“Fluid intelligence“, which is your ability to solve problems without applying outside information, peaks around age 20. I find this discouraging. But then “crystalized intelligence”, which is your ability to apply outside information to problems, does not decline until later in life (50, 60, 70? there are a lot of numbers in this article.) I find this encouraging as I am entering this window now. It also suggests to me that multi-generational teams may be good at problem solving if we can get past some of the cultural issues (seriously, what is with the avoidance of eye contact by 20-somethings today?)

The article goes on to describe measurement errors and biases that have affected past studies. A big one is that when you test the intelligence of a sample of people of different ages all at once, you get one answer, whereas if you follow the same people for decades and test them repeatedly using the same methods, you get a different answer. The latter approach is the better one, but of course it is difficult because it takes 50 years to carry out an experiment.