Reciprocal Altruism as the Foundation of Group Selection

This is an elaboration of the second point I wish to make in my paper at the APSA this fall.  As I stated in my previous post, to make group selection work you need mechanisms that enable altruists to benefit each other and to avoid being exploited.  Otherwise, altruists who work to benefit the group at their own expense will promote the reproductive success of cheaters within the group.  Since the cheaters pay none of the costs of benefiting the group, they will proliferate at the expense of the altruists.  I argue here that another evolutionary vector for cooperative behavior can help to explain how this problem was solved. 

Reciprocal altruism is one of the basic explanations for cooperative behavior among organisms.  Altruism is here defined in terms of a sacrifice and a benefit, both measured in the coin of the probability of reproductive success.  When organism A delivers some benefit to some other organism B, at some cost to A, because it is likely that the favor will be returned, that is reciprocal altruism.  I use the term because to indicate the selection pressure that sustains the trait in both organisms. 

A paradigmatic example is blood sharing among vampire bats.  These nocturnal hunters must feed every three days to survive, leaving them at the mercy of chance as the herds of animals they prey on move around.  The bats manage the risk by a system of sharing.  If one comes back hungry, she will cozy up to another who obviously sports a full belly.  The latter will share some of her bounty because this makes it more likely that the beneficiary will share later.  There are many such examples in nature, but almost all of them involves exchanges between individuals. 

Christopher Boehm has argued in two magnificent books (Hierarchy in the Forest and Moral Origins) that human social evolution was driven by a specific problem.  Human beings have always been extraordinarily good at cooperating with their fellows.  This, more than anything, explains why we have inherited the earth.  Once we began to cooperate in hunting, gathering, etc., a problem presented.  Some members of each group (free riders) were tempted to let everyone else shoulder the burden (pay the cost in reproductive fitness) while taking their share of the bounty.  Another kind of problem is the individual who, due to physical and perhaps psychological advantage, was tempted to take more than his share of whatever was of value.  If these problems could not be solved, the evolutionary emergence of cooperation would have been precluded.  The free riders and bullies would have proliferated in the populations and the cooperators would have withered until the social unit dissolved. 

The way that this problem was managed was group enforcement.  Cheaters were sanctioned by their comrades.  Slackers could easily be marginalized.  Bullies required more strenuous interventions; however, even the biggest primate cannot stand up to the crowd and anyway, he is vulnerable while he is sleeping.  Boehm proposes that group enforcement eventually became psychologically internalized and that this is the evolutionary origin of the moral sense in human beings. 

It occurs to me that this account is a special case of reciprocal altruism.  What is special about it is that the parties are not two individuals, as in the paradigmatic cases, but the individual and the group.  The individual sacrifices the temptation to take more than his fair share.  If someone always has a bum leg when it is time for hunting or war he conserves his energy and avoids risks, the better to invest them in reproductive success.  If he tries to push his weight around, again, he is exploiting the group.  Every good citizen sacrifices such advantages to the political whole.  The group in turn has to pay the costs of enforcement, which may not be negligible if the bully is really big and the slackers are more than a few.  If the group is successful, it becomes a powerful cooperative unit. 

The social contract has long been regarded as an abstract and artificial invention of philosophers.  To the contrary, it seems to be an emergent product of human evolutionary history.  I think that there are only two possible ways to make group selection work.  One is if the groups frequently break up and reassemble.  Those groups with more cooperators than not out compete those that chance to be pregnant with cheaters.  I suspect that such a process, continually repeated, might result in the proliferation of altruists. 

I suspect, however, that every genuine case of group selection requires enforcement mechanisms.  Cheating must be suppressed if cooperation is to flourish.  This is true even if the cooperators are mostly related.  There are always black sheep in the family.  The evolution of politics is proof that the problem of group selection can be solved.  

Kin Selection as Group Selection Updated

In the paper I presented in Vancouver, there were three new ideas.  They were new at least to me!  One is that kin selection is in fact group selection: the target of selection in cases of inclusive fitness (I risk my life to save three brothers) is clearly not me.  It is instead the genetic inheritance of my family. 

