists. His emphasis on genes as "replicators" (as they came to be known) is based on the concept of average effects, which is calculated by averaging the fitness of alleles across all individual genotypes and all social and environmental contexts. This "all things considered" calculation predicts which allele will evolve in the total gene pool, which is why Williams called it a "bookkeeping method". It is also useful for estimating the heritability of a phenotypic trait at the individual level.

To see how the concept of average effects figures in a group selection model, consider two alleles at a single locus that code for an altruistic and selfish trait, respectively. The population is subdivided into multiple groups. The gene for selfishness is selectively advantageous within groups, but groups with a higher frequency of the gene for altruism contribute more to the total gene pool. What evolves in the total gene pool depends upon the relative strength of within- vs. between-group selection. If within-group selection prevails, then the selfish allele has the highest average effect and evolves in the total gene pool. If between-group selection prevails, then the altruistic allele has the highest average effect and evolves in the total gene pool. That's what it means to call average effects a "bookkeeping method" that accounts for whatever evolves in the total population.

Thanks first to Williams [2] and then to Richard Dawkins [3], the concept of average effects became popularized as "the gene's eye view of evolution" and the gene was called "the fundamental unit of selection". This is what Pinker calls "gene selection". It is a useful perspective, as long as the right inferences are drawn from it. In particular, all population genetics models are "gene selection" models when they assume that phenotypic traits are coded by genes. That includes all genetic group selection models. When Pinker asks whether groups can be like genes and group selection can be like gene selection, he is comparing apples and oranges. Groups were never expected to be like genes in group selection models.

Following the rejection of group selection, it was common for authors to treat the concept of genes as replicators as an argument against group selection all by itself. This is wrong, and very few informed evolutionists would make the mistake today. It is difficult for me to tell whether Pinker is committing this error or whether he has something more nuanced in mind.

Pinker also comments on cultural evolution in ways that have little to do with group selection (e.g., his remarks on Lamarckism, human decision-making, and what evolutionary theory adds to the study of human history). In my opinion, Pinker needs to absorb the fact that the process of natural selection requires heredity and that genes are a particular mechanism of heredity. Other mechanisms also create a resemblance between parents and offspring. Some of these mechanisms are psychological and cultural [4].

David Queller, often cited as a contemporary critic of group selection, helped to coordinate the much discussed reply by 137 co-authors [5] to the article by Nowak, Tarnita and Wilson on group selection published in Nature. [6]. But Queller, in his response to Pinker's essay, notes that the days of pitting multilevel selection against inclusive fitness theory are over. Pinker and Nowak et al. are both wrong to argue for one to the exclusion of the other. Queller compares the two theoretical frameworks to two languages, such as English and Russian, and appreciates that both languages can be useful. The challenge is to become bilingual and to identify the important issues that can now be addressed within either framework. In general, it is a mistake to confuse the issues at stake in current discussions of group selection based on equivalence with the issues at stake during the 1960's.

The commentaries following Pinker’s essay on cultural evolution are extraordinarily diverse. If we crudely divide the conceptual space into group selection (pro or con) and cultural evolution (pro or con), all four combinations are represented. One way to achieve clarity is by reminding ourselves of the basic question originally posed by Darwin: If a phenotypic trait that benefits whole groups does not maximize relative fitness within groups, how can it be maintained in the total population? Because this question is posed at the phenotypic level, it is agnostic about the mechanism of inheritance. The most explicit models of cultural evolution show that group selection is an exceptionally strong force in human cultural evolution, even to the point of qualifying as a major evolutionary transition [7-9]

___

Abbott, P., Abe, J., Alcock, J., & al., et. (2010). Inclusive fitness theory and eusociality. Nature, 471, E1-E4. doi:doi:10.1038/nature09831

Dawkins, R. (1976). The Selfish gene. Oxford: Oxford University Press.

Jablonka, E., & Lamb, M. (2006). Evolution in Four Dimensions: Genetic, Epigenetic, Behavioral, and Symbolic Variation in the History of LIfe. Cambridge, MA: MIT Press.

Maynard Smith, J., & Szathmary, E. (1995). The major transitions of evolution. New York: W.H. Freeman.

Nowak, M. A., Tarnita, C. E., & Wilson, E. O. (2010). The Evolution of Eusociality. Nature, 466, 1057-1062.

Turchin, P. (2012). Multilevel selection is a productive framework for investigating human history. Evolution: This View Of Life  http://www.thisviewoflife.com/index.php/magazine/articles/multilevel-selection-is-a-productive-theoretical-framework-for-investigatin

Williams, G. C. (1966). Adaptation and Natural Selection: a critique of some current evolutionary thought. Princeton: Princeton University Press.

[1] Williams, G. C. (1966)

[2] Williams (1966)  

[3] Richard Dawkins (1976)

[4] Jablonka and Lamb (20060

[5] Abbott et al. (2010)

[6] Nowak, Tarnita and Wilson (2010)

[7] Maynard Smith, J., & Szathmary, E. (1995).

[8] Richerson, P. (2012). Commentary on Steven Pinker’s essay.

[9] Turchin, P. (2012), in Evolution: This View Of Life  http://www.thisviewoflife.com/index.php/magazine/articles/multilevel-selection-is-a-productive-theoretical-framework-for-investigatin

Social Psychologist; Professor, New York University Stern School of Business; Author, The Righteous Mind 

To See Group-Selected Traits, Look At Groupishness During Intergroup Competition.

If you want to see fish swimming, go to a coral reef. You will be surrounded by them. Likewise, if you want to see group-selected traits in action, look at groups that are actively competing with other groups.

Here is a much harder way to study group-selected traits in human beings: Bring individuals into the lab to interact with strangers in temporary groups that are not put into competition with other groups. Rather, make your experimental subjects play games in which their monetary interests are directly pitted against the interests of the other group members, even though they would be better off overall if they were to work as a group. Contributions tend to decline rapidly, and in his essay, Steven Pinker concludes from this research that "humans are not a group-selected adaptation which capitalizes on opportunities to make sacrifices for the common good."

Pinker notes that much has been made of the findings, within this paradigm, of "altruistic punishment." People will pay to punish cheaters—they bear a cost to help the group. He then asks if this one single finding should be "sufficient reason to revise the modern theory of evolution," and he offers alternative explanations for altruistic punishment based on individual-level selection. I grant that the research on commons dilemmas is open to multiple interpretations. But I strongly disagree that self-sacrifice in commons dilemmas is "the only empirical phenomenon which has been directly adduced as support for group selection."

Now, let us go to the coral reef. Let us flip the intergroup competition switch.

One of the few social psychological studies that actually put real, ongoing groups into real and protracted conflict was the famous "summer camp" study carried out by Muzafar Sherif [1], who brought two groups of twelve-year-old boys to a summer camp in a state park in Oklahoma in 1954. At first, the two groups did not even know of each others' existence, yet even so, each group started marking territory and creating a tribal identify for itself. Both groups engaged in some mild tribal behaviors that would be useful if the group were to encounter a rival group that claimed the same territory. That is what happened on day 6 when the "Rattlers" discovered that the "Eagles" were playing baseball on what the Rattlers took to be "their" ball-field. The Rattlers then challenged the Eagles to a game, which initiated a weeklong series of competitions that Sherif had planned from the start.

Once the competition began, it was as though a switch was flipped in each boy's head. As Sherif described it: "performance in all activities which might now become competitive (tent pitching, baseball, etc.) was entered into with more zest and also with more efficiency." Tribal behaviors increased dramatically. Both sides created flags and hung them in contested territories. They raided each others' bunks, called each other names, and even made weapons (socks filled with rocks.)

Were these acts altruistic? Technically yes, because each tribal behavior had some cost for the individual, and it benefitted the group's cohesiveness or effectiveness. But I think the opposite of selfishness in evolutionary terms should not always be altruism. For the purposes of the present debate, things get clearer if we contrast selfishness with groupishness. The hand of group-level selection is most vividly seen when we look at behaviors that impose some cost on the individual, but that do not transfer that cost as a benefit to one or several specific other group member (which would help the selfish individualists prosper in a multi-level analysis). Rather, mental mechanisms that encourage individuals to do things that help their team succeed, despite some cost to the self, are the most likely candidates for having come down to us by a path in which group-selection played a part.

I mean the sorts of things we do more of when our group or team is attacked—show the flag or the team colors, rally around the leader, join with others (at some cost to the self) to punish free riders and kill or expel traitors, and just generally becoming more of a "team player," as Sherif's campers did. Another vivid example: Gneezy and Fessler [2] were running studies of altruistic punishment among separate groups of Israelis and Palestinians back in 2006. Then the war between Israel and Hezbollah broke out, and suddenly people in both communities began to punish free-riders more harshly, and reward cooperators more lavishly. When you flip the intergroup-competition switch, people get more groupish.

I am not saying these things are good in a normative sense. I am just trying to understand some corners of our moral minds that can not readily be explained by kin selection and reciprocal altruism alone. Groups in which genes for groupish psychology co-evolved with cultural innovations for effective groupishness (such as initiation rites and body painting) probably outcompeted groups that lacked either the genes or the cultural innovations to maximize the effectiveness of those genes. (Note: Pinker takes group selection to be the claim that that mental traits can be shaped be a process in which natural selection acts on groups "akin to natural selection acting on genes." Most of us who favor multi-level selection do not think this way; we think that natural selection can sometimes act on groups akin to the way it acts on individuals.)

Pinker does indeed consider intergroup competition in warfare, and he is right to disabuse us of romanticized notions of warriors who willingly die for the group. He is also right that there is no innate tendency to become a suicide bomber—such people must be manipulated into such extreme self-sacrifice with environmental cues and pressures, including "the simulacra of family experiences, myths of common descent," etc. But he follows this point by asserting that "none of this wasteful ritualizing and mythologizing would be necessary if 'the group' were an elementary cognitive intuition which triggered instinctive loyalty." Well, many social psychologists believe that the group is an elementary cognitive intuition. The "minimal group" studies of Henri Tajfel [3] showed this long ago—people will make more positive attributions about strangers assigned to their own group even on the basis of arbitrary criteria, no more meaningful than a coin flip. And more recently, work with infants shows that they note markers of group identity [such as accent] and prefer people who are members of their group [4]. This preference for in-group members is at least as strong [5] as the well-known preference of infants for helpful characters (rather than harmful or "mean" characters [6]).

In sum, most of our social psychology, and even most of our moral psychology, was shaped by individual-level selection. There has always been competition among individuals within groups, competing for status, mates, and the trust of potential partners for cooperation. But if you examine the psychological traits that motivate and enable cohesion, trust, and effective coordination, and if you do this during times of intergroup conflict, you will find many behaviors and mental mechanisms that are much harder to explain using only individual-level mechanisms. You will find yourself swimming among group-selected traits.

___

1. Sherif, M., Harvey, O. J., White, B. J., Hood, W., & Sherif, C. [1961/1954]. Intergroup conflict and cooperation: The Robbers Cave experiment Norman, OK: University of Oklahoma Institute of Group Relations.

