They were analyzing a bunch of data from a big Australian survey of twins.
This included the twins, the twins’ family, and the twins’ mates “height, BMI, age, education, income, personality, social attitudes, and religiosity”, which were the dimensions along which they tried to predict mate choice.
If we want to play the dangerous game of trying to explain differences between contradictory studies instead of just dismissing everything as noise, I might argue that this looked at some pretty different variables compared to the last set.
Instead of looking at facial similarities, it’s looking at things like social attitudes and religiosity; young children trying to imprint on their mother’s image can maybe be forgiven for not knowing her opinion about Asian immigrants (one of the “social attitudes” questions they asked).
They find…well, some things come out heritable, but the confidence intervals are really wide.It’s not consistent with a simple genetic theory where you just get both parents’ genes.It might be consistent with a more complicated genetic theory where mate preferences are on a sex-appropriate chromosome or get chromosomally imprinted such that you only care about your father’s preferences for women and your mother’s preferences for men, but this is hard and I haven’t seen any analysis of whether it’s evolutionarily worth it.Spence & Smith replicated this finding with zebra fish raised by differently-colored zebra fish.So the research shows conclusively that sexual selection is based on learned imprinting, at least in animals whose names start with the string “zebra fi*”. I can’t find the study itself, but multiple reviews cite Jedlicka 1984, who looked at children of mixed-race couples (white and native Hawaiian).