The second is that in primate groups and especially in our hunter-gather ancestral groups a very sophisticated form of reciprocal altruism is presenting itself.  When the group policies the behavior of the individual members, a social contract is in force.  The parties in this case are not two individuals, as in one animal doing a favor for another when it is likely that the other will return the favor.  A group that protects its members by sanctioning bullies and free riders is enforcing a social contract between the individual and the group itself. 

The third idea is that the once the group protects each individual within it, one of the things that can be protected is the access of each individual male to his mate.  This is, effectively, the social construction of marriage.  It seems likely that the nuclear family as we understand it is a product of the emergent social contract.  This is vital because, once it happens, the group can recognize not only the bonds between couples within the group, but bonds that extend beyond the group to the wife’s family. 

Here I discuss the first idea.  Early in Human Evolutionary Psychology (Louise Barrett, Robin Dunbar, & John Lycett) group selection is briefly considered and summarily rejected.  It is only the fitness of individuals that counts in natural selection.  Almost immediately the text complicates the matter by considering the selfish gene argument.  I will argue that it is complicated in the other direction as well. 

These are the basic models for the evolution of cooperation.  Kin selection (or inclusive fitness) indicates cooperative behavior that is selected for because it benefits closely related individuals.  In the paradigmatic example, if I call out because I see a tiger and so save the lives of five brothers, I may get more of the genes that code for my familial sacrifice into the next generation than if I survive and sire children of my own. 

Reciprocal altruism indicates a tradeoff: one individual pays a cost in order to benefit another because it is likely that the cost will be returned.  When one vampire bat shares blood with another because the other be expected to reciprocate later, that benefits both parties. 

Group selection indicates cooperation on the part of a member of the group that benefits the group but not necessarily the altruist.  Consider a group of organisms bound together and living in a pond.  When they run out of oxygen, the group must rise to the surface.  Producing the substance that allows them to rise requires energy and thus involves a sacrifice.  The sacrifice only pays off if enough individuals make the sacrifice to get the group where it needs to go. 

What occurred to me as I was writing this paper is that in all these cases, the target of selection was more than one individual.  The term “target of selection” indicates the actor whose behavior or trait was selected for.  If a well-camouflaged insect avoids predators and so successfully mates and reproduces, we can say that individual was the target of selection. 

In cases of group selection, the group is by definition the target of selection.  It is the fitness of the group that is favored by natural selection; individual members benefit only in so far as they are along for the ride. 

The same is true, however, for reciprocal altruism.  Consider the case of cleaner fish and their predatory clients.  The former consumes parasites off the skin and jaws of the latter.  Each must give up something: the cleaners eat only the parasites and not the healthy tissue; the predators refrain from eating the cleaners.  What is the target of selection in this case? I submit that it is the alliance.  Together they are selected for; divided they are selected against. 

In the case of kin selection, the target is not the individual who calls out the warning, for she is more likely than the others to attract the predator’s gaze.  Nor is the target of selection any individual sibling or cousin.  According to Hamilton’s rule, cooperation among kin can be selected for when c < rb.  Here c is the cost of the cooperative act, measured in the probability of successful reproduction on the part of the cooperator.  If a certain action is likely to result in reproductive failure one out of four times you do it, then c = .25.  If reproductive failure is almost certain (I charge the lion to save my brothers and sisters) c = close to one.  How closely I am related to the individuals I am taking a risk for is r.  Siblings are .5.  How many siblings I am saving is the benefit, or b. 

This explains the paradigmatic example deployed above.  If I sacrifice my life to save three brothers, then 1 < .5 x 3.  What is key here is that the target of selection, the T whose reproductive success is being selected for, is not any individual.  The target of selection is the family. 

Two things stand out for me.  One is that kin selection seems to provide a very limited selection pressure in organisms like ourselves (as opposed to social wasps, bees, and ants).  For siblings, r = .5; but for cousins it is only 12.5.  Only if the clan were a very strong institution, so that such a sacrificial act could reliably benefit a large number of relatives, would it be able to work. 

The other is that it runs into some of the same problems that group selection models face.  In the latter, the altruists in the group enable the group to grow larger due to their sacrifices; however, non-altruists in the group benefit more since they are making no sacrifices.  Eventually, the cheaters will outnumber the altruists and the system will collapse.  This is why many theorists suppose that group selection cannot work.  To make it work, you need mechanisms that enable altruists to find each other and to exclude cheaters. 

