Let me change my reading of St. Alphonsus Liguori's statement, "An opinion of sentiment that is morally certain, is that which excludes all prudent fear of falsity; so that the contrary opinion is regarded as wholly improbable." In effect, I'm going to pretend that semicolon isn't there.
In my earlier post, I said a fear is prudent if it's "proportionate to both the magnitude of the evil and its potential of occurring." But I think it's more sensible to pull the proportion-to-the-magnitude part completely out of the definition of moral certainty, because it operates in our moral reasoning in a way we wouldn't naturally think of directly as certainty.
Suppose: A medical patient is twice as likely to have illness A as illness B, neither illness being especially dire if treated. The treatment for A doesn't work for B, so if he is treated for A but has B, it will be known that he has B when he doesn't respond within a day. (I'm trying to gussy up the old "Take two aspirin and call me in the morning" wheeze.)
We wouldn't say the doctor is "certain" the patient has illness A, but because the gravity of the patient having B when treated for A is not very great, the doctor can be morally certain in prescribing treatment for A. He is not certain the treatment he prescribes will cure the patient; he is certain he should prescribe treatment A.
If the patient is twice as likely to have illness A as illness B, but treatment for A will kill him if he has treatment B, then we're watching House I'm not sure what should be done; moral certainty that the doctor should prescribe treatment A seems to require additional conditions.
Maybe the way to put it is this: The gravity of the effects of acting as though something is true, when it happens to be false, affects the certainty that acting that way is morally good.