I wonder if this is actually just some sort of statistical effect. Suppose intelligence is normally distributed in a high intelligence population and also in a low intelligence population but with two different medians. The guy at +2 sigma in the latter may be at 0 in the former. He may then reasonably conclude that he is extremely intelligent. If, on top of this, he doesn't then get the chance to mix with where the HI population resides (can't get into university, chooses to live outside of HI population centres), he's never going to be disabused of the fact that he's not really incredibly smart.
It's like being the best basketball player in your school. If you never go to the summer training classes, you'll never discover that there are thousands like you, and that the guys who didn't make the cut at a different school are way better than you are.
It's like being the best basketball player in your school. If you never go to the summer training classes, you'll never discover that there are thousands like you, and that the guys who didn't make the cut at a different school are way better than you are.