David Sloan Wilson’s essay Human Cultures are Primarily Adaptive at the Group Level is helpful in calling attention to the fundamental role that the human social group has played throughout our evolutionary history. But Sloan Wilson is mistaken, in my view, in seeming to use the phrase “primarily adaptive at the group level” to mean that humans have acquired a suite of social and psychological dispositions for promoting the fitness of their groups even if it means suffering a cost to their own fitness.
There is agreement that there is something important to explain. Human beings are distinguished among all animals for having forms of cooperation that, on the face of it, pose a challenge to conventional Darwinian accounts of evolution. We routinely help others, we share our knowledge and skills, we give up seats on trains, pay taxes, hold doors for people, give money to charities, and even sometimes risk our health and well-being to fight wars.
The challenge to Darwinism is to explain how such apparent altruism can evolve when there are people who are only too happy to benefit from your aid but have no intention of returning it. This question is often answered by drawing on an analogy to the social insects – the ants, bees, wasps and termites – or even to the skin cells in your body.
Individual ants quite willingly, indeed sometimes enthusiastically, go to their deaths in support of their queen, and skin cells in your body do not have to be coaxed into giving their lives to protect you from the harmful rays of the sun. These actors’ high genetic relatedness to each other makes selfless altruism a good strategy for promoting copies of their genes that reside in others.
But humans are different – the multiple actors in the great ‘bodies’ we call our societies are not like cells in a body, nor even like a colony of ants. Indeed, the wonder of human cooperation is that we somehow manage to make our style of altruism work even among non-relatives. We have moved beyond the mere eusociality of the social insects to an ultrasociality, this term acknowledging that we cannot explain our actions as strategies for promoting copies of our genes in relatives.
Sloan Wilson is one of the leading proponents of ‘group selection’ as a way to explain this fascinating dilemma of human behaviour. The idea is that our groups have been as important to our survival and well-being throughout our evolutionary history as an ant’s colony has been to it. As a consequence we have been moulded by natural selection to do things that advance our ‘colonies’ even if it means suffering a cost to our individual fitness.
Our groups have been important to our success as a species – a point that is difficult to overemphasize. But have we really evolved to be willingly subservient to them? Is this what it feels like to you to be a human? This is really one of the most fundamental questions we can ask of our nature. Are we fundamentally ‘good’ or are we fundamentally self-interested?
In seeking answers to these questions, it is vital not to fall into the trap of assuming that what appears to be selfless group-level altruism and cooperation can only emerge from the sort of group-selection Sloan Wilson envisages. There can be selection between groups but this need not imply ‘group selection’ in the sense that Sloan Wilson uses it. In my book Wired for Culture, I discuss several alternative propositions that explain how apparent group-level altruism can evolve because it has returned individual benefits, not group ones at the expense of individuals.
In the simplest case, imagine you inhabit a group of two and that by helping each other you can achieve more than twice as much as the two of you working alone. Now ramp this scenario up to a larger group. The behaviours might range from cooperation in acquiring food to fighting battles with other groups. The returns from cooperation mean that what looks like altruism, is really a form of enlightened self-interest. Selfishness can never be widespread because groups with lots of selfish players cannot compete against cooperative ones.
Another scenario I call ‘enhancement selection’ and it seems to capture some of the more psychologically and socially nuanced, charming and puzzling aspects of our behaviour. What if, by virtue of being useful to the group in some way, you enhance your reputation, becoming widely known as the kind of person others like to have around. Perhaps you are a good hunter willing to share the meat you bring back, a good warrior, or you might simply be good at making arrows or navigating on the open seas. If as a consequence of your good deeds you attract kindnesses from others, then your apparent altruism can be more than repaid, and what looks like altruism is really in your best interest.
The appeal of this scenario is that once ‘altruism’ becomes a way of acquiring a good reputation, the problem of how to get altruism to evolve is subverted: if being altruistic attracts altruism from others, then people will actually compete to cooperate. In fact, we will become altruism ‘show-offs’, falling all over ourselves in an attempt to convince others of our worth. Our ultrasocial nature becomes the altruistic equivalent of a peacock’s tail, except where the peacock uses his tail to acquire a mate, we use our altruism to secure the spoils of cooperation.
