Archive for February, 2013

We began with a quick review of the history of the concepts of altruism and group selection. We first see these concepts pop up in Charles Darwin’s early works, but Darwin did not formalize these concepts in terms of the mechanisms or mathematics reinforcing them. Wynne-Edwards (1962) introduced the idea of inter-deme selection and the notion of “social conventions” of self-restriction, which act to restrain actions/behaviors at the individual level for the benefit of the group (e.g., Individual A forgoes reproducing so that the group size is restricted, so that the available food resources are not overexploited). It was later shown that Wynne-Edwards strongly overestimated the strength of group selection. Hamilton (1963) first formalized the idea of kin selection, suggesting that genetic reproduction matters, and the level of relatedness (or genetic similarity) impacts the indirect benefits of altruistic acts. In short, the more related individuals are, the more likely they are to be altruistic towards one another. Trivers (1971) was the first big proponent of reciprocal altruism, which involves a pairwise interaction with future benefits in mind. In reciprocal altruism, individuals actually benefit (individual selection) through future pay-back that ultimately benefits the altruistic individual. Wynne-Edwards and Hamilton both came at the idea of altruism from an indirect benefits (group/kin selection) point of view (i.e., altruistic acts lead to increases in the frequency of altruistic genes in the population), whereas Trivers’ reciprocal altruism implies that altruism results from individual selection because of individual pay-offs for the giver of altruistic acts.

This discussion of the history of these concepts led to an offshoot conversation on the differences between group- and individual-level selective pressures. This could be an issue of perspective—the larger your viewpoint (i.e., looking from meta-population level versus sub-population level), the lower the extinction rates and reproductive rates of systems in question. Because of this, lower levels of selection (i.e., at the individual) usually override selection at higher levels (i.e., at the group), when we consider multi-level selection. The group felt comfortable with multi-level selection, but note that some biologists do not believe that group selection is ever relevant in real-world cases. Here we brought up the example of virus evolution, wherein viruses evolve to reduce virulence as a consequence of inter-deme (or group) selection. Why is this a good example of group selection? Viruses must keep their hosts alive long enough to be transmitted into new hosts, otherwise the entire virus population kills itself out (i.e., extinction) by killing off hosts too quickly and therefore cutting off the possibility of transmission of virus particles to new hosts. Therefore, decreasing individual virulence increases the survival of the group of viruses as a whole.

Ok, so if biologists are so unsure about group selection, then why is altruism/group selection such a popular topic? Popular opinion is that group selection is inherent to natural systems. There is a “perfect storm” of public support and common sense knowledge that keeps the idea of group selection rooted in the literature, even though there is a “pro-math anti common-sense” backlash against group selection in the scientific community. So, biologists are concerned about the proliferation of pseudoscientific claims on group selection. Wynne-Edwards made this debate (between public opinion and scientists) worse because his models and ideas were so incredibly intractable and overstretched that they added to the idea that group selection theories belong in the realm of pseudoscience.

From this point, we revisited the concept of the Prisoner’s Dilemma, represented by the payoff matrix below:

       Individual B        Cooperate                            Defect

Individual A

Cooperate                    A = -1, B = -1                     A = -20, B = 0

Defect                         A = 0, B = -20                     A = -10, B = -10

As shown in the payoff matrix, cooperation can benefit everyone in the long run, but in the short run, individuals gain by defecting (or ‘cheating’). No matter what the other individual decides, it’s always best for the individual to defect in the short run, but if EVERYONE always defects then in the long run, average payoff is less than if everyone cooperates. Importantly, single-shot events versus repeated games lead to differences in optimal strategy. Subtleties in strategy changes are linked to differences in the rules of the game (e.g., repeat partners, repeat games, and the idea of tit-for-tat strategies). In fact, despite the presence of tit-for-tat strategists, cooperators may be maintained in a population because reciprocal altruism reduces costs of cooperating over time when cheating is avoided. Reliability of getting a repeat game or interacting with repeat partners also increases the likelihood of cooperation because of pay-off discrepancies. This process may interact with kin selection. If we think of the players as spatially static within a grid, repeat interactions occur between neighboring dots; in nature, kin remain spatially close, and those are at an increased probability of encountering repeat interactions.

At this point, we entered into an aside discussion of the “Tragedy of the Commons.” The phrase refers to the commonly shared resource of grazing fields (e.g., the Boston Commons), that are at risk of overexploitation by individual shepherds if any one individual allows their personal interest to outweigh the interest of the group (i.e., maintaining the resource by restricting herd size per individual).

The rest of the class was spent discussing six different definitions (“cases”) of altruism, summarized in the chart below (with special regard to differences in payoff (P) and cost (C) to the actor). Note that in all cases, the recipient gains a benefit greater than 0.


Defining by Payoff vs. Cost



Case 1 Operational Altruism P < C An observer sees the short-term cost to the altruist exceeding the short-term payoff to the altruist.
Case 2 True Altruism Pi < Ci[The subscript i indicates long-term inclusive fitness.] This may occur, but must be selected against through natural selection; the big question here is, DID you actually get a payoff for altruism or not, even if the payoff was unintentional. Note that in kin selection Case 1 is true but Case 2 is not. [P < C AND  Pi>Ci].
Case 3 Ethical Altruism Actor intends Pi < Ci In evolutionarily adaptive explanations of human altruism, Case 3 is often true, Case 1 is always true, and Case 2 is never true.
Case 4 Theological Altruism Combination of Case 2 and 3 Personal sacrifice for the sake of another is both intended and achieved.
Case 5 Signaled Altruism Giving signal of P < C, whether or not Pi < Ci is true or even P < C “Altruism” has been invoked in sexual selection and social selection as a costly signal.  That signal may be faked.
Case 6 Adapted Altruism P < C but, on average, for past interactions P> Ci[For any particular instance, past or present P> Cimay not be true.] Group, inter-deme, and kin selection as well as reciprocal altruism explain the rise of operational altruism in this way.  Other fully biological explanations of altruism exist, notably where it arises as a side-effect of an adaptation, or where signaling has failed (or been co-opted).

“Biological Altruism” can refer to any biological explanation of altruism.

Britt Singletary, Anna Dornhaus, Lucas Mix


Read Full Post »