This is an interesting one, in that with things like Must Be True Principle and Justify the Conclusion Principle (and even Strengthen Principle to some degree) the task is often so exceptionally similar as to be synonymous: match, as near to perfect as possible, the broad strokes of the principle with the specific details of the scenario presented.
So whether you're moving up (Justify/Strengthen) or down (Must), you're ultimately just playing that matching game, looking for a piece-to-piece concordance.
That isn't to suggest you can completely
ignore the construction/type itself (as I'll explain shortly), but merely to note that the distinction typically proves irrelevant in practice...something I think we can all be thankful for
As for the difference, it's the same precept we see with regular Must vs Strengthen: we classify questions as Must Be True when the stimulus, whether specific scenario or broad principle, is used to determine the correct answer; Justify and Strengthen move in the opposite direction, where the correct answer choice (again, whether specific or broad) is used to affect your opinion of the argument in the stimulus. The degree of help is where we draw the line between Strengthen (help) and Justify (proof).
So which way are we moving here?
I'd argue that it's downward, where the facts in the stimulus are used to find an answer choice principle that best describes, broadly, what they tell us. So for my money this is classic Must Be True! In fact, it operates an awful lot like a Method of Reasoning question (which themselves are just abstract Must Be True), where the correct answer simply describes, in generalized terms, the details in the stimulus
A final takeaway: stay vigilant in your desire to correctly identify and describe each question that you see by type, as so much of your success depends on knowing what the task at hand truly is, and what tools then are most powerfully deployed in completing it. But. Recognize too that every so often the type of question matters little (if at all) relative to the nature of the relationship that must exist between stimulus and answer choice. Principle questions are prime candidates in that latter regard, since, as I noted above, it's always just a game of matching pieces.
Hope this helps!