I would argue that a lot of the computational based problem solving , from middle school through early undergraduate years, focused on topics historically oriented to boys’ interests, aren’t a good measure of innate math talent either.

But those have historically left a lot of female students behind.

Male or female, most students are really looking to get through math requirements with plug-and-chug replication of algorithms to get to an answer - not genuine problem solving or abstraction. However, being able to reproduce an answer on a very slightly different problem, or just one with different numbers to plug in, does very little towards using mathematical as a means to model problems independently and find solutions.