Kin selection is supposed to be more reliable because I can be more certain who my real siblings are and so my sacrifice will promote the genes that code for my behavior.  How reliable is this?  Only once the family is institutionalized can I have a good idea who my brothers and sisters are.  If the nuclear family emerged first in the course of human social evolution, and human beings were more rather than less monogamous before they joined together in larger units, then kin selection might have been the original driving force in that social evolution. 

If, however, the family is the product rather than the cause of our evolutionary trajectory towards political animals, then kin selection is a much less significant (which is to say, not insignificant) factor in that trajectory.  Aristotle was surely correct to say that human beings are more a coupling than a political animal, since that is true of all sexually reproducing vertebrates.  Coupling, however, does not mean “familial”.  Kin selection presupposes and therefore cannot explain the emergence of a coherent family unit.  

On the human thing

I am about to propose a paper for the IPSA next year in Istanbul.  What follows are some reflections that I will distil into that proposal.  

Aristotle reflected that the various branches of philosophy often took their essential subject matter for granted.  A mathematician might never bother to ask what a number is and someone investigating physics (all motion and change, as the philosopher understood it) might never address the question whether motion and change are real.  Modern philosophy, having divorced itself from science, is almost exclusively devoted to topics which other fields take for granted. 

In the case of political science, someone studying voting behavior, for example, will probably feel little need to explain what government is and what politics is, let alone what human beings are.  Political philosophy, my racket, can and must address these questions. 

Politics in its full expression is an exclusively human activity and if the human being is the political animal, as Aristotle says, we cannot understand either the human or the political apart from one another. 

I have long thought the beginnings of Aristotle’s Nicomachean Ethics and Politics offer a clue to this puzzle.  Here is how the former opens:

πᾶσα τέχνη καὶ πᾶσα μέθοδος, ὁμοίως δὲ πρᾶξίς τε καὶ προαίρεσις, ἀγαθοῦ τινὸς ἐφίεσθαι δοκεῖ: διὸ καλῶς ἀπεφήναντο τἀγαθόν, οὗ πάντ᾽ ἐφίεται.

Every technology and every methodical inquiry, and similarly both practice and deliberate action, are regarded as aiming at some good; wherefore the good beautifully presents as that at which everything aims. 

I often ask my students in ancient political philosophy to tell me what “the good” is.  They are always stumped.  Aristotle makes it clear.  The good is some goal that explains some deliberate human activity. 

It is important to note that these activities‑technique, method, practice, and action‑are activities of individual persons.  In the most primary sense, politicians politic and deliberate; regimes do so only in a secondary if not a metaphorical sense.  If we take the Nicomachean Ethics to be an account of the human being, and it is certainly at least that, then it would seem that the human being is the natural person, the individual. 

Here is how the Politics begins:

ἐπειδὴ πᾶσαν πόλιν ὁρῶμεν κοινωνίαν τινὰ οὖσαν καὶ πᾶσαν κοινωνίαν ἀγαθοῦ τινος ἕνεκεν συνεστηκυῖαν （τοῦ γὰρ εἶναι δοκοῦντος ἀγαθοῦ χάριν πάντα πράττουσι πάντες）, δῆλον ὡς πᾶσαι μὲν ἀγαθοῦ τινος στοχάζονται, μάλιστα δὲ [5] καὶ τοῦ κυριωτάτου πάντων ἡ πασῶν κυριωτάτη καὶ πάσας περιέχουσα τὰς ἄλλας.  αὕτη δ᾽ ἐστὶν ἡ καλουμένη πόλις καὶ ἡ κοινωνία ἡ πολιτική.

Since we see every polis to be some community and every community is established for the sake of some good (for the good is regarded to be that for the sake of which everyone does everything) as it is clear that as all communities aim at some good, the one that aims especially at the most authoritative good is the most authoritative of all of them and embraces all the others.  This is called the polis, is the political community. 

If the Politics is an account of the human being, and it is surely at least that, then it would seem that the human being is the assembly of natural persons, self-organized in a political way. 

