Abstract
Although the approaches provided by Daniel Steel and Nancy Cartwright have the same goal―to tackle the challenges presented by the extrapolation problem―Steel complains that Cartwright’s approach is nothing more than simple induction and thus her approach cannot actually help us deal with the methodological problems confronted by extrapolative reasoning. This paper, by presenting a detailed analysis of what Cartwright means in a new account of evidence, aims to point out that the suggestion of Steel’s mechanism-based approach is quite similar to what is maintained in Cartwright’s capacities-based approach, and thus we can say that both approaches extrapolate their models to target populations―or, metaphorically, to use Steel’s own phrase, actually move across the boundaries between model and target―on the same ground.