Two foundational mathematical thinking constructing constructs (if I may say): * (Existential-Universal) quantification * (Epsilon-Delta) language Once people start to not only understand it, but actually think that way, they start to be scientists and mathematicians. It opens the door of refutable and constructive thinking. I've seen also a lot of people struggle with definitions .
One problem with epsilon-delta definitions is that they don't scale. In many topological spaces, there's no concept of distance and so epsilon-delta stuff has to be replaced with more general open sets. About the same reason why sequences have to be replaced with nets and filters.
Though it's intuitive for humans and its limits are quite broad, if I may say :)
Eg. we usually don't start defining the derivative with distributions.
I didn't mention geometry neither btw though it's foundational too imo.