Paradox of Animal Sociality,
Lecture #2003-12
THE TWO SOLUTIONS TO THE PARADOX
We ended last lecture with a definition of the paradox of animal sociality. It is:
if natural selection favors only selfish entities (or entities with selfish traits), why is it that so much behavior is (or appears to be) altruistic?
As I have indicated, the terms selfish and altruistic are here understood in a special Darwinian sense. A trait is selfish if it causes the bearer to have more offspring than it would have without the trait; a trait is altruistic if it causes the bearer to have fewer offspring than it would have without the trait AND it causes associates to have more offspring. The associates may be related or unrelated, just so long as they are not direct descendants. In offering these definitions, I admitted that they had numerous problems and that the model of altruism might not, therefore, be the best for presenting the paradox of animal sociality. We will never be completely comfortable with the concept of evolutionary altruism. Just when we are beginning to think it makes sense, it will suddenly lead us to embarrassing errors of thinking. Having said all this, however, the concept of altruism is admittedly the way in which the problem is usually posed in the literature and we will continue to work with it until I can think of a better kind of model.
The paradox of animal sociality arises in the telling of Darwinian stories. A Darwinian story, you will recall, is one that accounts for the existence of some contemporary trait by imagining a history in which a species contained two kinds of organisms, one kind with the trait and one kind without, and imagining that the trait was inherited and conveyed to its bearers some sort of benefit to reproduction, so that, in the long run, trait-bearers came to characterize the species. Darwinian stories can be represented as games tables that describe the outcomes of all the interactions between and among trait-bearers and non-trait-bearers. The trait of altruism is an obvious problem for Darwinian stories. Given that altruism by definition conveys an advantage to any non-trait-bearer with whom a trait-bearer interacts, how do we tell a Darwinian Story in which altruistic trait-bearers as a group increase in number relative to non-trait-bearers?
The paradox of animal sociality underlies a great many of the issues that are raised in the literature, even issues that don't at first examination seem to relate to it. For instance, the hawk-dove games table came about because some ethologists took the fact that animals did not fight to the death as evidence for species-level altruism -- as if, in a sense, members of the same species were members of a team and had one another's interests at heart. As you already know from reading your textbook, the hawk-dove games analysis suggests a very different reason for why animals often use conventional means when they fight.
A. A Broad Outline of Solutions to the Paradox
Two fundamentally different kinds of solution to the paradox are found in the literature. The first kind of solution is to argue that the altruism is only apparent and deny that the individual performing the behavior suffers any net reproductive consequences by comparison with its alternative. Within this kind of solution are two subtypes: (a) short-term self interest that only appears to be altruism because we aren’t smart enough to see how it is immediately benefiting the health and welfare of the individual performing the behavior; (b) long term self interest that appears to be altruistic because the individual is paying an obvious short term cost to its health and welfare now (a somatic cost) but that is revealed to be selfish when we examine the long term reproductive consequences. The second broad type of solution is to concede that altruism is taking place but to argue that altruism is maintained in the population because the benefits of altruism are falling disproportionately upon other bearers of the altruist gene. Within this broad type are three subtypes, known to the literature as (a) “kin selection”, (b) “reciprocal altruism”, and (c) “trait group selection”.
Each of these two broad types of solution is exemplified dozens of times in your textbook. As you encounter explanations of apparently altruistic behavior, see if you can sort them into these two broad categories and even, hopefully, further sort them into their subtypes.
1. Apparent Altruism as Self-Interest.
Particular cases of apparent altruism can be explained away by demonstrating that the apparent altruist is actually behaving selfishly and will have more direct offspring because of his apparent altruism. This can happen in two ways. One way, to use Dawkins terminology, is to demonstrate that the survival machine itself will be healthier and happier that those that do not perform the apparently altruistic behavior. Contrary to appearance, the animal actually experiences somatic and reproductive benefits from performing the behavior. The other is to demonstrate that the survival machine itself suffers from performing the behavior, but that as a consequence of its suffering, it has more offspring. Since altruism is defined in terms of the sacrifice of reproductive benefits, rather than somatic ones, the theorist demonstrates that the donor is not giving anything away by the apparently altruistic behavior. We will consider each of these alternatives in turn.