The two points of departure point to one idea.  The human being is not the individual nor is it the social group; it is instead the dynamic that involves both of them.  A person can live apart from other persons, as do hermits; however, to the extent that he lives off the culture baggage that he carries (does he take books with him into the woods?) he is not really alone and to the extent that he does not the life he lives is more animal than human.  If you don’t believe me, read Rousseau’s Second Discourse. 

Likewise, human beings can attempt to integrate themselves and others into a social whole so completely that their individuality disappears.  This latter trajectory is more limited than the former.  It is possible for one human being to become completely feral; it is not possible (at least not yet) for two or more human animals to become one human being.  Only the most awesome force can suppress human individuality and that only so long and so thoroughly as the force is applied.

Biopolitical theory, based on Darwinian evolution, can turn the two dimensional spectrum defined by the poles of human persons and human communities into a three dimensional and very real structure reaching backward into time.  Evolution is not goal-directed.  In so far as it has a direction, it teleomatic rather than teleonomic.  It is a mechanical force, pushing organic forms into new environmental niches.  It is not trying to do anything anymore than a balloon is trying to get high.  At any one point in time a biological lineage can branch toward simpler organisms or more complex ones or both.  That one or more branches moved into new niches by the emergence of more and more complex organisms is the reason that human beings (and bovine beings and canine beings, etc.) exist on this planet. 

Single celled organisms can combine into multicellular organisms and eventually their individuality can be entirely (or almost entirely) submerged into the whole.  Multicellular organisms can go the other way or parts of them can go their own way (viruses?).  Some, as the scarecrow says, do go both ways, as in the case of slime molds.  Multicellular individual organisms can also coalesce with other individuals into larger social wholes, though here complete assimilation is more challenging.  Eusocial insects combine into colonies and hives that act more or less like biological individuals.  A sterile cast of ant workers is a sign that individuality has been reproductively submerged in a social whole.  For the most part, social animals are more individual than social.  Sociobiology is all about the ways that individual interests are managed in a way that maintains social integration. 

Human beings are capable of a degree of social integration that is greater than anything other than the eusocial insects and is much deeper than that.  We are almost certainly the only animal that can try to submerge its individual self in a larger whole, only to fail.  I propose that this is the temporal dimension of Aristotle’s great dichotomy.  I also suspect that group selection is the key to understanding this existential dimension of the human being.  That is what I will present in Istanbul, if my paper is accepted.  

How Group Selection Works

To say that a human being must breathe, eat, and defecate, all on a regular basis, in order or write poetry, does not reduce poetry to respiration and digestion; it simply reminds us that poets are biological beings.  Biological processes such as these may rarely be relevant to the interpretation of poetry but they are constantly relevant to an historical interpretation of households and cities.  Caesar cannot recline and enjoy his feasts if food does not come in and poop go out and neither can Rome. 

If you want to understand the idea of justice, as Plato’s Socrates sets out to do in the Republic, you will sooner or later have to account for the existence of human souls and human communities and that cannot be accomplished without evolutionary theory.  My position (big surprise) is that the latter completes Socrates’ account. 

I have been reading David Sloan Wilson’s Does Altruism Exist?  It reminded me of a fairy tale I composed some time ago to illustrate group selection.  I offer it here. 

Consider a group of imaginary primates living and hunting in small groups in an area covered by tall grass.  I’ll call them puds.  Some of the primates in each group are taller than the grass.  Others are smaller than the grass. 

Now consider that this population of puds is subject to two selection pressures.  The puds are preyed upon by a large avian animal.  I’ll call it a dragon.  Dragons cannot spot a single tall pud but can easily spot more than a handful when the group is moving.  When a dragon descends on a pud group, however, it can feed on large and small with equal ease.  Groups with a lot of tall puds are at a disadvantage, so that is selection against those groups. 

Within each group, tall puds have an advantage.  They are stronger and can forcibly mate with more females than their shorter male pud compatriots.  So in any group with both tall and short males, the taller will proliferate over time.  So that is selection against small puds. 

If the groups remain isolated from one another, the species is on the road to extinction.  Every group will eventually have enough basketball stars to attract a dragon and so every group will be eaten and the dragons will go on disability.

