Relying in science and data is great, but relying in science avoiding rational thinking is questionable.
About the IQ test, what would make the test "unlearnable" in the sense that one cant train to do the test and be better at it? Thus enhancing his IQ.
Do those studies experimented this : measuring individuals who made the test (same rules, different exercices) multiple time. If their IQ increases over repetition, it definitely means that IQ test measures _in some way_ the ability to do the test, which makes the IQ test narrow.