a. Apparent altruism as short-term somatic self-interest
An argument of this form is sometimes used to explain a curious behavior performed by fleeing African antelopes called "stotting". An animal that is being chased by a predator leaps into the air, and holds its legs rigid and nearly perpendicular to the ground and its tail erect. When it lands, it may leap again in the same way, chaining together several leaps in a sequence. Many different theories have been offered of stotting, some of which presuppose that the antelope is doing something altruistic -- warning its fellow antelopes, or even sparing them by diverting the attention of the predator away from the group
But the Israeli ethologist, Amotz Zahavi, has offered a different explanation for stotting, one that suggests that it is an instance of plain-old, garden-variety selfishness. On Zahavi's account, the antelope stotts because its stotting alerts the predator to the fact that the running antelope knows that it has been seen. Since an antelope can always outrun its predator when the element of surprise is eliminated, and since predators are selected for not pursuing fruitless chases, antelopes will be selected for notifying their predators when they become aware of them, so that the predator will break of the chase and the antelope can get back to feeding. The stotting behavior is that signal, according to the theory. Thus, the stotting is not a form of altruism in any sense because it directly benefits the health of the stotter.
It's hard to imagine an explanation of this form for the curious behavior of Polistes. It seems so obvious that the helper females would be fatter and happier if they foraged for themselves, or even foraged on the eggs of the female they were helping. To hypothesize a way in which being a helper physically helps the individual female helper we might imagine that the foundress female controls some resource that the helper needs to be healthy, such as a territory that contains food and places to hide out. This sort of explanation has never been offered for Polistes, as far as I know, but it has been offered for similar behavior in Florida scrub jays. Like Polistes wasps, the young jays often don't start nests of their own but hang around and appear to help their parents raise siblings. One of the many explanations brought to bear for this odd behavior is that scrub jays live in a very spotty habitat, one where territories are very difficult to come by and widely separated. Essentially, the young scrub jays, according to this theory, are like college graduates that continue to live with their parents because they don't have enough money live on their own. Like human parents of such college students, the parents -- according to the theory -- will not tolerate their adult young in the territory unless they are helping with the raising of other young. So, it's help or starve. Thus, helping/staying is not altruism because it is selected by comparison with selfish/not staying behavior.
b. Apparent altruism as long-term reproductive self-interest.
The other way to deny that any altruism has occurred is to argue that even though the survival machine may be compromised in some way by the performance of the apparently altruistic behavior, still there were more direct offspring of the survival machine who survived because of that sacrifice. All parental behavior is of this form. An explanation of stotting that argued that the stotting animal diverts attention away from its offspring with a large benefit to them and a small risk to the parent would be an argument of this type.
Sometimes the short term cost for long term gain is born by the offspring of the traitbearer. We will encounter one famous explanation of this type when we talk about group selection a little later. Briefly, scientists had noticed that birds laid far fewer eggs (about 3 or 4 ) than they were capable of. Some theorists had offered the explanation that the birds were limiting their reproduction in order to protect the species’s food supply. The British Ornithologist David Lack offered a different explanation. He argued that that the birds were limiting the number of their eggs to the number of young that they could successfully raise. Thus, even though the number of eggs laid by an average bird was less than the bird was capable of laying, it was still the number of eggs that produced the greatest number of offspring in the long run. Lack tested his theory by artificially adding and subtracting eggs from birds nests in the wild. He was able to demonstrate that nests with greater numbers of eggs than the normal 3 or 4 actually fledged fewer young than the normal nests. Thus, he concluded that reproductive selfishness, not selection for altruism, was determining the number of eggs laid.
An explanation of this type can also be offered for Polistes behavior. Imagine, for instance, that it is the case that there are only so many good nest sites in a given locale. If you are a female who cannot get one of these sites, you have a choice between not breeding at all and helping a foundress breed. Now further imagine that there is a high predation rate and that foundresses often die. Alternatively, that egg laying is very stressful and leads to the early demise of the foundress. Under these conditions, the helper could now take over the nest , regenerate her ovaries, and begin laying her own eggs. So, the explanation would go, the reason that helpers help is that, while it offers only slight chance of reproduction, still it offers a better chance at reproduction than trying to fight a dominant foundress for her nest or leaving locale and searching for another site.
2. Discriminating altruism.
Our experiments with Darwinian Stories have already shown that indiscriminate altruism does not afford well-formed Darwinian Story because, while indiscriminating altruists give out their benefits to everybody, they take the costs of altruism only upon themselves. But what if an altruist were discriminating? What if it somehow an altruist could bias the individuals it met, so that it was selectively meeting altruists? Then any kindness done would be kindnesses to other altruists. The behavior might be altruistic, in the sense that the individual performing the behavior might suffer reproductive consequences, but it still might be possible to write a nearly well-formed Darwinian story about it because altruists as a class would end up having more offspring than non-altruists.
