My personal favourite hypothesis for language learning can be reused for this: There /is/ some pre-coded information about these problem domains in the brain.

My personal favourite hypothesis for language learning can be reused for this: There /is/ some pre-coded information about these problem domains in the brain. It just /isn't/ coded explicitly, but implicitly as a bias of the corresponding learning faculty. Even hard things can be easy to learn, if the learning mechanism itself is properly biased.