Tests can only tell you what some piece code is supposed to do, not why its doing it the way its doing. Why do you choose to define this particular constant to this particular value, why are you using that particular data structure, why are you rolling your own version of a certain function rather than using the one supplied in the library you are using, why are you using that seemingly round about way to get to that particular result?

There are lots and lots of ways to write a function that passes a test, I want to know if there is a reason why you chose the approach you used.