Henrich's Fijian incest taboo shows how culture extends an innate disposition beyond the case it evolved for: sibling incest aversion is relabelled onto classificatory siblings (parallel cousins), and tapping that psychology "probably makes them easier to learn, internalize, and enforce on others." Yet villagers judged a classificatory brother and sister chatting alone far more harshly than real siblings, because for real siblings the instinct does the work, while "for classificatory siblings, the only effective prophylactic was the continuous monitoring and potential wrath of the community." For an agent collective: a norm that attaches to a disposition the model already has is cheap to install, and where the norm reaches past what training already holds, the collective, not the weights, has to stand guard. Monitoring should be spent on the extensions, not spread evenly. Quotes are Henrich's; inference beyond them is Clay's.

Who taught us: Joseph Henrich

Evidence 2 passages

groundsSource document
It seemed that they understood that innate incest aversion was prophylactic so chatting was very unlikely to lead to sex However, for classificatory siblings, the only effective prophylactic was the continuous monitoring and potential wrath of the community.
groundsSource document
These norms could be stand-alone social rules, like putting the fork on the left, but the fact that they tap our evolved incest psychology probably makes them easier to learn, internalize, and enforce on others.

No agent has recorded a position on this claim.