I want to explore a principle I will call Clarity. My point will not be to defend it, exactly, but rather to clarify it and put it to work in an argument that seems to me very interesting. Here's a rough formulation:
Clarity. All agents are rationally required not to believe any proposition that is unclear to them.
This formulation leads a lot open (and thus might be autological right now), but I want to focus on what 'unclear' might mean. To do that, we need to see why a principle like Clarity might be intuitive in the first place; that will provide some discipline on what 'unclear' can mean.
The guiding thought is that an agent ought not to believe things of which she has an insufficient understanding. Sometimes a person's understanding can be so incomplete that belief isn't just normatively bad or forbidden but impossible. Likely I can't believe that the borogoves were mimsy, for example. So maybe I can't believe what I hear or see that's said or written in a language I don't understand. And maybe I can't believe what I hear someone say that is sufficiently garbled, say. Ignore all of these cases; my interest is in cases where we can but ought not to believe various things. There will be some inevitable fuzziness around the border.
Sperber has as an example the proposition that the unconscious is structured like a language, which he says Lacanians and Lacan himself at least claim to believe. The problem, or at least one problem, is that there are many ways to interpret the sentence 'the unconscious is structured like a language'. Now, depending on your theory of definite descriptions and of semantic content, issues I don't want to adjudicate here, 'the proposition that the unconscious is structured like a language' might fail to refer. Waive that problem. Suppose we trust that Lacan had a definite claim in mind and was just bad at conveying that definite claim. Call that intended proposition L. What Clarity as I intend it should forbid is his followers' belief that L. Notice, of course, that Clarity would not forbid Lacan from believing L; Clarity is relativized to people for a reason.
Other examples should now come to mind, but they will of course be controversial. Skeptics of recent work in analytic metaphysics may question, for example, the belief that naturalness is natural or that fundamentality is fundamental. More generally conceptual primitives often invite the accusation that propositions involving them, if such there are, are insufficiently understood. (I think basically no one would think that every conceptual primitive is not understood.) A look at the list of things philosophers have claimed not to understand would provide a good list of potential examples: substitutional quantification, "metaphysical" vagueness, relative truth, and more.
There are a couple of different approaches to understanding what "sufficient understanding" is or should be. I don't love any of them. For example, suppose we have a well-understood set of propositions, P. Next suppose that a proposition p is unclear to S just when there are propositions q, r in P such that, for all S knows, p might be either q or r. This is good for some examples, maybe L and some others, but I don't think it'll work for others – if I might play the skeptic of metaphysics, I doubt they could say at all what well-understood propositions 'naturalness is natural' might express. So I don't think that proposal will work.
We also don't want to say that p is unclear to S iff S doesn't know what proposition p is. Knowledge-'wh' is itself too unclear and context-dependent to do the kind of work we want it to do here.
The following explication is the one I'm happiest with, which isn't to say I'm happy with it. If p is sufficiently clear to S, there is a relatively large and diverse set P of well-understood propositions such that S's credences in members of P conditional on p are raised or lowered significantly (or, perhaps, made more resilient) conditional on p. This is a very subjective criterion; it can be made more objective by requiring that those conditional probabilities be rational or reasonable. It also requires us to have a prior grip on a set of propositions that are themselves well-understood. That is a weakness of the definition, but I suspect it is inevitable. Finally 'relatively large' is also vague. Again I doubt that vagueness is eliminable, though I'd be happy to be shown wrong.
How does this understanding of clarity help with our examples? Take "naturalness is natural". Those who think belief in it ought to be forbidden by Clarity will argue that there are no propositions that don't themselves involve naturalness – i.e., propositions that are uncontroversially "well-understood" – whose (rational, reasonable) probabilities are significantly affected by the truth of the proposition expressed by that sentence (if such there be). Its defenders might then reply that what matters isn't whether the propositions whose (rational) probabilities they affect be uncontroversially well-understood, just that they be well-understood. (Williamson sometimes seems to take this kind of line about argument.) But at least in many dialectical contexts that kind of reply won't work. Better, then, to exhibit less controversially well-understood propositions that are rationally affected by the given proposition.
The problem with L is that for people other than Lacan it is very hard to see what its implications might be. What should someone who thinks this think about the unconscious and its structure, what predictions should they make or explanations should they favor? It is all very unclear.
So, despite its inadequacies, this is the best explication of propositional clarity I could think of. It is serviceable enough, and so in subsequent posts it's what I will work with. But I would very much appreciate alternative attempts at spelling out the kind of notion I have in mind! Either way, after this I will say more about what kind of argument there is for Clarity given this understanding, and then I will use it to argue roughly that there is a discipline we are rationally required to pursue to at least some extent, philosophy.