2. Gneezy, A., & Fessler, D. M. T. [2011]. Conflict, sticks and carrots: war increases prosocial punishments and rewards. Proceedings of the Royal Society B: Biological Sciences.

3. Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. [1971]. Social categorization and intergroup behaviour. European Journal of Social Psychology Vol 1[2] 1971, 149-178.

4. Kinzler, K. D., Dupoux, E., & Spelke, E. S. [2007]. The native language of social cognition. PNAS Proceedings of the National Academy of Sciences of the United States of America, 104[30], 12577-12580.

5. Hamlin, J. K. [2012]. A Developmental Perspective on the Moral Dyad. Psychological Inquiry, 23, 166-171.

6. Hamlin, J. K., Wynn, K., & Bloom, P. [2007]. Social evaluation by preverbal infants. Nature, 450, 557-560.

Post-doc, Philosophy, Politics and Economics Program, University of Pennsylvania

The Moral Problem Of Group Selection

There is much to agree with in Steven Pinker's essay. In these remarks, I focus on the argument that group selection is at odds with what we know of human moral psychology and argue that indeed individual selection provides a better account moral behavior.

Putting psychology back in the debate is all the more important that the debate has often focused on group-selection evolutionary mechanisms (are they are theoretically sound? Were they at work during human evolution?) but less often on the psychological predictions that go with these mechanisms. Let's grant that group selection is theoretically possible and let's assume, at least for the sake of discussion, that it had an impact on human evolution; what would this really predict about human psychology?

For many group selectionists, the answer is obvious: group selection predicts that humans are moral creatures and since they are indeed moral, it strongly suggests that group selection is a better theory than individual selection. This would all be well and good if individual selection predicted that individuals are, like psychopaths, hiding their selfish motives in order to cooperate with others. But it doesn't. Quite the contrary, individual selection emphasizes that being genuinely moral is the best evolutionary strategy.[1]Today, evolutionary biologists indeed argue that moral behavior evolved to attract partners in a market where individuals could freely choose amongst potential cooperators. In this competitive context, being moral isn't just the right thing to do, it is simply your best option. Start cheating, hiding, or manipulating others, and you will soon find yourself stuck with second-class partners[2].

So the debate is not between a theory that predicts that humans are moral (group selection) and a theory that predicts they are not (individual selection), but between two theories of morality: one based on sacrifice for the group and the other one based on individual interests. This debate is not novel and echoes a fundamental question in moral philosophy: Is morality about maximizing the welfare of the community (biologists would rather talk about the fitness of the group) or about respecting individual rights (biologists would talk about individual fitness)? In other words, is morality utilitarianist or contractualist?

At first sight, utilitarianism is the winner: morality is all about helping others and being generous, right? But philosophers have highlighted that, again and again, human morality actually isn't about maximizing welfare. Here is a famous case in point put forward by moral philosopher Judith Thomson:

"A brilliant transplant surgeon has five patients, each in need of a different organ, each of whom will die without that organ. Unfortunately, there are no organs available to perform any of these five transplant operations. A healthy young traveler, just passing through the city the doctor works in, comes in for a routine checkup. In the course of doing the checkup, the doctor discovers that his organs are compatible with all five of his dying patients. Suppose further that if the young man were to disappear, no one would suspect the doctor." [3]

What would you do? No doubt ants and bees would agree to kill the young individual in the interest of the group. But we, humans, disagree. As John Rawls noted in the opening page of his famous Theory of Justice:

"Each person possesses an inviolability founded on justice that even the welfare of the society as a whole cannot override. For this reason justice denies that (…) the sacrifices imposed on a few are outweighed by the larger sum of advantages enjoyed by many".[4]

Of course, one can dismiss these intuitive remarks and relegate them to philosophical quibbles. In the last two decades though, empirical investigations of moral judgments have exploded and, properly interpreted, they have all lead to the same conclusion: humans are not utilitarian and morality did not evolve for the group.

Now, let's be a bit more technical, and consider two examples where utilitarianism and contractualism disagree: punishing wrongdoers and distributing resources.

1. Punishing wrongdoers

According to group selection, punishment evolved as a way to curb selfishness or, as utilitarian philosophers would put it, to deter futures crimes. By contrast, individual selection predicts that punishment is just a way to compensate the victim or, as contractualist philosophers would put it, to restore fairness. There are many ways to disentangle these two views of punishment. For instance, if punishment is really a second-order utilitarian trait, as cultural group selection scholars argue, sanctions should always be harsh enough to make it more advantageous to cooperate than to defect: if a crime is difficult to detect, the punishment for that crime ought to be more severe in order to counterbalance the temptation created by the low risk of getting caught.

Experimental studies relying on a variety of methodologies, however, have revealed that when people punish harmdoers, they generally ignore factors related to deterrence (likelihood of detection, publicity of the punishment, likelihood of repeat offending) and instead take into account parameters that are relevant to restore fairness between the criminal and the victim (magnitude of harm, intentionality).[5] Similarly, field observations have extensively demonstrated that the level of compensation in prestate societies is proportional to the harm done to the victim: it aims to force the wrongdoer to compensate the victim rather than to deter future crimes.[6]

Here, death penalty is a case in point. Although many people claim that their opinion about death penalty is based on efficiency (partisans argue it deters crimes, opponents that it has no effect), several studies have shown that, many people would actually continue to support death penalty even if it had no deterrent value.[7]People support death penalty first and foremost because it seems to them that it is the only proportionate penalty for certain crimes (murder, rape, etc.), not because they see it as a useful tool to deter future crime.

2. Distributing resources

When it comes to distributing resources, it seems natural to turn to economic games—experimental situations in which people are asked to allocate real money among different individuals. What are the predictions of the two theories here? In group selection, morality should aim at maximizing the welfare of the group through transfers of resources from individuals who have resources to those who need them. As Steven Pinker notes, this is precisely what we observe among eusocial insects and at first sight, this is what participants do in economic games: they voluntary transfer the money they have to the other participant, even though they could keep it for themselves. But is that so clear? Let's consider two scenarios. In the first scenario, you are in the street and an experimenter comes to you and says "Hi, this is John. He is not a beggar. He is just an average guy like you. Would you like to share the $10 I know you probably have in your wallet?" In the second scenario, the experimenter comes to you and says "Hi, here are $10. Take them. Oh, I forgot to make the presentations: this is John. John is a guy that also happens to be here. Would you like to share the $10 you just got?" In the first scenario, you own the money and you see no reason why you would share it; in the second scenario, you have just received a lump of money for no apparent reason and you do not feel entitled to keep it all to yourself.

The first scenario is close to real life, the second is close to most procedures used in standard 'dictator games'. In standard dictator games, people appear to be generous, but they might simply be distributing according to what appears to be fair given where the money comes from. Experimental manipulations of context in dictator games have indeed confirmed this view: When participants are told that the money belongs to them, they do not give anything; when they are told that it belongs to the other participant, they let her take it; when they work with the other participants to produce the resource, they distribute it according to merit, taking into account personal investment and talent.[8]In other words, how much money is transferred in dictator games does not reflect how much the dictator cares about the recipient but how much the dictator owes her.

Group selection is hailed as best explaining human morality. It does not.

___

_References

_{André, J. B., & Baumard, N. (2012). Social opportunities and the evolution of fairness. Journal of Theoretical Biology.}

_{Baron, J., & Ritov, I. (1993). Intuitions about Penalties and Compensation in the Context of Tort Law. Making Decisions About Liability and Insurance, 7(1), 7-33.}

_{Baumard, N. (2011). Punishment is a not group adaptation Humans punish to restore fairness rather than to help the group. Mind and Society, 10(1).}

_{Baumard, N., André, J. B., & Sperber, D. (in press). A mutualistic approach to morality.}

_{Carlsmith, K., Darley, J., & Robinson, P. (2002). Why Do We Punish? Deterrence and Just Deserts as Motives for Punishment. Journal of Personality and Social Psychology, 83(2), 284–299.}

_{Delton, A. W., Krasnow, M. M., Cosmides, L., & Tooby, J. (2011). Evolution of direct reciprocity under uncertainty can explain human generosity in one-shot encounters. Proceedings of the National Academy of Sciences, 108(32), 13335–13340.}

_{Ellsworth, P. C., & Ross, L. (1983). Public opinion and capital punishment: A close examination of the views of abolitionists and retentionists. Crime & Delinquency, 29(1), 116.}

_{Frank, R. (1988). Passions within reason: The strategic role of the emotions (Vol. 1st). New York: Norton.}

_{Rawls, J. (1971). A theory of justice. Cambridge, Mass.,: Belknap Press of Harvard University Press.}

_{Sperber, D., & Baumard, N. (in press). Moral and reputation in an evolutionary perspective. Mind & Language.}

_{Sunstein, C., Schkade, D., & Kahneman, D. (2000). Do People Want Optimal Deterrence? Journal of Legal Studies, 29(1), 237-253.}

_{Thomson, J. J. (1985). The Trolley Problem. Yale Law Journal, 94(6), 1395-1415.}

_{Trivers, R. (1971). Evolution of Reciprocal Altruism. Quarterly Review of Biology, 46, 35-57.}

_{___}

_{[1] See for instance Delton, Krasnow, Cosmides, & Tooby, 2011; Frank, 1988; Sperber & Baumard, in press; Trivers, 1971.}

_{[2] André & Baumard, 2012.}

_{[3] Thomson, 1985.}

_{[4] Rawls, 1971.}

_{[5] Baron & Ritov, 1993; Carlsmith, Darley, & Robinson, 2002; Sunstein, Schkade, & Kahneman, 2000.}

_{[6] Baumard, 2011.}

_{[7] Ellsworth and Ross 1983.}

_{[8] Baumard, André, & Sperber, in press}

Postdoctoral Scholar, Center for Evolutionary Psychology, University of California, Santa Barbara

Postdoctoral Scholar, Center for Evolutionary Psychology, University of California, Santa Barbara.

From the raiding parties of the Turkana to the gangs of West Side Story, group cooperation—and violent group conflict—interpenetrate human life and fiction. What is the origin of human group behavior? Steve Pinker's essay is an account of the serious problems associated with one explanation: group selection. We would like to amplify one of Pinker's points about the dearth of empirical evidence for group selection, and address several related issues that appear perniciously difficult to communicate.

Most organisms live in a world filled with photons and these photons have bounced off nearby surfaces, including the surfaces of mates, food, friends, foes, or predators. An animal that could use these photons to reconstruct the world around it would be at a marked advantage. The utility of seeing (and physical laws that organize photons etc.) are what evolutionary biologists call a selection pressure—a feature of the world that, if properly exploited, can cause faster biological replication. The eye and the visual system are what evolutionary biologists call an adaptation—an improbably well-organized feature of an organism that actually does exploit a selection pressure and that came into being through natural selection because of that selection pressure. Understanding human nature requires mapping the nature of human psychological adaptations.

