A Limited PSR As Applied to Modality
"For example, the modal non-rationalist (or, following Ayn Rand's naming conventions, irrationalist)"
Modal rationalism is the idea that what is possible and necessary is in theory knowable a priori. Not only could one figure out a priori—meaning without observing the world—that there can’t be married bachelors, they could figure out all of the facts about what is possible and necessary without looking at the actual world. I’ve always found modal rationalism self-evident—so much so that I had trouble understanding that type B materialism is a thing. Here, I’ll articulate why I think it is right. If you’re looking for a reply to various objections to it, see Richard’s series of articles.
It seems very clear that the principle of sufficient reason, which says that everything has some explanation, is true when applied to modality. All modal facts have some deeper explanation. Suppose someone, for example, said that they thought that a non-multiverse was impossible. They explain that if a non-multiverse is possible, then there’s a high probability there would be a single universe, and it’s very probable that we wouldn’t exist for fine-tuning reasons. Thus, the fact that we exist is evidence that a non-multiverse is impossible. You ask them why it’s impossible, and they say there’s no reason—we just have “empirical” evidence, from the fact that we exist, that it is, in fact, impossible.
This seems really unsatisfactory. It seems like modal facts have to have an explanation. But any explanation they could have seems knowable a priori. What else could the explanation lie in? It could perhaps be grounded in some other modal fact, but that fact seems to require an explanation.
One could reply by claiming that all accounts have facts without explanations. If every modal fact has some further explanation, that results in infinite regress. It’s true that every modal fact won’t have some further explanation, but this doesn’t mean some modal facts are unexplained. For example, it doesn’t seem that there’s a further required explanation for the badness of pain—it just resides in the character of pain that it’s bad.
Why does it not require some further explanation? Well, because you could grasp it just by knowing what pain is like. Thus, for any modal fact, there can be an explicable explanation if and only if one could figure it out a priori. You could figure out from the armchair that there can’t be contradictions; that doesn’t require an explanation, because it’s just contained within a proper understanding of what the concept of a contradiction is. Thus, the argument for modal rationalism is as follows.
All modal facts have sufficient explanations.
Any sufficient explanation of a modal fact can be realized a priori
If a modal fact has an explanation that can be realized a priori, that modal fact can be known a priori.
Therefore, all modal facts can be known a priori.
Think about how strange the modal non-rationalist picture is. For example, the modal non-rationalist (or, following Ayn Rand's naming conventions, irrationalist) thinks that there’s no deeper account of why zombies, ghosts, or inverted qualia scenarios are impossible. Nevertheless, they are impossible.
This strikes me as just as unsatisfactory as thinking that non-multiverses are impossible. Sure, the are some facts about the world that are more likely if it’s true, but it’s a terrible explanation. It leaves various modal facts totally unexplained, in a way that’s utterly unsatisfactory.
So, what do people think about this whole modal rationalism business? Any objections? Btw, Richard’s articles that I linked before decisively refute the appropriation of Kripke to prove that the modal space is not a priori derivable. In short, while you can’t know that water is H20 a priori, but it is necessarily H20, this is just because you can’t know what the word water will refer to a priori. You need to do empirical investigation to figure out what water actually is, because this will determine what things we refer to as water in other possible worlds, but this just means that what our concepts mean depends on the actual world, not that modal space is dependent on the actual world.