What will save the puds is that some of them will survive a dragon attack.  The survivors will find other survivors and form new groups.  Groups with more short pud fellows will survive longer and grow larger, increasing the supply of small puds in the total population.  If, however, every group has at least a few tall guys, the cycle will continue. 

One thing that might alter the trajectory is if the puds are smart enough to figure this out.  A bunch of small puds might realize that the two tall puds who made it to the founding council are a problem and vote them off of the island.  This rule, whether a product of pud culture or a slowly evolving disposition, would tilt the playing field in favor of puds against dragons.  How bad it would be for the latter depends on how tall genes are distributed in the population. 

I thought this up before I knew a lot about the research into group selection.  I intended it only as a demonstration that group selection was logically possible.  D.S. Wilson’s book gave me an example that neatly matches my fairy tale.  Pond skaters (Gerridae) are insects that skate across water in search of prey.  The males come in two flavors: rapists and gentlemen.  The former force themselves on any female they meet.  The latter wait until a female approaches to mate. 

Within groups of skaters, the rapists have an advantage.  They mate more often and produce more offspring.  So their sons proliferate in the group.  However, their appalling behavior doesn’t leave the females with enough time to feed, so they lay far fewer eggs than they would if they were well fed.  The result is that groups with more gentleman grow faster than those with more rapists. 

What keeps the gentleman in business is that females have some choice.  When groups are forming, females are more likely to join a group with more gentlemen.  That simple, biologically trigger preference, rewards chivalry with the precious coin of reproductive success. 

This explains so much.  It explains why cooperation is so difficult and how it is possible.  It explains why rules of justice are necessary among human beings and other social organisms and how justice is possible. 

Something more important now occurs to me.  The balance between group selection and individual selection can tilt either way.  When it tilts toward the group, eventually you get the assimilation of individual organisms into a new whole, as when a multi-cellular animal emerges from a cooperation of cells.  When it tilts the other way, the community with its social interactions disintegrates into individual competitors. 

Human beings are social animals but we are also individuals.  The entire field of justice and morality emerges from this fact.  All human activity, including poetry and politics, is possible because within group selection and between group selection balanced out, over a long period of our life on this earth.  

Group Selection and the Evolution of Morality

Perhaps the best single article I have read on the group selection debate is “Evolution ‘for the Good of the Group’”, in American Scientist, September-October 2008.  It is another collaboration between the two Wilsons, E. O. and D.S.  I would add that it is also a very good introduction to the general question of the levels of selection‑genes, individuals, and groups, within a population. 

Group selection is one explanation for the evolution of altruism.  Any time one organism (or any unit within an organism) behaves in such a way as to confer a reproductive advantage to another organism at its own expense, this is evolutionary altruism.  Honey bee workers who serve the queen but do not themselves reproduce are behaving altruistically.  A vampire bat who regurgitates some hard won blood to feed a hungry roost mate is another example.  Many examples of altruism are easily explained in terms of deferred gratification (reciprocity) or benefit to closely related individuals (kin selection). 

Group selection theory is based on the claim that some altruistic behaviors are selected for because they benefit the group without any return to the altruist whether direct (deferred gratification) or indirect (kin selection).  A group with more altruists will be more reproductively successful than a group with fewer and so altruists may increase in the total population, at least initially.  Increase in the total population is what we mean by evolution. 

There seems to be an insuperable problem.  While between group selection might well favor altruistically endowed groups, within group selection will favor the selfish over the public spirited organisms.  Altruists would seem to be doomed to inevitable extinction as their selfish fellows outbreed them.  In this view, which held the field for a long time, group selection is unsustainable. 

However, group selection does in fact occur.  Wilson and Wilson present a number of forceful examples.  My favorite is the bacterium Pseudomonas fluorescens.

