However, what I meant by the previous comment is the "proper abstraction" is the object of study and practice in all the disciplines I mentioned.
There are "good" abstractions and "bad" ones. There are measures of complexity of abstraction. There are measures of effectiveness, and accuracy, measures of similiarity, equivelence, subjugation, of various abstractions. There are measures of systematic bias, and various other aspects of abstractions themselves studied in all these various forms of "math modeling."
In sum, the whole aim of these branches of study are to teach people how to make abstractions. From my experience, these fields are rather successful in this endevour.