What are the adaptations that compose the mind? The eye/visual system is surely one (or an integrated set, if you're a splitter instead of a lumper). One aspect of the present debate—and the focus of Pinker's essay—turns on the question: Are there psychological adaptations that evolved due to group selection? Note that this is an entirely different question than whether group selection is a logically possible selection pressure. To this latter question, decades of modeling have given a definitive yes. But as Pinker argues, and in our view as well, there are no data that demonstrate the existence of group selected adaptations in humans. To do this, one would need evidence of special design—evidence that a feature of an organism was exquisitely and improbably well-crafted for generating benefits for the group as a whole, independent of how it might affect the individual.

This search for evidence of special design is critically lacking in the arguments of proponents of group selection. Often—as in many of the pro-group selection responses to Pinker's essay—facts such as that humans are a social species, prone to group living and social learning, and capable of high levels of coordinated action for cooperation and aggression are taken as prima facie evidence for group selection. But as Pinker makes clear, all of these behaviors could plausibly emerge from a psychology designed by forces other than group selection. This is important enough to say twice: social or group-oriented features of human nature, such as moral intuitions of fairness, valuing group loyalty, etc., could emerge in principle from selection acting at any level of organization, not merely the group-level [1-3].As such, the mere existence of such features (the vast majority of evidence generally marshaled in this debate) cannot possibly arbitrate between competing theories. The only evidence capable of doing so is evidence of special design.

For instance, we have shown mathematically how generosity in one-shot, anonymous encounters arises from individual selection for direct reciprocity, once realistic constraints on what an organism can know are taken into account [4].Thus, the mere existence of experimental evidence of cooperation in one-shot, anonymous encounters cannot count as evidence in favor of group selection. Further, we have shown with behavioral experiments that one-shot generosity is in fact conditioned on how valuable the participant perceives the partner to be as a potential cooperative partner [5]. Thus, at least this aspect of human generosity appears designed for individual rather than group benefit.

The debate is not about selfishness versus generosity or individualism versus groupishness. The debate is about whether generosity, cooperation, altruism, etc. are instantiated by a psychology designed by individual or group selection. If the former, then this psychology should have design features that, on average and under conditions that match ancestral conditions, eventually lead to net benefits for the individuals or their kin. If the latter, then this psychology should have design features that, on average, lead to group benefits even if the individual suffers.

When the evidence is examined for indications of special design in this way, the verdict for group selection is decisively negative. Yes, people will sometimes give money to others even in artificial, controlled situations where such generosity cannot be repaid. But they do so contingent on cues that their actions are observed by others [6,7],that their interactions may possibly continue [4],that the other has a reputation for good behavior [5],and upon other cues that cooperation will probabilistically return a benefit [8];this reveals the design of a mechanism crafted to generate individual returns from cooperative relationships, not group-wide benefits from inter-group competition. Yes, people will sometimes punish others' bad behavior at a personal cost even in artificial, controlled situations where such action cannot beneficially redound on the punisher (either preferentially or at all). But they do this contingent on cues that they could benefit by recruiting those other's good behavior [5,9],and upon cues of the possibility of improving their standing in the eyes of others [10];again, this reveals a design tailored for individual, not group-wide, returns. Yes, people recognize large, symbolically marked ethnic groups extending over many miles and thousands of individuals. But symbolically marking a group is not limited to large ones—and if you disagree perhaps the Sharks or the Jets would have a word with you. Yes, cooperative practices differ from culture to culture, but in ways that sensibly mesh with ecological variation that drives individual returns from cooperation [11,12].

In short, group selection (or as some prefer, multilevel selection) may be a completely coherent model that may suggest selection pressures that have acted to shape human nature. The problem for theories of group selected adaptations is that the preponderance of the evidence suggests that it simply hasn't done so.

___

_References

_{1. Petersen MB, Sznycer D, Cosmides L, Tooby J (2012) Who Deserves Help? Evolutionary Psychology, Social Emotions, and Public Opinion about Welfare. Political Psychology 33: 395-418.}

_{2. Petersen MB, Sell A, Tooby J, Cosmides L (2012) To Punish or Repair? Evolutionary Psychology and Lay Intuitions about Modern Criminal Justice. Evolution and Human Behavior.}

_{3. Boyer P, Petersen MB (2012) The Naturalness of (Many) Social Institutions: Evolved Cognition as Their Foundation. Journal of Institutional Economics 8: 1-25.}

_{4. Delton AW, Krasnow MM, Cosmides L, Tooby J (2011) The evolution of direct reciprocity under uncertainty can explain human generosity in one-shot encounters. Proceedings of the National Academy of Sciences of the United States of America.}

_{5. Krasnow MM, Cosmides L, Pedersen EJ, Tooby J (under review) What are reputation and punishment for?}

_{6. Hoffman E, McCabe K, Shachat K, Smith V (1994) Preferences, property rights, and anonymity in bargaining games. Games and Economic Behavior 7: 346-380.}

_{7. Haley KJ, Fessler DMT (2005) Nobody's watching? Subtle cues affect generosity in an anonymous economic game. Evolution and Human Behavior 26: 245-256.}

_{8. Yamagishi T, Kiyonari T (2000) The group as the container of generalized reciprocity. Social Psychology Quarterly 63: 116-132.}

_{9. Carpenter JP, Matthews PH (2012) Norm enforcement: anger, indignation, or reciprocity? Journal of the European Economic Association 10: 555-572.}

_{10. Kurzban R, DeScioli P, O'Brien E (2007) Audience effects on moralistic punishment. Evolution and Human Behavior 28: 75-84.}

_{11. Delton AW, Krasnow MM, Cosmides L, Tooby J (2010) Evolution of fairness: Rereading the data. Science 329: 389.}

_{12. Lamba S, Mace R (2011) Demography and ecology drive variation in cooperation across human populations. Proceedings of the National Academy of Sciences 108: 14426-14430.}

Distinguished Professor of Anthropology, UCLA; Cowan Professor, Santa Fe Institute

Postdoctoral Researcher, Centre for the Study of Cultural Evolution, Stockholm University

There is much to disagree with in Steven Pinker's essay. We focus on two important misconceptions that lie at the heart of his argument.

The first misconception is his claim that it is always a mistake to think of function or design at the level of groups. This flies in the face of more than 150 years of thinking in biology. The many examples of intragenomic conflict leave no doubt that there is selection acting on genes within organisms, and this sometimes reduces the fitness of individuals carrying those genes. However, this does not mean that we should abandon the notion of function at the level of individuals. We teach our students in physical anthropology that we can understand the design of the human pelvis by asking how the shape of the pelvis affects the energetics of bipedal locomotion. From the gene's point of view, this is group function. Variation at hundreds (or maybe thousands) of genetic loci affect variation in the anatomy of the pelvis, but their action is coordinated by selection to create function at the level of the individual. Ever since Darwin, physiologists, morphologists, and behaviorists have made much progress using individual functionalist thinking to provide causal explanations for the diversity of life-forms.

The same goes for the function of some kinds of social groups. Just as there is intragenomic conflict within individuals, there is also conflict among individuals within social insect colonies. For example, worker bees sometimes lay unfertilized eggs that develop into male reproductives, while other workers police the hive killing such eggs. Despite such conflicts of interest, many attributes of the colony only make sense in terms of group function—the exquisite design of the cooling systems in termite mounds, the hexagonal space-saving shape of honeycomb, or the amazing made-out-of-ants structures used by army ant colonies in their travels. These are true group level designs, not simply aggregated effects of individual design like a herd of fleet deer. The ingenious mechanisms that colony members use to coordinate their behavior are of great interest, but this does not change the fact that these behaviors are best understood in terms of group function.

The reason that there is design at the level of individuals is that meiosis ensures that much of the genetic variation in populations is among individuals, and this means that the interests of genes at these many loci are aligned. The same goes for social insect colonies (although things are a bit more complicated in some cases). This "multi-level" description based on genetic variation within and among groups is usually equivalent to an "inclusive fitness" description based on the relatedness within groups (See David Queller's comment in this discussion for witty summary of this idea). The approaches are equivalent, so it is a matter of taste or expediency which approach to apply to a particular problem.

Our view is that thinking in terms of group function is most useful when there is relatively more heritable variation among the higher level entities (individuals, colonies) and the adaptation under consideration does not generate much conflict of interest between the lower level entities (genes, individuals). For the social behavior of most mammals, we prefer the inclusive fitness framework, but for humans we believe that both individual and group level approaches are essential.

This is where Pinker's second misconception comes into play. Pinker asserts, "It's only when humans display traits that are disadvantageous to themselves while benefiting their group that group selection might have something to add." This is wrong because it conflates the evolution of altruism with the evolution of design at the group level. Selection among groups can create design at the group level which does not entail altruistic sacrifice at the individual level. This is particularly important for understanding human cooperation that is sustained by norms enforced by third-party rewards and punishments. Turkana warfare [1]illustrates why. The Turkana form war parties averaging about 300 men to raid neighboring ethnic groups and steal livestock [2].These raids produce collective benefits, but there are opportunities to free ride during the raid—desertion, cowardice in battle, and unfair appropriation of the loot. However, free riders are punished by members of their community. If punishment were sufficiently severe to make free-riding costly, then people wouldn't cheat and participation in raids would be beneficial to individuals. Then, according to Pinker, we'd be done: men fight because it is in their interest to do so. There is nothing left for selection at the group level to explain.

 But think—systems of rewards and punishments can stabilize a vast range of outcomes, including non-cooperative ones, on a wide range of scales. As long as the cost of being punished exceeds the cost of following the norm, obeying the norm will be self-interested. It doesn't matter what the norm requires. Mutually enforced sanctions could maintain cooperative or noncooperative norms: "You may steal your neighbor's cows to feed your family", or "You may not steal your neighbor's cows to feed your family". Similarly, punishment can maintain norms at different scales. "Do not steal a clan member's cattle, but the cattle of other clans are for brave men to steal" or "Do not steal the cattle of someone from your tribe, but the cattle of other tribes are for brave men to steal". These are both group-beneficial norms, but one benefits clans, the other benefits tribes. The list of possible variations is nearly endless.

When a vast range of outcomes are consistent with self-interest, any real account of human behavior must specify what gives rise to the norms that are actually observed—an equilibrium selection mechanism in the jargon of evolutionary game theory. Selection among groups is one such mechanism: groups vary, these variations affect their success in competition with other groups, and the descendant groups resemble their parents. Selection among groups will lead to the spread of normative systems that enhance the group's competitive ability. There are lots of examples of "cultural group selection" and it may be an important equilibrium selection mechanism in human populations [3].There are a number of competing hypotheses about equilibrium selection in human societies [4].These mechanisms are not mutually exclusive, and their relative importance over the long run of human cultural evolution is an important, but unanswered empirical question.

Pinker's answer to the equilibrium selection problem is: leave it to the historians. (There is some irony in the fact that Pinker shares this view with Richard Lewontin, one of evolutionary psychology's most vociferous critics.) This is really not satisfactory. Historians focus on concrete narrative accounts of particular historical sequences. With some exceptions, historians eschew generalizing mathematical theory and typically ignore societies without written records. But people have been cultural organisms for a long time and the great majority human societies lack written records. Understanding how culturally transmitted norms and institutions have shaped human social life requires generalizing, preferably mathematical, theory that can generate predictions that can be compared to the broad patterns in the archaeological and ethnographic records.

