As a Math/CS graduate (I focused on mathy aspects of CS): math cs people struggle at working with real world abstractions. They abstract and simplify, but then tend to convince themselves that simplified model and its conclusions are totally all there is to reality.
And if they accept that model is not super perfect, they completely reject model - to find another simplified model that they convince themselves if flawless.
They struggle with the "this is model with such and flaws we need to keep in mind while interpreting".