## Our model Fig is a finite population analogue of

Inclusive fitness theory (Hamilton, 1964, Rousset and Billiard, 2000, Wakano et al., 2013 and Lehmann and Rousset, 2014) is another approach to studying the Romidepsin of social behavior. In this approach, each individual?s fitness (expected number of viable offspring) is expressed as a sum of portions of fitness due to itself and each other individual. An individual?s inclusive fitness is then defined as a weighted sum of fitness portions bestowed on self and others, where the weights represent relatedness to the recipient.

Inclusive fitness theory is regarded by its proponents as a general and powerful framework for understanding the evolution of cooperation. Howevever, Nowak et al. (2010b), building on earlier critiques by Cavalli-Sforza and Feldman (1978), Uyenoyama and Feldman (1982), and Matessi and Karlin (1984), showed that fitness is not generally equal to a sum of portions due to separate individuals, and thus the quantity of inclusive fitness is only well-defined in special cases. Some proponents of inclusive fitness theory responded (Abbot et al., 2011 and Gardner et al., 2011) that such portions of fitness can always be identified using linear regression (Hamilton, 1970, Queller, 1992 and Frank, 1998; see also Birch, 2014). Yet Allen et al. (2013b) showed that this regression method relies on invalid use of statistical inference tools and leads to false conclusions.