1. Group Selection: A guide for the confused OR: How I Learned to Stop Worrying and Ignore Multilevel Selection.
People have been arguing about group selection for decades. What’s going on? Is it something you need to know about? <thread>
People have been arguing about group selection for decades. What’s going on? Is it something you need to know about? <thread>
2. tl;dr No. There’s nothing to see here. New-fangled ‘group-’ or ‘multilevel-selection’ is just a complicated way of saying ‘selection for cooperation’. Go and do something more interesting instead.
3. It all starts with the ‘problem of altruism’. Altruists pay a cost to benefit others. Egoists do not–they accept benefits but pay no costs and provide no benefits. So altruists appear to be at a disadvantage to egoists, and it’s unclear how evolution could ever favour altruism
4. The solution is that, although altruists can’t survive on their own, they can survive (and thrive) if they stick together with (and direct benefits to) other altruists, and avoid egoists.
5. There are various ways of ‘sticking together’ – or, to put it more technically, there are various causes of ‘Positive Assortment’ (PA). They include:
6. Kin Selection (KS): Altruists breed more altruists, who remain in close proximity doi.org/10.1086/497114
7. Reciprocal Altruism (RA): Altruists are conditionally altruistic to other altruists, but stop being altruistic to (or ‘walk away’ from, or punish) egoists jstor.org/stable/2822435
8. Population Structure (PS): Even in the absence of KS or RA, in some fission-fusion populations, it is possible for enough altruists get shuffled together by chance alone (as in John Maynard Smith’s ‘haystack model’) doi.org/10.1038/201114…
9. If ‘enough’ altruists ‘stick together’ then the costs (of being altruistic) are outweighed by the benefits (of interacting with other altruists), and altruism will evolve. ‘Nice guys finish first if they cooperate.’
10. So far, so good. As far as I am aware, there is no substantive disagreement about any of this? So what are people arguing about? Three things: a) the utility of the Price Equation, b) the prevalence of Population Structure altruism, and c) which language to use.
11. First, is it useful to model this process with George Price’s equation (PE)? doi.org/10.1038/227520…
12. The Price Equation distinguishes between: i) the individual costs of altruism, and ii) the benefits of interacting with other altruists, and identifies the point at which the costs are outweighed by the benefits, and hence altruism will evolve.
13. (Technically, the Price Equation “partitions the variance in fitness” into i) ’individual’ and ii) ’group’ components (ie PA: KS, RA, PS), each of which can vary.)
Some people think the Price Equation is useful, others don’t. 🤷♂️
Some people think the Price Equation is useful, others don’t. 🤷♂️
14. Second, does Population Structure alone explain altruism in nature? This is an empirical question. Most people think that it can in theory, but doesn't in practice. Population Structure is a ‘weak force’; there are few if any clear examples of it.
15. (Whereas, there’s a lot of evidence for Kin Selection, and some evidence for Reciprocal Altruism.)
16. The third and final issue, and the one that causes all the confusion, is largely semantic.
17. Some people, like @richarddawkins, use the term ‘Group Selection’ to refer to the ‘Naive View’ (NV) that there is no problem of altruism, and/or to the view that Population Structure (PS) explains altruism.
18. They argue that ‘Group Selection’ (meaning the Naive View and Population Structure) has been rejected in favour of Kin Selection and Reciprocal Altruism.
19. Other people, like @David_S_Wilson, use the term ‘Group Selection’ to refer to the general claim that the evolution of altruism requires Positive Assortment (hence some combination of Kin Selection, Reciprocal Altruism, and Population Structure).
20. More specifically, @David_S_Wilson refers to the individual cost of altruism as ‘individual-level’ or ‘within-group’ selection. And he refers to the benefits of Positive Assortment as ‘group-level’ or ‘between-group’ selection. Hence ‘multi-level selection’ (MLS).
21. @David_S_Wilson even refers to Kin Selection & Reciprocal Altruism as types of Group Selection… because they are types Positive Assortment. Under this nomenclature, ‘reciprocal altruism’ is a ‘group-level adaptation'…
22. … after all, you can’t be altruistic on your own, you have to do so in with others, in a ‘group’. Mother and child, or you and your reciprocating partner, constitute a ‘group’, hence @David_S_Wilson calls it ‘group selection’.
23. So, when @richarddawkins says “Group Selection (NV & PS) is dead, long live KS and RA”, and @David_S_Wilson says “No! Group Selection (PA, including KS and RA and maybe PS) is alive and kicking”, they are talking at cross purposes 
24. They both take a gene-centred view, and reject the Naive View. They both agree that altruism requires Positive Assortment. And they both agree that Positive Assortment is caused by Kin Selection & Reciprocal Altruism. They just disagree about the role of Population Structure.
25. And that's it, that's all, that's all there is. ‘Multilevel Selection’ is just a way of saying ‘selection for cooperation’. Confused by @David_S_Wilson? Just translate MLS-talk into cooperation-talk and it becomes clear… clear that nothing of substance is being advanced.
26. For example, when @David_S_Wilson says ‘altruism evolves when within-group selection is outweighed by between-group selection’, all he is saying is 'altruism evolves when the costs (of being altruistic) are outweighed by the benefits (of interacting with other altruists)'.
27. So, which ‘language’ to use? As far as I can tell, there are no advantages of using the terms ‘group selection’ or ‘multi-level selection’… they provide no new explanatory power, they identify no new causes of altruism. They just repackage old theories.
28. There are only disadvantages of using this language. The general terms ‘group selection’ and ‘multilevel selection’ obscure the various more specific theories of altruism that have been developed. For example…
29. While it's true (ex hypothesi) that ‘within groups selfishness beats altruism, between groups altruism beats selfishness’, it’s not true that the ‘rest is commentary’ @David_S_Wilson. ‘The rest’ is the various specific, distinct theories of how & why altruists form groups
30. Worse, the terms ‘group selection’ & ‘multilevel selection’ sow confusion, and invite misunderstandings, like:
31. The ‘naive view’ has been vindicated; ‘group level selection’ is an alternative to ‘gene level selection’; there is ample evidence of population structure; PA/PS requires or implies direct group competition (eg warfare); ‘groupishness’ is evidence of PS; and so on.
32. So my advice would be, if you want to argue about the causes of altruism, use the precise theory and terminology that’s available (and demand that others do too). Do not use the terms ‘group selection’ or ‘multilevel selection’ because you will be misunderstood.
33. Better still, leave this tired old debate. The battle is over. Go and do something more interesting — go and investigate other fascinating forms of (pro)social behaviour, like coordination and conflict resolution, that have been unfairly neglected. <the end, hopefully>
