Multilevel selection – the idea that evolution acts at several organisational levels – is a controversial idea that has been put forward to explain the evolution of behaviours like altruism (McAndrew, 2002). Although theoretically plausible, the controversy rages over whether it is of any practical importance, given that it will only work under restricted circumstances (large numbers of small groups that do not interbreed, Nowak 2006). David S Wilson, author of Darwin’s Cathedral, is a champion of the idea that group selection can explain the evolution of religious behaviour.
Rick O’Gorman, in a paper co-authored by Wilson, has taken a look at the implications of multilevel selection for evolution of human behaviour (O’Gorman 2008). The paper looks at some of the evidence for this in the interactions between ‘intrinsics’ (people who value intimacy, community and personal growth) and extrinsics (people who value money, beauty and popularity). In head to head competition in economic games, extrinsics tend to outcompete intrinsics because they’re more likely to act selfishly to get what they want. But when you put people together in groups of different compositions, and make the groups compete, the result’s more complex. Groups with high numbers of intrinsics do better than those with high numbers of extrinsics, but within a group, the extrinsics still win out.
By itself, this doesn’t mean that nice guys win. Within each group, the extrinsics are more successful, so they would reproduce faster (especially if the groups are not reproductively isolated). If group-selection did drive altruism, intrinsics would need to evolve ways to detect and punish these ‘free-riders’ – preferably by excluding them from the group.
So the authors review the evidence to show that this is exactly what happens:
… there is an extensive body of research showing that humans are willing to punish free-riders in public-goods situations. Moreover, it appears that humans will do so, even when the act of punishing is itself costly. Such altruistic punishment is evoked in controlled lab studies, where participants are anonymous to each other and do not interact more than once with any other participant, avoiding the possibility for reputation to be developed and for signals of future intent.
They go on to talk about the power of gossip as a tool to identify and share information about free riders, and the fear that people have of being gossiped about or ostracised.
They also look at the evidence for ‘group-functional behaviour’ – in other words, evidence that humans have evolved for optimal performance in a group. For example, there is some evidence that groups can make better decisions than the same numbers of individuals. The ground here is decidedly dodgy, though. It’s clear that humans can do some tasks better in groups, but that doesn’t require group-level selection. In fact, much of the optimisation for group behaviour could be learned, since all the test subjects will have been brought up in a a society dominated by interactions within groups.
McAndrew, F.T. (2002). New Evolutionary Perspectives on Altruism: Multilevel-Selection and Costly-Signaling Theories. Current Directions in Psychological Science, 11(2), 79-82. DOI: 10.1111/1467-8721.00173
Nowak, M.A. (2006). Five Rules for the Evolution of Cooperation. Science, 314(5805), 1560-1563. DOI: 10.1126/science.1133755
O’Gorman, R., Sheldon, K.M., Wilson, D.S. (2008). For the good of the group? Exploring group-level evolutionary adaptations using multilevel selection theory.. Group Dynamics: Theory, Research, and Practice, 12(1), 17-26. DOI: 10.1037/1089-26220.127.116.11