The article makes it seem as though there are absolute measurements being made of certain elements in the brain, but it's almost more than certain to be pattern matching or machine learning that is being done. You find some people that are good at doing X and have them take a test. Then you take new people, have them perform the same test, and if their scans are similar to the scans of the initial people, then you say these new people are pre-disposed to do X well.