Such theory is essential even for evolutionary psychologists like Pinker. To see why, consider his claim that evolutionary theory predicts that people will be concerned about their reputations. Third-party punishment is absent in other animals, and in humans it is virtually always regulated by culturally transmitted norms. What gets you a good reputation depends on the content of these norms. People won't care about being seen as cooperative unless over the long run human moral systems have supported cooperative behavior. Thus, Pinker's claims about human psychology depend on evolutionary predictions about the content of norms. Here Pinker relies on models of indirect reciprocity that apply only to pairwise interactions, and depend on a number of dubious assumptions including: (1) a single mutation can shift an individual from defection to a complex reciprocating strategy, (2) individuals have accurate knowledge of past behavior of all other members of their social groups, and (3) there is never significant disagreement about past behavior. We believe that a useful theory of reputation must apply to collective action, and be robust to changes in assumptions about the genetics and actor's knowledge of past behavior, and this gets you right back to the equilibrium selection problem.

_{De Silva, H., C Hauert, A. Traulsen, K. Sigmund, 2010, Freedom, enforcement, and the social dilemma of strong altruism, Journal of Evolutionary Economics, 20, 203-217.}

_{Mathew, S. and R. Boyd. 2011, Punishment sustains large-scale cooperation in prestate warfare. Proceedings of the National Academy of Sciences (USA) 108, 11375–11380.}

_{Richerson, P. J. and R. Boyd, 2005, Not by Genes Alone, How culture transformed human evolution, University of Chicago Press, Chicago.}

_{Young, P. 1998, Individual Strategy and Social Structure: An Evolutionary Theory of Institutions, Princeton, NJ: Princeton University Press.}

_{[1] Mathew and Boyd 2011.}

_{[2] Matthew & Boyd 2011.}

_{[3] Richerson and Boyd 2005.}

_{[4] e.g. Young 1998, De Silva et al 2010.}

Ecologist and Evolutionary Biologist; Distinguished Research Director, Centre National de la Recherche Scientifique; External Professor, Santa Fe Institute

There Is No (Single) Holy Grail

Much of the discussion regarding group selection has been fueled by its fuzziness for many, yet the appeal to many of holy-grail, single-process explanations for human social behaviors. I believe that in his recent essay "The False Allure of Group Selection" Steven Pinker is incorrect in applying Occam's razor to say that simpler, ground level, individual selection is sufficient to explain traits in human groups. There are counter-arguments that selection on different levels is involved (or as I argue below, was involved) in the establishment of behavioral and institutional adaptations.

Challenge – Signal – Response.  In understanding the origin and current relevance of group traits, one must consider the context in which they are elicited, the elicitor, context-dependent signals between individuals, and their behavioral responses. Take for example feelings of patriotism, elation, and resolve when in a like-minded group, hearing and singing one's national anthem and seeing one's national flag raised. These unifying signals elicit emotional and physiological responses (e.g. adrenaline secretion). What for? Those groups that more effectively react to challenges will tend to survive and spread, even if the precise signal is altered, for example, when colonizing a new habitat (e.g., the colonizers may invent a new anthem and new flag, but have the same group level traits). What counts is whether the specific stimulus and response are shared by many or most members of a given group, and whether their evocation changes the prospects of group survival, growth and spread.

Group Traits: Memes, Genes and Institutions.  Group selected traits may have either a cultural and/or genetic basis. Although there is not perfect equivalence between the two, both function in similar and sometimes complementary ways. "Culture" is one of the defining characteristics of human groups, but that it is not a one-off occurrence is vouched by one important aspect of culture—tool use—in other primates, mammals, and even some birds and cephalopods. The question is whether group traits in humans depend only on individual minds though social learning and to what extent our genetic material gives us proclivities towards prosocial behaviors. For example, why would groups with higher proportions of altruistic punishers tend to win-out over those with fewer? Or, rather, do certain individuals only stand to benefit from this and other similar social behaviors? The answer is likely to depend on context (see below). But the key reason why our minds (individual and collective) and DNA retain the seeds of such adaptations and for why emergent features such as social norms and institutions exist at the group level are that those groups which effectively coordinate and communicate tend to survive, grow, and spread. This does not necessarily mean that all social behaviors are for mutual benefit or the good of the group! Rather, those groups that tend to have higher frequencies of these behaviors tend to win out over those that do not.

Origins.  Why do some people actually volunteer to be in armies, help others in distress risking (whether aware or not) their own possible injury, politely open doors for others, etc.? More innocuously, why do people enjoy sporting events and fervently support the home team? All of these behaviors seem at odds with individually selected adaptations in today's world, and they also may very well appear to be at odds with the process of group selection. The likely reason for why is that many or most group traits originated in small, hunter-gatherergroups. Such traits may still be under selection, but are evoked in (slightly) different contexts to those in which they were originally selected. At the time when these traits were originally selected, if you asked someone to identify herself, then she may have given you her name, her tribe's name and her function within the tribe. That is, units of selection were few and distinct. Today, she might give you her name, city and state of birth, country, religion, club membership, profession, etc. This is one possible reason for why selection for group traits will be slower and more transient today than in the distant past. Indeed, it is possible that many group-level traits no longer evolve; they are the fittest decedents of past.

_{Related reading}

_{Choi J.K. & Bowles S. 2007. The coevolution of parochial altruism and war. Science 318: 636-640.}

_{Danchin É. et al. 2011. Beyond DNA: integrating inclusive inheritance into an extended theory of evolution. Nature Reviews Genetics 12: 475-486.}

_{Gintis H. 2012. The evolution of human cooperation. Social Evolution Forum.} 

_{Hochberg M.E. et al. 2003. Socially mediated speciation. Evolution 57: 154-8.}

_{Wiltermuth S.S. & Heath C. 2009. Synchrony and cooperation. Psychol Sci 20:1–5.}

Professor, Department of Ecology and Evolution, University of Chicago

The Problem With Group Selection

Steven Pinker's elegant essay points out the theoretical problems with group (or "multilevel") selection, how it is regularly misunderstood and misapplied, and how group selection promoting genetic evolution is regularly conflated with processes promoting cultural evolution—nongenetic change in human society.

I've written in detail about the problems with the recent revival of enthusiasm for group selection on Why Evolution is True ^{[1, 2]}, a revival accompanied by indictments of its alternative, kin selection. As Pinker notes, models of group selection are either mathematically equivalent to those based on kin selection but less tractable, or are so nebulous that they can't be analyzed at all. Further, claims that kin selection is less useful than group selection in understanding nature are simply wrong. Two recent papers: "Social semantics: how useful has group selection been?" by S. A. West, A. S. Griffin & A. Gardner ^[3], and "The validity and value of inclusive fitness theory" by Andrew F.G. Bourke" ^[4] compare kin versus group selection in their ability to promote useful research and help us understand animal behavior; both show decisively that while the concept of kin selection has fostered tremendously productive research on evolution, the notion of group selection has come up dry.

Because I am an experimental biologist and not a theoretician, I see the biggest problem with group selection as its failure to explain anything about nature. The hypothesis is, as Pinker notes, is designed to explain those features of evolution—especially traits like altruism and cooperation in our own species—that seem disadvantageous to individuals but good for groups. We can predict, then, that if group selection were common and kin selection rare, we would often observe altruistic behavior in nature between individuals who were completely unrelated. That is, individuals would often sacrifice their lives (or reproduction) to help those who don't share their genes.

And that is exactly what we don't see. When such altruism does occur, as in ants, bees, termites, or naked mole rats, it is almost always between close relatives. "Eusociality" in insects—societies in which sterile castes of workers defend one or a few reproductive individuals—has evolved only in those lineages whose ancestral females mated once rather than multiply. Because single matings increase relatedness among offspring (they are full siblings rather than half siblings or more distant relatives), this is very strong evidence for the operation of kin rather than group selection in the evolution of eusociality and "altruism" in social insects. Other phenomena, such as the sex ratios seen in the nests of those insects, also support kin selection as a powerful stimulus for the evolution of social insects.

The data, then—as well as the absence of data for pervasive altruism—point to group selection as a rather unimportant force in nature. Except in species comprising groups of relatives, or those that practice reciprocal altruism ("I'll scratch your back; you scratch mine"), we simply don't see animals sacrificing themselves for nonrelatives. (We expect reciprocal altruism to evolve by normal, "individual selection" in groups of animals that are stable and whose members recognize each other, like primates.) True altruism in humans, which does exist, is the sole exception, but, as Pinker notes, this is almost certainly because of cultural features unique to our species.

In fact, I'm unaware of a single behavior in animals that is completely disadvantageous to individuals but useful for groups, and that doesn't involve dispensing aid to relatives. That itself is powerful evidence against the pervasiveness of group selection. It is curious but telling that proponents of group selection always concentrate on its logical and mathematical soundness rather than on its usefulness in explaining features of real organisms in nature.

If group selection is so impotent in helping us understand nature, why is it so popular? Here we wade into the marshy hinterlands of psychology, but I suspect several factors are at play. First, its proponents are well known scientists and psychologists who promote group selection in best-selling books, while the detractors are constrained to publish their criticisms in technical journals that aren't read by the public (Pinker's essay is a rare exception). And those criticisms are technical, so they're not only inaccessible to the layperson, but largely incomprehensible to those without training in mathematics and biology.

Too, it seems to me that people want to believe in group selection. We like to think that stuff like religion, cooperation, and altruism have spread by group selection because that involves the concept of harmonious and cooperating groups. Such a notion is deeply appealing to those who have a dislike for the idea of the "selfish gene," mistakenly confusing that notion with the idea of selfish individuals. As all evolutionists know, or should know, cooperation and altruism can evolve via selfish genes!

Finally, much of the work on group selection has been funded by the John Templeton Foundation, an enormously wealthy organization with an agenda to harmonize faith and science. The idea of group selection, with its spiritual and religious connotations—the process is often used to explain the prevalence of religion and societal harmony—is right up their alley. So the proponents of group selection monopolize not only the megaphones but the funding. In science, money talks.

In the end, group selection, while innately appealing, has not helped us understand very much about nature. We could reply to advocates of group selection as Laplace replied to Napoleon when queried about why God was absent from Laplace's great book on celestial mechanics: "I have no need of that hypothesis."

_____

_{1. "The demise of group selection", in Why Evolution Is True, June 24, 2012}

_{2. Did human social behavior evolve via group selection? E. O. Wilson defends that view in the NYT, ", in Why Evolution Is True, June 26, 2012}

_{3. in "Proceedings of the Royal Society B"}

_{4. in "J. Evol. Biol." 21 (2008) 374-385}

Distinguished Professor Emeritus, University of California-Davis; Visiting Professor, Institute of Archaeology, University College London

In these remarks I concentrate on the Steven Pinker's misconceptions about cultural evolution, cultural group selection, and gene-culture coevolution.

