That's a very interesting thought. I'd love to see the experiments confirming this hypothesis. If you point me to some peer reviewed papers, I will add them to the page in the link I mentioned. From the minimal research I did, professional psychologists seem to believe there isn't scientific evidence backing Myers Briggs.

I would tend to agree with that. In fact, I would be shocked if the MBTI was found to be the optimal way to carve up the correlations between the multitude of small traits into 16 arbitrary categories. I would love to see more research done collecting all of the information that 'personality test' questions collect, adding in some DNA test results and seeing what correlations can be found. For now I've seen enough (eg. Twin tests on the big 5) to conclude that a significant amount of personality is genetic but I know the systems we have for describing personality right now are abysmal.