This can explain the peculiar and repeated acts of altruism that most of us display throughout a typical day. All those doors we hold open, seats we give up, coins we drop into charity boxes and cats we rescue from trees are ways that we display our ‘long tails’ of altruism. The wonderful irony is that, as a self-interested tactic, this kind of ‘altruism’ can happily take its seat alongside all the other self-interested things we do, like cheat on taxes, exceed speed limits, lie to people, or pay huge sums of money to have our children educated. It is hard to explain these in the ‘nice’ world of Sloan Wilson’s style of group selection – you certainly wouldn’t see ants behaving like this.
So, yes, many of our distinctly human traits are adaptations that make it more likely our groups will be successful, but they need not be ‘adaptive at the group level’ in the sense that I think Sloan Wilson has in mind here. There might be an emerging consensus about the former, but certainly not about the latter, and it would be mischievous to suggest otherwise.
References
Pagel, M. 2012. Wired for Culture: Origins of the Human Social Mind. W.W. Norton, USA and Penguin Press, UK.
Pagel, M. 2012. Adapted to Culture. Nature, 482, 297-299.
I agree that Wilson is a bit off the wall–I never did buy his group selection–but the idea that human altruism is all self-interested hypocritical show-off is a total failure. It does not explain why anyone would value that in the first place. There has to be a value on male health to make male crickets compete to display health by chirping for hours. There has to be a value on singing (for whatever reason) for male birds to compete in song. Similarly, people obviously value altruism and want it in others, hence the need to show it off. More to the point, though, every philosopher who actually looks at human behavior recognizes that most acts of altruism are done because they feel good, not because they get you anything. However, most of those acts are done for your children, close kin, or people in a kin-like role (which now includes professional colleagues!)–so I remain fairly open to, if not downright in favor of, the old-fashioned kin selection idea of Hamilton. Showoff altruism seems done only to impress strangers or potential mates. Mothers don’t take care of their kids just to show off (though that is not to say they never show it off). So, Wilson out, altruism in, and kin-selection extended widely as still the most likely explanation.
There is no question that multilevel selection has ‘hijacked’ kinship feelings and terminology. So a ‘band of brothers’, ‘brothers in arms’, blood brotherhood etc denote relationships between non-relatives. But it is not kin selection in the Darwinian sense.
Furthermore, there are many situations when kin selection and group selection are acting against each other, and it is not a foregone conclusion that kin selection would always win. My favorite story is about one Ottoman sultan, who had to exterminate all his brothers upon acceding to the throne. He had to personally strangle his beloved younger brother, tears streaming down his face. But the raison d’etat prevailed over brotherly feelings.
“Band of brothers” and “brothers in arms” are strongly associated with cultural kin selection. Kin selection in the “Darwinian sense” is an ambiguous phrase, since most of the kin selection ideas didn’t come along until long after Darwin. However, it’s *essentially* the same as kin selection, complete with Hamilton’s rule, parental care, kin recognition, kin competition, etc, only applying to memes – rather than genes.
As for the idea that “the kin selection model is merely a special case of the multilevel selection theory” – that turned out to be a fallacy. The same with “there are many situations when kin selection and group selection are acting against each other”. These things are widely considered to be essentially the same thing. There can be struggles – but not between distinct forces associated with kin and group selection.
I think the using the terms “altruism” and “selfish” leads to much of the confusion within the field and with its impact outside the field.
Humans are complex beings with good memories, language and great fraud detection mechanisms. We evolved in small bands of only partially related individuals, where displays of cooperativeness were essential. The reason as Mark clarifies is that we are a species which can achieve substantially more in collaboration with others than alone. Thus we want to portray clear status as a good cooperator so we will be selected for cooperation and not lose the support of potential spouses and band members. The best way to project cooperativeness among people that we live with 24 hours a day our entire lives is to actually be reasonably cooperative. Add gossip into the mix, a positive sum process where people exchange valuable information while bonding, and free riders, exploiters and cheaters are pretty easily rooted out of a band.
Boehm has written extensively about the danger with being rooted out as a non cooperator. You risk not just lost cooperative opportunities, but outright expulsion or abandonment as well as capital punishment. When the band knows you are cheater, can profitably tell other bands you are a cheater, and can abandon you, it is best to really be cooperative.
But being a good cooperator is neither “altruistic” nor “selfish” in common usage of the words. The winning strategy is to establish a reputation as a valuable member of society by being a valuable member of society. Human societies can form positive sum institutional arrangements, and trying to describe it in a zero sum selfish/altruist terminology just gets in the way.