The problems in his essay begin with the idea that mutations have to be random with respect to fitness for natural selection to occur. Since cultural evolution manifestly includes the inheritance of acquired variation (if I learn something interesting, I can teach it to others), defining natural selection in this way seems to exclude anything cultural from the effects of natural selection. This is illogical. The problem is quantitative not qualitative. Natural selection works on any pattern of heritable variation. When animals rely only their own exploration and trial and error learning to gain the knowledge and skills necessary to survive, then the products of their learning are not heritable. This is approximately the case in many species where social learning is absent or of marginal importance. But for humans, learning from others is very important. We inherit most of what we know from others. Human are a veritable cultural adaptive radiation based on information that has been passed down the generations by social learning.

When a culture includes complex technology, such as the ocean going canoes of Polynesians, or complex social institutions, such as the Polynesian ranked lineage system, most members of the population follow cultural "recipes" acquired from their elders. Natural selection can work on such cultural traits. Polynesians who sailed canoes that were badly designed or built were less likely to have left offspring or imitators. Communities that had faulty institutions were more likely to have disbanded due to environmental deterioration, social collapse, or conquest by competing groups. Of course, Pinker is right that individuals sometimes make deliberate innovations. But the products of an inventor's ingenuity do not often supplant existing devices and techniques ⁽¹⁾. It is not easy to improve on complex adaptations. Many religious innovators preach a new doctrine for every Joseph Smith whose teachings led to the formation of Mormonism. In human culture, non-random variation and natural selection can both play roles in the evolution of cultural variation so long as the effects of non-random variation are not so strong as to overwhelm the transmitted aspect of culture.

Moderate rates of non-random innovation and natural selection together have a neat synergy in cultural evolution. In a population trying to exploit a new environment, no one yet knows what the best practices might be. In these circumstances, natural selection cannot act because there are no adaptive variants to select. Given enough time, adaptive variants will eventually emerge at random. But suppose a few innovative members of the population introduce fitness-enhancing variation at a faster, but still low, rate compared to random variation. Such a non-random process provides useful cultural variants so that selection can act sooner than if had to wait for random variation to produce adaptive variants.

Natural selection is a slow and painful process, even if it gets a boost from non-random innovation. In the case of culture non-natural selection based on innate or cultural biases can act to pick out favorable cultural variants. If such biases have evolved under the influence of natural selection, they will often act in the same direction natural selection would act. Biased learning or teaching thus also speeds up the cultural evolution of adaptations relative to what natural selection could achieve by itself, even including non-random variation. The human capacity for culture creates a system that ties decision-making to inheritance. This provides humans with a great advantage by speeding up the adaptive process without expecting individual innovators and innovation adopters to perform cognitive miracles. The evolution of cultural variants can lead to complex adaptations much faster than the evolution of genes alone because the work of creation and diffusion of favorable variants is distributed among many minds. Colleagues and I elsewhere argue at greater length than is possible here ⁽²⁾ that Pinker's view of culture depends on an impossibly high degree of individual cognitive prowess that amounts to a magical "skyhook" in Daniel Dennett's ⁽³⁾ felicitous phrase. Even if natural selection plays a subordinate role, as Darwin believed it did in "civilized times" ⁽⁴⁾, a rich formal theory of cultural evolution (a theory of history in other words!) has been built on the framework of combining non-random innovation, selective innovation adoption, and natural selection ^{(5, 6)}.

Natural selection on large scale patterns of cultural variation is plausible because the cultural variation between neighboring groups that might compete is typically much larger than the genetic variation between the same groups ⁽⁷⁾. The reasons are not hard to see; all human groups are more or less open to immigration. Groups intermarry and intermarriage is a very effective conduit for genes. This is less true of culture. Because culture evolves more rapidly than genes, groups will continue to differ despite migration. A large body of social psychology research has characterized the active mechanisms that damp down variation within groups and protect between group variation from the effects of migration. Human social groups are psychologically very salient entities ⁽⁸⁾as Pinker acknowledges. Groups often have different norms and institutions. Being able to conform to local norms and institutions is important for individual success, because institutions include carrots and sticks encouraging conformity to their dictates. Even individuals who personally dissent may nevertheless obey the group's rules. In many hunting and gathering societies, egalitarian norms prevent would-be dominant males from taking the resources of others or even hoarding resources they themselves have acquired. This has the effect of reducing individual selection within groups and making it easier for selection to act on any existing cultural variation between groups ⁽⁹⁾.

Not surprisingly, children are adapted to efficiently learn norms the norms of their group ⁽¹⁰⁾. Hence, immigrants, especially immigrant children, typically assimilate to the groups they or their parents join without appreciably diluting their host cultures. Infant chimpanzees raised as children, by contrast, are no more able (or willing) to acquire norms than language ⁽¹¹⁾.

Darwin ⁽⁴⁾ proposed that tribal scale selection was important in "primeval times" in the evolution of pro-social "instincts" such as empathy and patriotism. We certainly have ample ethnographic evidence of such between-group competition. ⁽¹²⁾ Such primeval selection might have been based mainly on cultural differences between groups as it has been in ethnographic times.

Contrary to Pinker's argument, the evidence for other-regarding dispositions in humans extends beyond the results of the economic game experiments he mentions. On the experimental side, see Batson's experiments testing his empathy-altruism hypothesis against purely individualistic alternatives ⁽¹³⁾. Humans routinely share useful information with their fellow group members and assume information given to them to be provided honestly and in good faith. The fact that we have complex adaptations such as language that facilitates such cooperative behavior suggests that it has a long history in our species ⁽¹⁴⁾.

It is interesting to consider what human life would be like if people were fundamentally selfish. Some individuals behave in ways that we characterize as "psychopathic" and this behavior has a devastating effect on the functioning of the groups they are part of ⁽¹⁵⁾. Psychopathy involves a lack of empathy and habitual disregard of norms. It is highly disruptive to the organizations psychopaths inhabit. On many accounts based on the behavior of chimpanzees ⁽¹⁶⁾, human psychopaths (perhaps 1% of living populations) rather resemble our last common ancestor with the apes (and the selfish egoists of the bare-bones economic and evolutionary theory that was influential in the latter part of the 20^th century). Psychopaths themselves typically suffer because their excessively self-regarding behavior is checked by institutions. It is hard to see how human societies could function as they do if even a large minority behaved like psychopaths.

Robert Boyd and I have proposed that in the human species, a pro-social psychology arose by cultural group selection and gene-culture coevolution. Once our ancestors were taking some advantage of cultural transmission and evolution, simple social institutions would have become part of their adaptive repertoire, such as stable mating bonds that would have the effect of encouraging patrilateral as well as matrilateral kin interactions ⁽¹⁷⁾. Then, social selection within groups, operating through primitive social institutions, would have generated selection on genes in favor of Darwin's pro-social instincts. In ethnographically known hunting and gathering societies, people who ignore social norms are shunned as mates; sexual selection based on conformity to norms would have been a very powerful force shaping the innate features of human social psychology. Many rounds of gene-culture coevolution would have eventually built living humans who, given the right norms and institutions, are capable of considerable feats of cooperation. Bowles and Gintis have proposed a different gene-culture coevolution scenario ⁽¹⁸⁾.

I am not aware of any writings of Pinker's that confront the hypotheses and evidence for the importance of cultural evolution, cultural group selection, and gene-culture coevolution. By not giving culture a reasonable role in human evolution in the first instance, he falsely relieves himself of any need to deal with further evidence.

_{1.         Petroski H (1992) The Evolution of Useful Things (Vintage Books, New York) p xi + 288.}

_{2.         Boyd R, Richerson PJ, & Henrich J (2011) The cultural niche. Proceedings of the National Academy of Sciences USA In press.}

_{3.         Dennett DC (1995) Darwin's Dangerous Idea: Evolution and the Meanings of Life (Simon & Schuster, New York) p 586.}

_{4.         Darwin C (1874) The Descent of Man and Selection in Relation to Sex (American Home Library, New York) 2nd Ed p 868.}

_{5.         Mesoudi A (2011) Cultural Evolution: How Darwinian Theory Can Explain Human Culture & Synthesize the Social Sciences (University of Chicago Press, Chicago) p xv + 264.}

_{6.         Turchin P (2006) War and Peace and War: The Life Cycles of Imperial Nations (Pi  Press, New York) p viii + 405.}

_{7.         Bell AV, Richerson PJ, & McElreath R (2009) Culture rather than genes provides greater scope for the evolution of large-scale human prosociality. Proceedings of the National Academy of Sciences USA 106(42):17671-17674.}

_{8.         Haslam SA (2001) Psychology in Organizations: The Social Identity Approach (Sage Publications, London) p xvi + 411.}

_{9.         Boehm C (1997) Impact of the human egalitarian syndrome on Darwinian selection mechanics. American Naturalist 150(supplement):S100-S121.}

_{10.         Chudek M & Henrich J (2011) Culture–gene coevolution, norm-psychology and the emergence of human prosociality. Trends in Cognitive Sciences 15(5):218-226.}

_{11.         Hayes C (1951) The Ape in Our House (Harper, New York,) p 247.}

_{12.         Otterbein KF (1985) The Evolution of War: A Cross-Cultural Study (Human Relations Area  Files Press, New Haven CT) p 165.}

_{13.         Batson CD (2011) Altruism in Humans (Oxford University Press, New York) p vi + 329.}

_{14.         Richerson PJ & Boyd R (2010) Why possibly language evolved. Biolinguistics 4(2-3):289-306.}

_{15.         Babiak P & Hare RD (2006) Snakes in Suits: When Psychopaths Go to Work (HarperCollins, New York) p xv + 336.}

_{16.         Vonk J, et al. (2008) Chimpanzees do not take very low cost opportunities to deliver food to unrelated group members. Animal Behaviour 75:1757-1770.}

_{17.         Chapais B (2008) Primeval Kinship: How Pair-Bonding Gave Birth to Human Society (Harvard University Press, Cambridge MA) p xv + 349.}

_{18.         Bowles S & Gintis H (2011) A Cooperative Species: Human Reciprocity and its Evolution (Princeton University Press, Princeton) p xii + 262.}

 Anthropologist; University Professor; Director, Institute of Cognitive and Evolutionary Anthropology, University of Oxford; Author, Arguments and Icons

Senior Lecturer, Department of Psychology, University of London

According to Pinker, group selection "adds nothing to conventional history" as an explanation of cultural change. Rather than arising from processes of random mutation and indifferent selection, he argues, cultural traits arise and spread as a result of the complex intentions and interactions of agents: "Conquerors, leaders, elites, visionaries, social entrepreneurs, and other innovators use their highly nonrandom brains to figure out tactics and institutions and norms and beliefs that are intelligently designed in response to a felt need".

