I remember something I read a long, long time ago about how the first IQ tests tested for a type of reasoning that was what was taught in school, and as formal schooling became more common (we are talking XIXth-early XXth century here) and more and more kids were taught that reasoning the numbers on some parts of the tests started shifting. Which when they moved from absolute numbers to percentiles, I guess.
The example that stuck in my mind was that on a similarities test a farm boy would be more likely to associate horse and plow (functional similarity) while a school taught child would associate horse and dog (same category).