Much of disagreement between Mark Pagel and David Wilson is semantic in nature. Multilevel selection is part of ‘conventional’ Darwinism – in fact, the first proponent of it was Darwin himself. Group selection is a form of ‘enlightened’ individual selection – by behaving in prosocial manner, an individual is favoring the spread of its genes, because its group will outcompete others.
Since I am dwelling on semantic issues, I want to reiterate two definitional points that have been discussed at the SEF by me and others. First, it is much more useful to talk about multilevel selection, rather than group selection, because the first emphasizes that selection can act at different levels of social organization simultaneously, rather than creating an artifical distinction between ‘individual’ and ‘group.’ Second, ‘altruism’ is not really what is at stake. Altruism means helping others, but the really interesting question is how human societies evolve. Jon Haidt proposes that we use the term ‘groupishness,’ which I initially did not like, but now I came over to his point of view. So all humans behave in ways that combine selfishness with groupishness in different mixtures and in different settings (except for sociopaths who are completely selfish). What is interesting is explaining why and under what conditions we sometimes behave more selfishly, and in other settings more groupishly.
I don’t have Mark’s book at hand, because I am traveling, so I can’t reproduce his exact wording, but he several times writes about groups as collective survival vehicles. That’s group selection!
Joining groups because it is good for you is most frequently called the “Allee effect”. Invoking Warder Clyde Allee is a nice way of paying homage to the history of the idea – and I think people should do it more often.
Reputations and virtue signalling are only part of the story of human cooperation. Cultural kin selection (kin selection on memes) is a huge deal, explaining why soldiers in battle sacrifice themselves for their “brothers” at the request of father-life figures, why teachers invest so much in their students, why school and hospital uniforms exist, and much about our social and political lives. Anyone who thinks that kin selection doesn’t apply to memes is wrong – they need to think that one through again.
Cultural group selection attempts to explain similar phenomena using similar mechanisms – though it is surrounded by “group selection muddle” – which broadly mirrors the situation with kin vs group selection in mainstream biology. However there, kin selection has the upper hand, whereas the evolutionary social sciences are dominated by group selection. That is a lamentable situation, and the cause of awful confusion surrounding the very species we care most about.
While reputation management is undoubtedly an important part of what makes human groups so cooperative, it seems insufficient in itself to explain human “ultrasociality.” For one thing, it’s not clear from Pagel’s account here how reputation management alone would overcome the free rider problem. Indeed, at ever larger scales of social organization reputation management would seem to become more vulnerable to free riders as interactions with anonymous others increase and monitoring reputation becomes increasingly more difficult.
As societies become larger and more complex, we need to supplement earlier reputation systems with more institutional solutions. Credit reports, user rating scores, brands, money back guarantees and so forth. In addition, we can supplement these with positive sum divisions of labor. People learn to project the ability to display that cheaters will be punished by investing in the salary of a person assigned the role of enforcer or sheriff. We then just cooperate with those agreeing to fund the sheriff and the judge. In other words, societies can form institutional arrangements which are positive sum and which protect against free riding and exploitation.
And yet these systems, too, are vulnerable to exploitation by free riders so it’s hard to see how they could arise on their own without the pressure of between group competition.
But once our late Pleistocene ancestors developed the capacity for cultural problem solving and prosocial behavior, then the arms race of cheating and detection could shift to the cultural playing field. I certainly agree with Boyd and Richerson that cultural selection can operate on the level of groups. I am less confident that biological selection needs to be group based. Certainly I see it as possible, but not really necessary.
The idea that the cultural realm is somehow different from the organic realm with regards to group selection is a conceptual mess. Nobody would claim kin selection applies to the cultural realm and not the organic realm. Yet we see the corresponding claim repeatedly from group selection advocates – e.g. see Boyd and Richerson’s 2005 book. This idea is wrong: both ideas are applicable in both domains.
I agree with what you have written on the two realms. They share key underlying principles, but also differ in important ways.
Group selection seems to be an area which is much more significant in the cultural realm, especially in regards to the area of cultural solutions which we would call institutions and protocols. They are collective group phenomena that resist infiltration by alternative institutional solution sets. It doesn’t work to drive on the left side of the street when everyone else drives on the right, and it doesn’t work to speak Hawaiian when everyone else speaks French. Organic group selection may be possible, but it doesn’t seem as important in explaining emergent patterns.