Pinker takes the view that natural selection isn't natural selection unless mutations are random (or, as he clarifies, blind to their effects[1]). As he rightly observes, however, nobody 'owns' the theory of natural selection. On our view, to refer to the processes by which cultural variants arise and proliferate using the "verbiage" of natural selection is not to indulge in poetic but pointless redescription. As we see it, the utility of the theory (and its associated verbiage) does not derive from the randomness (or foresightlessness) of the variability, but rather from the efficiency of explanation the theory affords.

While we agree with Pinker that the multifarious intentions of agents shape the emergence and spread of cultural traits, these are merely aspects of proximate causation in cultural evolution and do not preclude the possibility of selection by consequences. Moreover, a comprehensive account of cultural evolution would embrace all of Tinbergen's Four Whys. Consider the problem of explaining why the moving pistons in a car drive the main axle. A proximate explanation is that the pistons in an internal combustion engine deliver power to the car's wheels via a series of mechanical devices such as a crankshaft and gearbox. (And of course these features were "designed" that way.) An ultimate explanation is that vehicles equipped with this arrangement actually move and the design was selected for. But a more complete explanation must also consider developmental questions concerning the nature and sequencing of the car assembly process, such as how pistons, crankshafts, gearboxes, and axles come to be installed and connected up. And we also need to consider the constraints on design imposed by prior forms of motor vehicle construction, which are essentially questions about phylogeny.

The Four Whys provide explanatory power in the biological domain but can also do so in the cultural domain. Perhaps the reason why Pinker finds it hard to recognize this is that intelligent design is commonly associated with supernatural theories of creation that have no place in science. If so, that impediment to understanding can surely be dismissed. To appreciate that designed creations evolve does not expose us to the charge of theistic creationism.

[1] Genetic mutations are arguably no more "random" than deliberate alterations in artifact form – both are subject to a host of constraints within which new variants can occur.

Author, A Cooperative Species: Human Reciprocity and Evolution

On the Evolution of Human Morality

Steven Pinker's thoughtful remarks concerning group selection present a useful occasion for clearing some misconceptions surrounding recent developments in the behavioral sciences concerning our understanding of moral vs. self-interested behavior. Initiated in 1966 by George C. Willams' Adaptation and Natural Selection and followed a decade later by Richard Dawkins' The Selfish Gene, evolutionary biologists in the last quarter of the Twentieth century came to view humans as fundamentally selfish, contributing to society only when socially-imposed rewards and punishment render it in their self-interest to do so. Dawkins, for instance, opines in the opening pages of The Selfish Gene, "We are survival machines—robot vehicles blindly programmed to preserve the selfish molecules known as genes.... a predominant quality to be expected in a successful gene is ruthless selfishness. This gene selfishness will usually give rise to selfishness in individual behavior.... Anything that has evolved by natural selection should be selfish."

Of course, it does not appear in our daily life that everyone is selfish, and if we introspect, most of us will agree that we try to behave, however successfully or unsuccessfully, as moral beings willing to sacrifice personal amenities in the pursuit of truth, justice, loyalty and compassion. Dawkins' explanation is that human morality is a cultural facade laid upon our basically selfish human nature. "Be warned," he states, "that if you wish, as I do, to build a society in which individuals cooperate generously and unselfishly towards a common good, you can expect little help from biological nature. Let us try to teach generosity and altruism, because we are born selfish."

But why do fundamentally selfish beings, which is what humans are according to the selfish gene theory, accept cultural norms that contradict their natural strivings? Richard Alexander answered this question in 1987 in his The Biology of Moral Systems with his concept of indirect reciprocity, according to which we all continually evaluate others for possible future gainful interactions, and we reject individuals who violate norms of reciprocity. The somewhat more general answer offered by Pinker is that is that each of us conforms to social norms out of fear of losing our good reputation. What appears to be self-sacrifice is thus simply a superficial veneer covering our selfish natures. "Scratch an altruist," biologist Michael Ghislin eloquently wrote in 1974, "and watch a hypocrite bleed."

Pinker frames the issue in terms of sacrificing personal interests on behalf of the group. "What we don't expect to see," he writes, "is the evolution of an innate tendency among individuals to predictably sacrifice their expected interests for the interests of the group." This is not the correct way to frame the issue. People do not generally "sacrifice on behalf of the group." Rather, people have moral principles that they strive to uphold, and that compete with their material interests. When I behave honestly in a transaction, I may have no intention whatsoever of sacrificing on behalf of my transaction partners, much less on behalf of my society. I just do what I think is the morally correct thing to do. When I bravely participate in a collective action against a despotic regime, I am upholding my moral principles, not sacrificing on behalf of the group. Indeed, it is no sacrifice at all to behave morally, because we humans care about our moral worth in much the same way as we care about our material circumstances.

The past few decades have seen the massive accumulation of evidence in favor of the view that human beings are inherently moral creatures, and that morality is not a simple cultural veneer. Humans are born with a moral sense as well with a predisposition to accept and internalize moral norms their society, and often to act on these moral precepts at personal cost. In our book, A Cooperative Species, Samuel Bowles and I summarize a plausible model of human nature in which "people think that cooperating is the right thing to do and enjoy doing it, and that they dislike unfair treatment and enjoy punishing those who violate norms of fairness." Most individuals include moral as well as material goals in their personal decision-making, and they willingly sacrifice material interests towards attaining moral goals. It is this view that I will defend in my remarks.

Pinker does not present, and indeed makes light of the body of research supporting the existence of a basic human moral sense, suggesting that there is only one piece of evidence supporting the view that people behave morally when their reputations are not at stake: "It seems hard to believe," he says, "that a small effect in one condition of a somewhat contrived psychology experiment would be sufficient reason to revise the modern theory of evolution, and indeed there is no reason to believe it. Subsequent experiments have shown that most of the behavior in these and similar games can be explained by an expectation of reciprocity or a concern with reputation." Because expectation of reciprocity and concern for reputation are basically selfish and do not involve a fundamental respect for moral values, Pinker is simply reiterating Dawkins' message of a half-century ago that we are the selfish product of selfish genes.

1. Morality and Human Nature

Today the economics and psychology journals, including the most influential natural science journals, Science and Nature, are full of accounts of human moral and prosocial behavior. Pinker dismisses this evidence by asserting that "Any residue of pure altruism" beyond self-interested reciprocity and reputation building "can be explained by the assumption that people's cooperative intuitions have been shaped in a world in which neither anonymity nor one-shot encounters can be guaranteed." In other words what looks like moral behavior is just a form of mental error due to imperfections of the human brain.

The empirical evidence on cooperation in humans does not support Pinker's view. The social interactions studied in the laboratory and field always involve anonymity, so subjects cannot help or harm their reputations, and they usually are one-shot, meaning that subjects cannot expect to be rewarded in the future for sacrifices they make at a given point in time.

Pinker does cite a few studies that support his position. "Subsequent experiments have shown that most of the behavior in these and similar games can be explained by an expectation of reciprocity or a concern with reputation." Let us see what these studies in fact say. Reciprocity, says Pinker, "is driven not by infallible knowledge but by probabilistic cues. This means that people may extend favors to other people with whom they will never in fact interact with again, as long as the situation is representative of ones in which they may interact with them again." The only published paper he cites is by Andrew W. Delton, Max M. Krasnow, Leda Cosmides and John Tooby, "Evolution of Direct Reciprocity Under Uncertainty can Explain Human Generosity in One-shot Encounters." ¹ This paper (and several related papers coming out of the Center for Evolutionary Psychology at Santa Barbara, California) show that, in the authors' words "generosity is the necessary byproduct of selection on decision systems for regulating dyadic reciprocity under conditions of uncertainty. In deciding whether to engage in dyadic reciprocity, these systems must balance (i) the costs of mistaking a one-shot interaction for a repeated interaction (hence, risking a single chance of being exploited) with (ii) the far greater costs of mistaking a repeated interaction for a one-shot interaction (thereby precluding benefits from multiple future cooperative interactions). This asymmetry builds organisms naturally selected to cooperate even when exposed to cues that they are in oneshot interactions."

This statement is of course not only true, but completely obvious, and does not require sophisticated academic papers to validate its truth. However it does not explains human generosity. It is elementary logic that to say that P explains Q does not mean that if P is true then Q is true, but rather the converse: whenever Q is true, then P is true as well. In the current context, this means that whenever subject A sacrifices on behalf of stranger B in an experiment, it must be true that A is sufficiently uncertain concerning the probability of meeting B again, and A would incur a sufficiently large cost should A meet B again in the future, that it pays A to sacrifice now. The authors have not even attempted to show that this is the case. Nor is it plausible. The experiments under discussion assume subject anonymity,  subjects will never knowingly meet again. Pinker's supposed counter-evidence is thus invalid. To my knowledge, there is simply no credible counter-evidence.

2. Human Morality in Everyday Life

Many readers will doubtless wonder if our view of human moral behavior, which is based on controlled laboratory and field studies, extends to real life.  Consider, for one example among many, political activity in modern societies.

In large democratic elections, the selfish individual will not vote because the costs of voting are positive and significant, but the probability that one vote will alter the outcome of the election is vanishingly small. Thus the personal gain from voting is vanishingly small. The cost, however, is a significant amount of time and energy that could have been devoted to other, materially rewarding, purposes.  It follows also that a selfish individual will generally not bother to form opinions on political issues, because these opinions cannot affect the outcome of elections.

Yet people do vote, and many do expend time and energy in forming political opinions. This behavior does not conform to the selfish gene model. Of course it could be argued that we only vote to enhance our reputation as a good citizen, but since who votes is normally not public information, and one's voting history of little interest to employers and other social intimates, this is not a very plausible explanation.

It is a short step from the irrefutable logic of selfish political behavior that selfish individuals will not participate in the sort of collective actions that are responsible for the growth of representative and democratic governance, the respect for civil liberties, the rights of minorities and women in public life, and the like, that are characteristic of many modern societies. In the selfish gene model, only small groups of individuals who seek social dominance will act politically. Yet modern egalitarian political institutions are the result of such collective actions. This behavior cannot be explained by the selfish gene model.

Except for professional politicians and socially influential individuals, electoral politics is a vast morality play to which models of the selfish actor are a very poor fit.

Defenders of the selfish gene theory may respond that voters believe their votes make a difference, however untenable this belief might be under logical scrutiny. Indeed, when asked why they vote, voters' common response is that they are trying to help get one or another party elected to office. When reminded that one vote cannot make a difference, the common reply is that there are in fact close elections, where the balance is tipped in one direction or another by only a few hundred votes. When reminded that one vote will not affect even such close elections, the common repost is that "Well, if everyone thought like that, we couldn't run a democracy." Agreed. But this is just the Kantian categorical imperative, an eminently moral value. People vote because it is simply the right thing to do.

Politically active and informed citizens appear to operate on the principle that voting is both a duty and prerogative of citizenship, an altruistic act that is justified by the categorical imperative: act in conformance with the morally correct behavior for individuals in one's position, without regard to personal costs and benefits. Such mental reasoning, which has been called "shared intentionality," is implicated in many uniquely human cognitive characteristics, including cumulative culture and language. Shared intentionality rests on a fundamentally prosocial disposition.

