Evolution and Kantian morality

What kind of preferences should one expect evolution to favor? We propose a definition of evolutionary stability of preferences in interactions in groups of arbitrary finite size. Groups are formed under random matching that may be assortative. Individuals' preferences are their private information. The set of potential preferences are all those that can be represented by continuous functions. We show that a certain class of such preferences, that combine self-interest with morality of a Kantian flavor, are evolutionarily stable, and that preferences resulting in other behaviors are evolutionarily unstable. We also establish a connection between evolutionary stability of preferences and a generalized version of Maynard Smith's and Price's (1973) notion of evolutionary stability of strategies.