When this species is cultured in an unstirred broth, the cells soon consume most of the oxygen in the bulk of the medium, so only a thin layer near the surface remains habitable. A spontaneous mutation called wrinkly spreader causes cells to secrete a cellulosic polymer that forms a mat and helps them colonize the water surface. Production of the polymer is metabolically expensive, which means that nonproducing “cheaters” have the highest relative fitness within the mat; they get the benefit of the mat without contributing to its upkeep. However, if the proportion of cheaters grows too high, they are undone by their own success. The mat disintegrates, and the entire group sinks into the anoxic broth. Experiments by Paul B. Rainey and Katrina Rainey have shown that the wrinkly spreader trait is maintained in the population by group selection, even though it is disadvantageous within any one group.

This example illustrates the fact that the “free rider problem” is real.  The benefits of altruism in between group selection can indeed be undone when selfish cheaters crowed out the altruists.  At the same time, the very fact that mats form at all demonstrates that group selection was a powerful force in the evolution of this microbe.  Wrinkly spreader can only be maintained by its benefit to the community. 

Obviously, what is needed to maintain group selection is some mechanism for suppressing cheating.  I have no idea how this is done by bacteria but Christopher Boehm has a good idea how it is done among human hunter gatherers.  He argues in Moral Origins that social selection (reproductive benefits that result from a reputation for altruistic behavior) and sanctions against bullies (free riders) functioned to protect altruists from cheaters. 

Human beings are extraordinarily capable of altruism toward unrelated individuals.  Explaining this is a big challenge for evolutionary theory.  Boehm considers a number of explanations that are current in the scholarship.  He doesn’t reject them, but argues that some of them work only when cheating is suppressed by the mechanisms mentioned above. 

In Plato’s Gorgias, Socrates argues that justice is analogous to medicine: it is a response to dysfunction in the social body.  I am inclined to think that the theory of group selection is beginning to uncover something like the Platonic idea of justice.  It may be that retribution is something that shapes all life on earth. 

The Two Wilsons on Group Selection

In connection with the paper I have been writing, I have been thinking a lot about group selection.  I have incorporated this quote from E. O. Wilson’s The Social Conquest of the Earth:

An iron rule exists in genetic social evolution.  It is that selfish individuals beat altruistic individuals, while groups of altruists beat groups of selfish individuals.  The victory can never be complete; the balance of selection pressure cannot move to either extreme

This strikes me as a very powerful formulation.  It includes the fundamental problem presented by the evolution of morality.  Here is how David Sloan Wilson and Edward O. Wilson put it in RETHINKING THE THEORETICAL FOUNDATION OF SOCIOBIOLOGY (Quarterly Review of Biology, December 2007).  

During evolution by natural selection, a heritable trait that increases the fitness of others in a group (or the group at a whole) at the expense of the individual possessing the trait will decline in frequency within the group. This is the fundamental problem that Darwin identified for traits associated with human morality, and it applies with equal force to group-advantageous traits in other species. It is simply a fact of social life that individuals must do things for each other to function successfully as a group, and that these actions usually do not maximize their relative fitness within the group.

The question here is how to explain the emergence of genuinely altruistic behaviors in human beings and other animals.  What separates the Mensheviks from the Bolsheviks in contemporary evolutionary theory is that the former try to explain it away whereas the latter try to explain how it can be what it seems to be. 

Explaining it away means interpreting apparently altruistic behavior as selfish behavior, keeping the rule that any behavior selected for must advance the genetic interests of the individual within the group.  If I share my kill with the rest of the tribe, that is because big hunters get lots of pussy. 

Explaining altruistic behavior while accepting that it is, genuinely, altruistic, may require group selection.  Cooperation is a trait not of individuals but of partnerships.  If you and I make sacrifices for each other, our business model turns a profit.  Groups with a lot of altruists out compete those with few, resulting in an increase of altruists in the larger population. 

I suppose that both approaches have value, but that puts me squarely on the side of the Bolsheviks.  The Mensheviks will allow only the selfish interpretation. 

The other thing that is presented in the E. O. Wilson quote is that group selection best makes sense of the moral dimension in human beings.  If group selection works, it will work against directly selfish behavior and the resulting schema will produce a being at odds with himself.  That would be me and you and all of us human beings. 

I also note that E. O. Wilson’s quote contains the paradox that we find in the prisoner’s dilemma.  In that diabolical issue of game theory, it makes perfect sense for each of us to defect in our partnership; nonetheless, we would both be better off if we cooperated.  I think that the Wilson’s are on to something.