Human beings acting in the public sphere are, then, neither avid reputation mongers nor personal gain maximizers. Rather, they are in general what Aristotle called zoon politikon—political beings. And political beings are moral beings.

3. Cultural Evolution Theory is not Just History

"Most of the groupwide traits that group selectionists try to explain," says Pinker, "are cultural rather than genetic.... Instead, they are traits that are propagated culturally... group selection ... is not a precise implementation of the theory of natural selection... Instead it is a loose metaphor, more like the struggle among kinds of tires or telephones. For this reason the term 'group selection' adds little to what we have always called 'history'." There are two misconceptions in this statement. The first is that group selectionists are for the most part uninterested in genetic evolution. The second is that the concept of cultural evolution is a simply a metaphor.

In fact, the general framework within which we work is called gene-culture coevolution. This framework is quite clearly delineated in scientific terms (see, for instance, my overview, "Gene-culture Coevolution and the Nature of Human Sociality") ² which I can summarize as follows.

First, as elucidated by Richard Lewontin in 1970, natural selection applies to any entity that follows certain rules. "The principle of natural selection as the motive force for evolution... embodies three principles." These principles are first, phenotypic variation: "Different individuals in a population have different morphologies, physiologies, and behaviors." Second, differential fitness: "Different phenotypes have different rates of survival and reproduction in different environments." And finally, fitness must be heritable: "There is a correlation between parents and offspring in the contribution of each to future generations." Note that there is nothing about genes in this account of natural selection. Indeed, Darwin himself never heard of genes when he wrote The Origin of Species by Means of Natural Selection. "It is important to note," writes Lewontin, "a certain generality in the principles. No particular mechanism of inheritance is specified... The population would evolve whether the correlation between parent and offspring arose from Mendelian, cytoplasmic, or cultural inheritance."

Second, many forms of culture do indeed follow these three principles. Different societies have different cultural norms and technologies (phenotypic variation), some cultural forms contribute more to the fitness of individuals in the societies that embrace these forms, and therefore the cultural forms themselves have varying rates of having copies in succeeding generations (differential fitness), and finally, there is a degree of faithfulness in copying cultural objects from one generation to the next (heritability).

Third, there is a common underlying unity to genetic and cultural evolution. Genes transmit information from one generation to the next, encoded in DNA, that is used by new individuals for ontological development and adaptation to the environment. Cultural forms transmit information from one generation to the next, encoded in human brains, artifacts, and documents, used by new individuals for similar purposes.

Finally, when humans develop new cultural forms, such as language, tools, lethal weapons, control of fire and cooking of edibles, the long-term effect is the transformation of the human genome itself. In other words, cultural evolution leads to genetic evolution. Examples are cooking, which led to a vast reduction in the size of the human gut, language, which led to radical changes in the human larynx and tongue as well as a considerable increase in brain size, and lethal projectile weapons, which led to changes in the morphology of hand and shoulders, as well as a reorganization of the upper torso musculature. Consider, for instance, that our closest relative, the chimpanzee, spends five hours a day digesting comparing to one for humans, lacks the ability to produce complex vocalizations, and cannot throw a stone with more than minimal accuracy or force.

Gene-culture coevolution provides a plausible scenario for the development a moral sense in humans, a quality that appears to be absent or extremely rudimentary in other species. Of course, we may never know with certainty because the paleoanthropological record is extremely scanty. However, it is clear that the development of tools, weapons, and cultural objects required cultural norms promoting cooperation. Individuals who violated these norms were doubtless punished and shunned, and hence less likely to pass their genes on to the next generation. A genetic predisposition to conform to social norms was thus likely to be biologically fit, and hence to evolve through natural selection. Insofar as social norms contribute to the fitness of the groups that embrace these norms, so will the genetic predisposition to follow these norms. We call such a predisposition a moral sense.

4. The Group Selection Pseudo-controversy

Pinker begins his discussion of the group selection issue with the following question. "Human beings live in groups, are affected by the fortunes of their groups, and sometimes make sacrifices that benefit their groups. Does this mean that the human brain has been shaped by natural selection to promote the welfare of the group in competition with other groups, even when it damages the welfare of the person and his or her kin? If so, does the theory of natural selection have to be revamped to designate 'groups' as units of selection, analogous to the role played in the theory by genes?" There are two misconceptions in the very posing of this question.

If an altruistic behavior reduces the net fitness of the altruist and his kin, it cannot evolve. The "group selectionists" argue that while the altruist may be less fit that the selfish individuals in his group, groups with many altruists will expand at the expense of groups with few or no altruists, and this expansion can more than offset the fitness loss of the altruist. Because the altruists kin are more likely to be altruists and also are more likely to be in the altruist's (this is called 'limited dispersal'), the net effect of the altruistic act may be to increase the average fitness of the altruist's kin.

The first misconception here is the view that group selection is incompatible with kin selection. It is not. Kin selection says that the fitness of an individual depends on the genes of his kin and not just his own genes. Group selection says the fitness of an individual depends on the characteristics of the group he is in, not just his own genes. The second misconception is that group selection means that the group is a "unit of selection." This is not true. Group selection occurs when the fitness of individuals may be higher in one group rather than other, depending on the social structure of the group and its and distribution of genomes.

A third misconception is that if genes are the only true replicators in evolutionary biology, and if genes are in some sense purely selfish replicators, then all biological species must ultimately sacrifice only for their close genealogical relatives. We can thus admit the complex division of labor and altruism in such eusocial species as termites and honeybees, but we must deny altruism in the case of humans, who cooperate widely with non-relatives. In fact, a careful development of gene-level fitness dynamics in a recent paper by Andy Gardner and J. J. Welsh, "A Formal Theory of the Selfish Gene" ³ ) shows that even an inclusive fitness maximizing selfish gene can support altruistic behavior in its owner.

5. The Group Selection Issue is Scientific, not Political

The group selectionists, says Pinker "have drawn normative moral and political conclusions from these scientific beliefs, such as that we should recognize the wisdom behind conservative values, like religiosity, patriotism, and puritanism, and that we should valorize a communitarian loyalty and sacrifice for the good of the group over an every-man-for-himself individualism." If this were true, it would, in my eyes, be a violation of scientific principles. Even if every known society has followed a certain practice (for instance, eating meat), and even if we can find evolutionary roots for this practice, it would not follow that this practice is morally defensible or practically desirable in our society. Indeed,, there is not a single paper or book that I authored or coauthored that drew any such political conclusions, or indeed any political conclusions at all. Nor am I aware of others who work in this tradition who have drawn such conclusions. Sociobiology is not liberal or conservative, or even middle of the road. It is just good science.

6. Conclusion

Some academic fields, including sociology, anthropology, and psychology, have no problem dealing with the fact that human beings are moral creatures, and that this morality is an important element in our success as a species. Others, including evolutionary and population biology, have had a harder time with this fact, because they have over the years developed theories that appear to show that evolutionarily fit behavior is necessarily selfish behavior. However, the moral nature of human society, and the key role of morality in our success as a species, can be accommodated without requiring evolutionary biology to abandon its cherished accomplishments. The selfish gene versus group selection issue, when properly formulated, has little to do with the nature of human sociality.

[1] Proceedings of the National Academy of Sciences 108,32  (2011):13335-13340|
[2] Proceedings of the Royal Society B 366, 2011
[3] Journal of Evolutionary Biology,24, 2011

Philosopher; University Professor, Co-Director, Center for Cognitive Studies, Tufts University; Author, Breaking the Spell

Steven Pinker sets out the problems with the current enthusiasm for group selection clearly and fairly, except for what are, in my opinion, two regrettable but easily corrected misplaced emphases.

First, while evolution by natural selection definitively and demonstrably works under conditions of utterly random (non-directed) variation, non-random variation does not prevent the algorithmic process of natural selection from grinding out its well-tested innovations, or render it superfluous, as Pinker suggests. Non-random variation typically just speeds up the process, a little or a lot. The cognitive brilliance that Steve Pinker mistakenly sees as usurping the role of natural selection in human cultural change is indeed an important factor, but one should never forget Orgel's Second Law:  evolution is cleverer than you are. Brainchildren die like flies, thousands or even millions of hopeful ventures extinguished for every innovation that truly"goes viral" and then goes on to fixation in the shared culture we pass on to our descendants. This process of differential adoption, not just by intelligent consumers, but just as much by uncomprehending (or semi-comprehending, or even heedless) hosts is "ordinary causation" in one sense—the same sense in which every birth, death and rejection in the process of genetic natural selection is ordinary causation—but it is also a clear case of natural selection of cultural items—memes, unless you find yourself allergic to that term, as some thinkers apparently are.

Fitness-reducing memes are just as possible as fitness-enhancing memes, and one of the chief virtues of adopting the meme's eye perspective is correcting the systematic blindness to this possibility in many theorists who can only see the (often imagined) benefits of established cultural institutions because they think natural selection would sweep away all toxic variants. If only! Like the common cold, many cultural items persist because they can, in spite of not being consciously and rationally accepted or adopted or promoted by the minds they infest, and in spite of having lost any redeeming features they may ever have had. They have their own fitness as replicators, largely independent of their contributions to the fitness of their hosts.

Pete Richerson's comment  articulates the details well, but muddies the water by speaking of cultural group selection. There are reasons for calling these phenomena a variety of group selection, reasons ably recounted by Boyd and Richerson in many publications, but better reasons—in my opinion—for avoiding the label, precisely because it seems to give support to the vague and misguided ideas of group selection that Pinker exposes so effectively. These phenomena consist in the evolution by natural selection (both cultural and genetic) of what might be called groupishness adaptations, dispositions (or traditions) of cooperation and the punishment of defectors, and the like, but not by a process of differential reproduction of groups. What differentially reproduce in these phenomena are groupishness memes, not groups. The establishment of these memes may then enable the genetic evolution of enhancements in hosts—like the lactose toleration that evolved in response to the culturally spread tradition of dairying. This is no more group selection than the differential reproduction of the flora in our guts is group selection.     

The second misplaced emphasis is Pinker's phrase "the illusion of design in the natural world." Richard Dawkins, in a similar vein, says "the illusion of design conjured by Darwinian natural selection is so breathtakingly powerful" in The Ancestors' Tale (p457), and elsewhere proposes to speak of "designoid" features of the natural world (eg., Climbing Mount Improbable, p4). I disagree with this policy, which can backfire badly. I recently overheard a conversation among some young people in a bar about the marvels of the nano-machinery discovered inside all cells. "When you see all those fantastic little robots working away, how can you possibly believe in evolution!" one exclaimed, and another nodded wisely. Somehow these folks had gotten the impression that evolutionary biologists thought that the intricacies and ingenuities of life's processes weren't all that wonderful. These evolution-doubters were not rednecks; they were Harvard Medical students!  They hugely underestimated the power of natural selection because they had been told by evolutionary biologists, again and again, that there is no actual design in nature, only the appearance of design. This episode strongly suggested to me that one of the themes that has been gaining ground in "common knowledge" is that evolutionary biologists are reluctant to "admit" or "acknowledge" the manifest design in nature. I recommend instead the expository policy of calling nature's marvels design, as real as any design in the universe, but just not the products of an intelligent designer. There could be a good use for "designoid," to refer to the truly only apparent design manifest, for example, in all the complicated chemical apparatus that cartoonists draw when they need to illustrate a laboratory—it looks impressive to the naïve eye, but is just a nonsensical hodgepodge of tubes, bunsen burners, retorts and the like. That is apparent design; the design in nature, in contrast, is typically as good as, or even much better than, the designs we "intelligent" artificers have yet devised. They work really well, which is as good a criterion of design as any, in my opinion.

