Another point Los makes (see post below if you don't know what I'm talking about here!) is that the sense of satisfaction we get from realizing some truth is good justification for asserting it. Of course, as he points out, that sense is not infallible -- but we can only judge those instances where we get it wrong in light of a general ability to get it right. (As Wittgenstein noted, in order to be wrong about snakes in some sense we must be right about them in most ways!)
This is relevant in light of a recent book I saw on the shelves at Barnes and Noble, which contended that our sense of "getting it right" is meaningless in light of the many occasions upon which it fails us. (I forget the name of the book, and when I went back to B&N to buy it, it was off the shelf -- anyone who knows the name, please post it in a comment!) But how in the world does this fellow "know" he's gotten this conclusion right?! He may try to claim, "Ha, well, I have lots of evidence." But how has he decided the evidence is for his claim, other than his own judgment?
That is the same problem faced by efforts to reduce all epistemology by "formal epistemology." The models may be relevant and relevatory, but there is no way that the model itself can tell us that -- to decide "this model really says something important!" one has to turn to one's unformalized judgment.