For me it's the difference between a preponderance of evidence suggesting such, and something being applied and proven until any doubt is removed.

For example, I was trying to find studs in drywall recently (last house was plaster and lathe), and looking at things Socratically, I could use a stud finder but I might be drilling into conduit or a pipe. So I was like "I can use magnets to hit drywall screws to try to confirm the presence of a stud", and it seems reasonable, but I've never attempted it in practice, and there could be all sorts of things a magnet could hit, since I've no experience with drywall, how close a steel pipe could be, any of that. So it's a belief. It'd be rather arrogant of me to accept this as a reliable method without testing this method, drill through a pipe and wind up with egg on my face.

So, I tested this by getting two magnets to stick vertically, then measured 16" out, got 2 more magnets to stick vertically, kept doing that until I hit half a dozen spots, all 16" apart. Drilled a pilot hole, felt resistance and the smell of wood, drilled a couple more.

I think somewhere between mounting a flat screen to fixing 3 closet shelves it became knowledge, not sure exactly when, but all the doubts were removed and it never blew up in my face. I can just waltz in a room and sink a bunch of holes in the right spot now without being skeptical of some electronic stud finder.

I guess what I mean to say is that testing something and having it consistently work and be reproducible is what leads to knowledge imo