When in Rome, do as Bayesians do: Statistical learning and parochial norms
- Scott Partington, Cornell University, Ithaca, New York, United States
- Shaun Nichols, Cornell University, Ithaca, New York, United States
- Tamar Kushnir, Cornell University , Ithaca , New York, United States
AbstractIt’s a familiar point in anthropology that many norms are parochial, meaning they apply to people in certain groups (e.g., one’s ingroup) and not to others (e.g., one’s outgroup). One explanation for such parochialism is that people are just innately biased against outsiders. But it’s also possible that, given the evidence, people infer the parochiality of norms in statistically appropriate ways. This paper uses a Bayesian learning framework to investigate inferences of normative scope both experimentally and computationally. An experiment in which adult participants (n = 480) viewed sample violations of a novel rule among novel groups reveals that both sensitivity to statistical evidence and prior knowledge of relevant social categories are integral to computations of normative scope. In tandem with the experimental results, computational analysis supports the notion that degree of prior inclusivity bias (i.e., an expectation that a norm will be broad, rather than narrow, in scope) is another key factor. Together, these novel insights raise intriguing possibilities for integrating perspectives on norms research.
Return to previous page