> If there are largely no changes between humans in recent millennia, then the differences in testing scores are due to cultural biases in the construction of the test and the education of people.

Not being sarcastic, I suppose things you see in IQ tests like spacial manipulation are culturally biased in that certain cultures give kids shape-in-the-hole toys, teach them shapes, have geometry classes, etc. You could say improvements in intelligence scores are teaching to the test. You could also say we're standing on the shoulders of giants.