Inspired by Martin Nowak’s new book “Supercooperators“, I would like to outline possible basic conditions for cooperation. First, Martin Nowak offers us five well-known mechanisms.

  • Kinship selection
  • Direct Reciprocity
  • Indirect reciprocity
  • Spatial cooperation
  • Group selection (multilevel selection)

Biologically, kin selection is by far the most important of these five mechanisms. It depends on who benefits from my cooperative efforts. The more closely related I am to the person who benefits from my altruism, the more I increase my own fitness because my genetic make-up partly overlaps. An oft-cited example is ants or bees, whose puzzlingly high rate of cooperation is perfectly explained by Hamilton’s equations. This is theoretically sound, but it is equally clear that there are many altruistic acts that cannot be explained in this way.

Things get more difficult with reciprocity (mechanism 2, direct reciprocity): “Tit for tat” is an example of an implementation of DR. It is an extremely successful strategy in plants, animals and humans because it can withstand many other strategies (it is therefore robust). This is also beautifully captured in the Golden Rule, which deals with moral issues and exists in a similar form in all world religions: “Do unto others as you would have them do unto you” (or in positive terms). The problem is that there are hardly any good examples of this outside of humans – even the famous vampire bat example of sharing food, which is also found in Nowak’s work, is now highly controversial or can be explained more simply with mechanism 1.

Direct reciprocity (mechanism 2) should not be confused with mutualism. In DR, I can always cheat / defect – due to the time lag between the altruistic action A and the subsequent action B. In mutualistic systems, this is also possible. In mutualistic systems it is also possible, but it would be stupid, since by definition both partners benefit from it, and would therefore be worse off, not better off, in the case of defection.

Indirect reciprocity (IR, mechanism 3) works through a reputation mechanism. When A does something good for B, C notices it and may pass it on to A. Trustworthy sellers on eBay benefit from their reputation, i.e. the good ratings they receive from others. This mechanism becomes particularly interesting in larger populations, where direct reciprocity can no longer be fully effective because there are too many individuals who no longer meet. Of course, IR also requires some basic cognitive equipment (recognition of interaction partners, memory, etc.).

Nowak is most interested in mechanism 4, i.e. cooperation, which works primarily because a population is spatially arranged in a certain way. This has been well demonstrated in bacteria: If cooperators can develop undisturbed, protected from too many immigrating defectors, they enjoy the many advantages of this relative isolation and the exclusive contact with “like-minded people”. Unfortunately, with a higher rate of migration, mutation or mixing (well-mixed populations), the bacterial utopia quickly becomes unstable.

Finally, the last mechanism, group selection, is viewed with scepticism by most biologists. Trivers, Hamilton, Dawkins and others point out that selection necessarily acts on the individual, not on the group or some higher unit. Nowak also disputes this. He believes that he (and some other colleagues) can show that selection can be effective at different levels (so-called multilevel selection). In my opinion, this is the weakest chapter of his otherwise exciting book.

After this rather general but necessary description of the basics, I would like to take up some of Martin Nowak’s other ideas in more detail in the next article. This fits in with a very interesting article on the experimental investigation of the spatial structure of human social networks. Fitting, because Nowak mostly does computer simulations of evolutionary processes. The problem is that with a little fiddling with the parameters, a wide range of results can be generated. Whether all the important parameters have been captured and whether they are realistic, however, is a completely different question that is usually not asked.