I note the irony in the apparent tension between my two recommendations: drop the term "group" from the label for cultural selection, since it seems to give aid and comfort to the forces of darkness (the proponents of bogus forms of group selection), but maintain the word "design" in spite of the fact that it seems to give such aid and comfort to the intelligent design crowd. The situations are only superficially similar, however. The design due to human designers and the design due to natural selection  are often responsive to the same reasons and conditions (think airplane wings and birds' wings) to the point that the former are often quite dutiful copies of the latter. It may even be impossible to tell, without direct knowledge of history, whether a design feature found in some organism is due to natural or artificial selection (think cheetah and greyhound). On the other hand to call a phenomenon group selection which seldom if ever involves the differential replication (as contrasted with survival or growth) of groups seems to me to be gratuitously misleading.

Evolutionary Biologist, Washington University in St. Louis

Two Languages, One Reality

There is indeed a false allure surrounding group selection. From its inception, part of its appeal has been that it seems to tell us that nature is not as coldblooded and cruel as we feared.  Evolution can produce good behavior towards others!  Group selectionists sometimes overemphasize the good that can come of group selection.  One can almost hear the Hallelujahs.

But this false aspect of its appeal should not be taken to mean the theory is false.  Modern group selection theory is as mathematically rigorous as individual selection or inclusive fitness theory.  I say this despite being someone who favors the inclusive fitness approach and whose entire career has been based on it.   I think of these less as alternative theories that make different predictions than as two different languages describing the same world.  They simply divide up fitness in slightly different ways – inclusive fitness into effects on self versus others, and multilevel selection into between-group and within-group parts – and a simple partition of fitness should not alter predictions (1).  Inclusive fitness became popular, despite the head start enjoyed by multilevel selection thinking, because it successfully weighted the relative importance of its two fitness components, using genetic relatedness.   Without a similar set of weights, group selection advantages could not be accurately judged, and their strength and importance was often overemphasized.  I join Steven Pinker in opposing this loose sort of groupish thinking.  However, modern multilevel selection theory does have such weights, the between-group and within-group genetic variances, whose ratio happens to be relatedness of the actor to its groupmates (including itself).  Once the proper weights are accounted for, the two approaches give essentially identical results.  

Pinker is therefore correct that multilevel selection results can usually be seen as restating things we already knew in a different language . But I am loath to say that just because I speak English, others cannot speak in (as homage to Peter Kropotkin) Russian.  There is a problem here of course; we all have to learn (at least) two languages if we are to communicate with each other.  This is made more difficult by false friends, words that appear the same yet have different meanings in the two languages.  The chief troublemaker here is "individual selection", which inclusive fitness measures with respect to the whole population and multi-level selection measures with respect to the group.  Pinker confounds these in talking about group selection predicting that humans are selected to benefit their groups (Russian) at the expense of themselves (English meaning of individual).  Good group selection theory does not make that prediction, although it is true that some group selectionists have argued that way.

If the two languages are completely equivalent, we'll probably eventually converge on one of them.  But I suspect that they carry somewhat different potentials, if only psychologically.  I am with Pinker in being a big fan of inclusive fitness. It provides an agential kind of language that seems to me better for ferreting out conflicts among individuals, and I think this is one reason that it has historically generated more successful hypotheses than multi-level selection.   But real insights have also come from multilevel selection, for example on the role of population viscosity in selection for cooperation (2), and on the major evolutionary transitions (3).  I will probably continue to favor English, but maybe Russian has some useful ways of framing things that come more naturally to Russian speakers than to me.

None of this should be taken as an endorsement of E.O. Wilson's recent flip-flop towards group selection, for two main reasons.  First, Wilson makes the converse error that Pinker does – that if he describes a phenomenon in Russian, then the English account must be wrong.  Second, Wilson's Russian is itself rather garbled and the mathematical "support" from his collaborators is not group selection but pure population genetics that uses W.D. Hamilton's neighbor-modulated selection (let's call it Basque because it's hard to learn).  They in turn make the error that their laborious Basque invalidates simple English (but not, somehow, Russian).

I am not an expert in human psychology, but I can insert a useful note into the debate.  It consists of the reminder that these are empirical issues. Actual human psychology is what we need to figure out, and the theories will dutifully follow (or else die).  We need to guard against the overly strong psychological appeal of group selection, but perhaps also against a sometimes excessive relish for the hardheaded view that human nature must be selfish and conniving.  There is so much that is special about the human animal – language, culture, including cultural group selection (which we could also re-frame as cultural inclusive fitness!), shared intentionality, norms, institutions, punishment, gossip – that conclusions appropriate for other animals may not apply.   We may find that there are some truly altruistic or cooperative components of human psychology, even if those came about only because " people's cooperative intuitions have been shaped in a world in which neither anonymity nor one-shot encounters can be guaranteed" (i.e. reputation is always a potential issue).  Selection for individual selfishness can sometimes lead to cooperative outcomes, so why not cooperative components of psychologies?  These will of course be tempered by other parts of our psychology that look out for number one.  Still, even a partly cooperative human psychology would be a finding of high importance, though not one inconsistent with either group selection or inclusive fitness. 

1.              Queller DC (1992) Quantitative genetics, inclusive fitness, and group selection. Am. Nat. 139:540-558.
2.              Wilson DS, Pollock GB, Dugatkin LA (1992) Can altruism evolve in purely viscous populations? Evol.
Ecol. 6:331-341.
3.              Buss LW (1987) The Evolution of Individuality (Princeton University Press, Princeton).

Linguistic Researcher; Dean of Arts and Sciences, Bentley University; Author, Language: The Cultural Tool

In this essay Pinker convincingly argues that the idea of group selection, as popularized by leading scientists such as E.O. Wilson and Jonathan Haidt, fails to explain a number of group characteristics as effectively as standard natural selection. But in so doing, I believe that he falls prey to what Karl Popper and other philosophers of science have criticized as "essentialism," an inconsequential concern that a term can only be correctly used if all of the attributes that one writer associates with the term are found in all uses of the term. If not, the word has been used incorrectly. Pinker may be correct that natural selection was not intended to refer to group selection (the spread of alleles throughout a population because of their benefits to that population). And he therefore prefers to stick to the traditional idea that natural selection does not operate at levels higher than the gene. I am not convinced, however, that this debate encompasses the most useful range of alternatives. For example, in addition to the cases I discuss below, I would urge Pinker and others debating these issues to reconsider their blanket dismissal of Lamarckianism ideas about the genetic inheritance of acquired traits. Eva Jablonka and Marion Lamb make a strong case for a version of this idea in several stimulating works.

Before focusing on the main points where I believe that the debate over selection models has gone astray, I want to point out a potentially important example of group selection that Pinker overlooks: human languages. The function of language is to build communities and groups, as I have argued in many places. If an individual lacks the ability to talk, he or she will still survive. But a group of Homo sapiens that cannot talk will not be competitive with another group that can. Pinker might reply that this is beside the point since if an individual cannot talk then their genes will be eliminated due, perhaps, to sexual selection, so that language only appears to be group-selected. I reject such reasoning because languages are group traits. If there are genes for language, then this would support group selection—the language genes are there to ensure the survival of groups (even if only the original band of humans). On the other hand, I have also argued in many places (especially in my recent book Language: The Cultural Tool) that language is not carried on the genes—in other words, that there is no innate grammar or language instinct—so I won't press Pinker hard on this one, even though he does believe that core components of language are part of the human genotype. (One reason for rejecting language as part of the human genotype, for example, is that if this were the case, then we would expect to find populations of humans whose linguistic genotype differs from other populations', due to cultural selection, just as we find other culturally selected genetic traits in human populations, such as lactase persistence and oxygen processing at high altitudes. This would predict that some populations would not be able to learn the languages of other populations, which is false.)

But I digress. My next principal objection to Pinker's essay is that he ignores one known way in which individual traits are selected for by groups, the selection of genetic mutations by cultural pressures, also known as dual-inheritance theory. I have already mentioned two potential examples in which dual-inheritance produces results that benefit an entire group, the ability to digest milk beyond infancy and the oxygen-processing ability of some Tibetan populations. Daniel Dediu and Robert Ladd published a paper a few years back arguing for a similar culturally selected genetic change in some populations that facilitates tone perception, useful for populations of tone-language speakers. Yet another candidate is dyslexia. This likely has a genetic origin. Before the invention of writing systems, most forms of dyslexia would have been undetectable. But as the importance of literacy increases in a culture, the dyslexic will find life harder, from employment to even greater difficulty finding marriage partners. Perhaps dual-inheritance will eliminate the previously innocuous dyslexia gene from literate populations.

It thus seems to me mistaken for Pinker, Wilson, Haidt and others to quibble about the proper definition of natural selection in this context while ignoring dual-inheritance theory. Are there ways in which genes can be selected for the benefit of entire groups? Yes. Lactose tolerance is one clear example. Do the relevant selection mechanisms precisely fit the original definitions of natural selection? Not always. But Pinker's concerns here are less about science than about words. We need to get out of the dictionary and into the field, to go beyond essentialist discussions of natural selection vs. group selection and expend greater effort on documentation of what it is that we need to explain, at least among human populations

Founder, The Whole Earth Catalog; Co-founder, The Well; Co-Founder, The Long Now Foundation; Author, Whole Earth Discipline

A good critique, and helpful.

Mostly, to my eye, it helps sort out cultural "evolution" (perhaps better thought of as "progress") from biological evolution. That's a real service.

Pinker does not address genomic mechanism much. Particularly absent is anything on kin selection, the foundation of "selfish gene" theory in William Hamilton and Richard Dawkins. Ed Wilson's main attack is on kin selection—he finds it insufficiently explanatory and research-constricting.

My prediction is that new research in microbial ecology and evolution will change everything in how we think about genes and evolution. Because of the prevalence of "horizontal" gene transfer (by six different ways) in micro-organisms, they don't have Darwinian species, and their evolution looks Lamarckian—traits are acquired on the fly and passed on to offspring.

"Multi-level selection" is likely to extend downward below Darwinian species as well as upward into groups. Like everything in biology it will be messy and squishy. Simplistic Darwinian selection-by-mutation got pummeled by sexual recombination, chromosome doubling and tripling, kin selection, extended phenotype, endo-symbiosis (Margulis), regulatory genes, mitochondrial genomes, transgenic gene flow, and doubtless more to come.

The longer and closer you look, the gnarlier it gets, so far.