Whether this idea is plausible depends on the idea of selectivity. Altruists might be selective in two ways. One way (we'll call it "a") increases the relative frequency that altruists meet other altruists, thus increasing the weighting of the payoffs in the upper left hand corner of the table. The other way (we'll call it s) decreases the relative frequency with which selfish individuals meet altruists. This decreases the weighting of the payoffs in the lower left cell. Both ways increase the relative fitness of altruists, one by increasing the frequency with which altruists deliver benefits to altruists, the other by decreasing the frequency with which non-altruists get benefits from altruists.
Figure 2003-12-01Payoff received by Individuals… / Against individuals playing
Discriminating
Altruist with probability p / Selfish with probability (1 - p) / Total
Payoff
…playing … / Discriminating
Altruist / (p+a)(b-c) / (1-p-a)(-c) / (p+a)(b-c)+(1-p-a)(-c)=
pb-pc+ab-ac-c+pc+ac=
pb+ab—c
Selfish / ( p-s)(b) / (1-p)( 0) / ( p)(b)-(s)(b)+(1-p)( 0)= pb-sb
Discriminating Altruists will come to characterize the population ONLY if the total payoff to them is greater than the total payoff to Selfish individuals: i.e., when pb+ab-c > pb-sb or when the sum of the two discrimination coefficients exceeds the ratio of costs to benefits -- when (a+s) > (c/b).
When all the interactions among and between discriminating altruists and selfish individuals are considered, the evolutionary games table shows us that discriminating altruism will come to characterize the species when the sum of the two discrimination coefficients -- the increase in the proportion of the time that altruists meet altruists and the decrease in the proportion of the time that altruists meet selfish individuals -- is greater than the ratio of costs to benefits: when a+s > c/b.
This formula has two dramatic implications. The first is that the proportion of altruists in the population makes no difference to the success of discriminating altruism. This implies that altruism is as likely to increase in a population mostly full of selfish individuals as it is in a population mostly full of altruists. The second implication is that increasing the number of altruists that altruists encounter has the same effect as decreasing the number of altruists that selfish individuals encounter. There is no advantage for altruists to pursue a strategy of approaching other altruists by comparison with a strategy of avoiding selfish individuals. The two strategies work equally well.
Lets see how discriminating altruism would work in practice. Would altruism be favored if, say, if it increased the proportion of altruists encountered by altruists by .2, decreased the proportion of selfish individuals encountered by altruists by .1 and if the ratio of the costs incurred by altruists to the benefit received by their beneficiaries were 1/2? The answer is, "No!", because .2 + .1 = .3 and 1/2 = .5, and therefore a+s is not > c/b.
A Darwinian story involving discriminating altruism would claim that altruism has come to characterize the species because in the history of the species, altruists have met altruists more often than by chance and/or selfish individuals less often than by chance and because, as a consequence, the benefits of altruism have fallen so much more often upon other altruists as to compensate for the costs of altruism. . One implication of the word "discriminating" needs to be guarded against. The word suggests that the bias which assures that altruists meet other altruists more often than they should arises from the behavior of the altruists themselves. This might be the case, but it is not necessary. Any happenstance that brings altruists together non randomly, whether created by the altruists or not, makes the altruism discriminating in this sense.
Whether a Darwinian Story about discriminating altruism is plausible or not hinges, of course, upon whether circumstances can be found which would cause altruists to meet one another more often than by chance or selfish individuals less often by chance or some combination of the two. So far, sociobiologists have considered three ways in which these biases can be brought about: kin selection (in which altruists meet other altruists more often than by chance because they are meeting their kin), reciprocal altruism (in which altruists meet other altruists more often than by chance because they associate with individuals that have been nice to them), and trait-group selection (in which altruists deliver their benefits more often to fellow altruists because groups vary in the number of altruists and groups that have the most altruists deliver the most benefits to one another).
B. Conclusion.
So where have we come? I began this latest series of lectures with a restatement of the paradox of animal sociality. Animals do perform behaviors that are apparently altruistic and these apparently altruistic behaviors demand an explanation. Our analysis has shown that they can be explained in two fundamentally different ways. One way is that we can try to show that the apparently altruistic behavior is not actually altruistic in any sense: either the apparently altruistic creature isn't giving up anything at all, or the apparently altruistic creature is giving away something in the short term but getting more back in the long term. The other way is to show that behavior is genuinely altruistic, in the sense that the altruist would have more offspring if it did not perform the altruism and that other individuals have more offspring because the altruist is performing the altruism, but to show that altruists as a class are favored because altruists tend to behavior altruistically in the presence of other altruists.