The Arche Modality Conference has just finished, and has been excellent fun. I've put some photos on my website. In the end, my comments on Scott Sturgeon's talk mainly focussed on the claim that the infallibility of idealized conceivability methods is the wrong focal point for his worry. He's worried that if we are realists, we shouldn't think that metaphysics and epistemology will fit together the way the infallibilist requires - there should be room to make a mistake. For idealized conceivability (as he thinks of it) involves only mental idealization (allowing unlimited time, capacities etc.). How can we have a guarantee of getting a true belief, if we have only idealized these mental processes, and said nothing about their relationship to the world?
But I argued that we still have a worry of the same kind for fallibilist conceivability views. My point here was just an analogue of what we tell students who think Hume’s worries about induction show that we aren’t guaranteed to get true beliefs by inductive methods, but still think these beliefs are likely to be true. What I claimed is that, for the same reasons Sturgeon thinks we should be worried about the infallibility of conceivability, namely because epistemology and realist metaphysics don’t fit together like that, we should also be worried about the claim that it’s likely that conceivability will deliver true beliefs. Epistemology and realist metaphysics don’t fit together like that either – you don’t get good ways of finding out about the independent world when you idealize along only mental dimensions. Good ways of finding out about the independent world require input, which is a world-involving relationship, and not purely mental.
So infallibility is not necessary to generate the problem. In fact, I claimed, it’s not sufficient either. Sturgeon’s worry will go away if we can somehow add a mind-world input element into our account of ideal conceivability. This sort of ideal conceivability can unproblematically be infallible, for the same reasons Sturgeon thinks idealized vision (which is, among other things, always veridical) can unproblematically be infallible.
The deep problem in the vicinity of Sturgeon's worry is a very old one: how can a priori reflection be a source of knowledge of facts which are construed realistically? But of course, there are many answers to this and debates about them which would have to be engaged with before we could claim to have pressed the point in a new way.