This is an issue for clever people on the LSAT, where there is a technique to solving the problems that takes practice, and when you do a course on it, the thing STEM people tend to have to unlearn is solving the problem as a class of problem instead of just recognizing which question form it is and grinding through it fastest. The other trick is to recognize which ones are designed as a time suck. When I saw my first set of logical reasoning practice questions I thought, "ah, a state machine," while the other students had already put down their pencils.

Abstraction solves a lot of hard problems, but it is also costly.