Homo Moralis and regular altruists - preference evolution for when they disagree

Alger and Weibull (2013) present a model for the evolution of preferences under incomplete information and assortative matching. Their main result is that Homo Moralis - who maximizes a convex combination of her narrow self-interest and "the right thing to do" - is evolutionarily stable, if it assigns a weight on the right thing to do that is equal to the assortment parameter. We give a counterexample against their central result, and a way to repair it. We also show that the result ceases to hold if we allow for mixed equilibria or coordination on asymmetric equilibria. Allowing for mixed equilibria, we show that if there is a stable preference, it will be behaviorally equivalent to a regular altruist that puts a positive weight on the payoff of the other that is equal to the assortment parameter. We also consider the cross-species empirical evidence.