Sampling Assumptions and the Size Principle in Property Induction
Philip M. Fernbach · eScholarship (California Digital Library) · 2006
The 'size principle' emphasized in recent Bayesian models of inductive generalization (Kemp & Tenenbaum, 2003;Sanjana & Tenenbaum, 2003;Tenenbaum & Griffiths, 2001) is tested in the domain of property induction.As predicted by the model, the size principle is obeyed more frequently when a strong sampling assumption is explicit than when sampling is weak or unspecified, but it is not followed consistently.This implies that, although people are sometimes sensitive to sampling assumptions as specified by the Bayesian framework, models based on a strong sampling assumption may not provide general accounts of property induction.