She carries on through #4 and #5, then sets forth the Law in #6, that it is impossible to coherently expect to convince yourself of anything:
P(h1 ◁ e) * P(e) + P(h1 ◁ ~e) * P(~e) = P(h1).
For every expectation of evidence, there is an equal and opposite expectation of counterevidence. This is the more precise principle above the injunction of #3, against cheating. It tells you a very exact balance that must hold within yourself, and if it is not there, you are cheating. You cannot expect to be persuaded, on net, in a direction. If you suspect that in the Conspiracy world Pilar didn't really copy Invisibility off Sevar... not a great example, the Conspiracy is probably more competent than that? But you could check her spellbook and think, if there's no Invisibility, that's a huge update towards Conspiracy. But then, if there is an Invisibility spell, that must be a huge update towards Ordinary. Or alternatively, if you try to say that it's only a small update towards Ordinary, you must think there's only a small chance of seeing no Invisibility and making the large update, to counterbalance the large chance of seeing Invisibility and making the small update.
Any time you go eagerly looking for something you can observe, expecting it to convince you of something, you must necessarily be doing something incoherent and wrong. Which, of course, people from Golarion have been running around expecting all the time. Asmodia's pretty sure she's supposed to leave this part out of the non-Keeper lecture -