This, from the “Adapted to Culture” article illustrates where the problem lies:
“In the rest of the animal kingdom, cooperation is generally confined to helping relatives. The theory of kin selection explains why: actions that support your relatives benefit copies of your genes. But this theory is mute in the face of the human propensity to help strangers.”
That’s not really correct. Shared *memes* can influence even total strangers to help each other – for example if they are wearing a company uniform, a religious gown, or a military outfit. The process is known as cultural kin selection. The process is kin recognition applied to memes. The theory of kin selection is quite capable of explaining this type of behaviour – you just have to apply it using units of inheritance which are applicable to the situation.
This might sound like a non-sequiter and maybe I am missing something but,
I cant help but think to myself that part of this whole debate stems from the fact there has to be individual genetic differences in the amount of ‘other oriented’ prosocial behavior expressed in people. So why don’t we look at this whole debate as having some kind of self selecting process to determine which side you stand on. Some people really do phenomenologically feel group adapted and internalize a general ethos of prosocial behavior, while for others prosocial behavior is far more contingent on context.
Lets just imagine that cultural group selection has operated on our species but that it is only sustained through punishment of social norms (ie. Boyd & Richerson). The result would result a scenario that our prosociality is determined by context (ie, how many people are watching) and the extent to which the person has internalized prosocial norms. I hope that we can agree that there is going to be a genetically based differences in the rate at which people internalize norms. For some people, prosocial norms al. la. religious morality is deeply embedded in their world views and results in generalized prosociality even when no one is looking. While for other people, we just follow the norms because want to avoid punishment. I would hazard that it is an empirical fact that these two strategies exist in the population and thus at some point in our history they could have been at some kind of equilibrium.
So if the question is, do we really feel that we are group adapted, I bet you some people really do, while for others more we are more contingent and contextual with our prosociality. These differences may just be at the heart of this debate.
The discussion is likely to get into “ad hominen” territory if anti-social traits are attributed to the supporters of one side of the argument or the other. The usual advice in technical discussions is to stick to addressing the arguments – and leave the personality traits of those involved in the discussion out of it.
Thanks to Pagel and the other respondents. I have devoted a considerable part of my general reply to Pagel’s commentary and will elaborate here.
All aspects of Pagel’s commentary–and also his book Wired for Culture–need to be subjected to the equivalence test. You can’t evaluate multilevel selection theory unless you compare selection differentials within and among groups. Pagel appears unwilling or unable to do this. Yet, his examples include sufficient information for the reader to make the appropriate comparisons. When this is done, Pagel turns out to be a group selectionist. In other words, most of the prosocial traits that he posits do not evolve by virtue of benefitting individuals, relative to other individuals within the same group. They require the differential contribution of groups to the total population.
Pagel frames his ideas in terms of selfish gene/meme theory by calling cultures “group level vehicles of survival”. The concept of “vehicles” within selfish gene/meme theory crudely maps onto the concept of units of selection in multilevel selection theory. For example, Dawkins would say that group selection doesn’t work because groups do not qualify as vehicles of selection. In my frequent jousts with group selection critics such as Jerry Coyne, this is treated as common knowledge–yet Pagel wants to describe cultures as vehicles while disavowing any connection with group selection, thereby parting company with Coyne and Dawkins. This kind of inconsistency is inexcusable in an intellectual environment where multiple equivalent theoretical frameworks are being employed. I and most other multilevel selectionists are adept at translating among equivalent theoretical frameworks such as selfish gene theory, inclusive fitness theory, and multilevel selection theory. If Pagel insists on speaking only one language, then he is either ignoring or misunderstanding a large fraction of the scientific literature relevant to his thesis.
The modern literature is replete with articles on how group selection can be a strong force in human genetic and cultural evolution without resulting in ant-like eusociality. A better comparison is with the rules of meiosis, which (largely) prevents the differential fitness of genes within individuals, thereby causing the individual to be the primary unit of selection. Pagel’s suggestion that humans aren’t group selected because they don’t resemble eusocial insect colonies ignores this literature entirely.
Dawkins *does* talk about “vehicle selection” and “units of selection”. However, I *suspect* that “vehicles of selection” is a term which is the result of these ideas having sex in the minds of his critics. If you search you will probably see what I mean.
FWIW, my preferred terminology to describe a collection of phenotypes that tends to die at the same time is “target of selection” – following on from Mayr’s 1997 essay.