Much of our understanding about intelligence seem to come not from intelligence itself, from how we live it and experience its works, but instead from rationalizing the activity a posteriori. So for example, almost anyone who is really good at physics will not go at it by following rules and formulae, but will actually have a weird “feel” for the thing. But, when asked about their own performance, they will rationalize it into an red-tape-y picture of abstract, clean, impersonal logic — a picture that is as far as the real thing as possible.
From my experience, getting to understand monads (or any abstract notion for that matter) is about developing an intuition about when and why they might be useful. This intuition has to be developed gradually on a series of examples. Only if I can see a pattern there, will I be willing to move to a more abstract level, otherwise I won’t bother. Burrito analogies are useful, but only as a sort of private post hoc explanations.