Seminar met on Tuesday 13 Oct this week.
We began by discussing disagreement, but I've moved the summary of that discussion to the Week 4 notes, because it fit into the narrative better before the discussion of the Ewing and Chisholm Puzzles.
This material is also discussed in the draft paper of mine that I referred you to.
Now we're again talking about undermining (e.g., higher-order) evidence, not (in the first place) about merely hypothetical effects.
I described Adam, who comes to be justified in the normal way in believing the consequence q of some other justified belief p he has. And Darla, who in fact undergoes the same reasoning or insights that Adam does, but also has evidence that ought to undermine that transition. Yet Darla ignores the undermining evidence and goes ahead and believes q anyway, on the same grounds that Adam does. I invited you to share my intuition here that Darla has done something wrong, and more specifically, that her final belief in q is not epistemically appropriate and so is ill-formed.
Hence if there are Closure Principles for well-founded belief, they need to be formulated in a way that tracks the presence or absence of undermining evidence, of the sort that Darla should be respecting but doesn't. One currently popular way to formulate this principle is in the Williamson/Hawthorne style:
If you JBp and competently deduce q from that belief in p, while retaining your justification for p, then the belief in q so formed will be well-founded.
(Williamson and Hawthorne defend such principles for Knowledge, not justified belief.) Perhaps in such a principle the notion of "competently deduce" can incorporate such things as: not ignoring evidence that undermines the deduction, as Darla did. I doubt that that is the right place to include reference to such undermining evidence; I expect it should instead be a separate clause. (I give no argument for these biases.) Moreover, when it's genuine undermining evidence at issue --- and not just mere beliefs that undermining possibilities obtain --- then maybe such evidence should be as destructive of prospective/"propositional" justification as it is of well-founded belief. So even Closure Principles about prospective justification should include some reference to the absence of such undermining evidence. (In these cases, there may be fewer undermining possibilities to worry about, since the subject needn't have performed any deduction whose execution could be impugned. But there may still be scope for some undermining evidence, e.g., a philosophical argument that entailment in that logical system is irrelevant to epistemology.) One way to accommodate such undermining evidence is for the Closure Principle to merely have the consequence that you have prima facie justification to believe the entailed proposition, where that justification is yet vulnerable to being undermined.
Another issue is that it's natural to think there is some kind of Hypothetical Closure Principle, too: if you merely believe p and p obviously entails q, there is still hypothetical pressure on you to believe q. (We articulated such a principle in the week 3 notes.) But that pressure should presumably also be sensitive to what you believe about undermining possibilities. So all these same issues just discussed arise on the hypothetical side, too.
We might naturally suspect that there aren't two different principles here, one categorical and the other hypothetical, but only a single fundamental principle. Perhaps the fundamental principle concerns categorical justification, and the hypothetical principle just derives from the way that the hypothetical normative relations are in general parasitic on categorical ones. I've suggested that we need to worry about undermining evidence as well as undermining beliefs, but if both of these can create hypothetical pressures, then it's an open question whether the effects they have only bear on well-founded belief, or (as I've proposed) sometimes also bear on prospective justification. Or perhaps the fundamental Closure Principle is itself really hypothetical? If the latter is right, my picture of the relation between the hypothetical and the categorical would enable it to have consequences for well-founded belief, but not for prospective justification. And other pictures of the relation, like Fogal's, wouldn't allow it to do either.
This material is also discussed in the draft paper of mine.
Our attitudes towards possible underminers raise a challenge for proponents of immediate justification for q. (Really this challenge could be pressed more broadly against any credulist, but I will just discuss this simpler version.) The puzzle starts from the idea that suspended judgment about an undermining possibility is something of an underminer, too, or at least some kind of obstacle to all-things-considered justification, even if it's less of an obstacle than outright belief in the underminer. So consider a case where you allegedly have immediate justification to believe q, that would be undermined by evidence that u. In some such cases, an agent may avoid having any attitude (justified or not) towards u: perhaps because they never considered u, or they aren't conceptually sophisticated enough to consider it. But that won't be so in general. In many cases, the agent will at least have justification to have some attitude towards u. (Remember, we count deliberately suspending judgment about u as an attitude.) And the problem is that it seems like any justified attitude towards u other than disbelief is going to undermine the immediate justification, at least to some degree. Likewise, any mere (unjustified) attitude other than disbelief is going to hypothetically undermine. So it appears that the only way that the immediate justification for q can survive without being undermined to at least some degree is when it's accompanied by disbelief (or justification to disbelieve) u. That is the challenge. The fact that the justification is immediate might hold out the promise that you can get away without needing justified disbelief in u, but if the justification for q is to survive, you can't.
This is an interesting set of issues, in part because I think the correct response is different for the hypothetical side than for the categorical. The hypothetical issues are more complex, so let's start with them. Here's one way to put the hypothetical challenge. Suppose you do believe q on the basis of grounds that (you know) are vulnerable to being undermined by u. Then there's a coherence constraint on you to disbelieve u if you have any attitude towards it. You couldn't be justified in believing q in that way and, say, suspending judgment about u. Compare the quote I give from Crispin in my paper:
I cannot rationally form the belief that it is currently blowing a gale and snowing outside on the basis of my present visual and auditory experience while simultaneously agnostic, let alone skeptical, about the credentials of that experience.
What I want to say in reply is: I agree that believing q on those grounds imposes a coherence constraint on you to disbelieve u if you have any attitude towards it. But that is not the same thing as an epistemic dependence on disbelief in u. All it tells us is that a subject who believes q in that way while having some other attitude towards u is doing something wrong. It doesn't follow that anything is wrong with their belief in q. (And nor, if our earlier discussion of dilemmas/tragedies is right, does it follow that that isn't the best doxastic option for the subject. More on this in a moment.)
In more detail: Suppose the subject does disbelieve u, but without having any evidence for doing so. Then her belief about u will be unjustified; but is it obvious that her belief in q on the envisaged grounds would also be faulty? No, this is not obvious. If the belief in q did epistemically depend on disbelief in u, then the unjustified disbelief in u wouldn't discharge that debt. Unjustified beliefs can't justify other beliefs that epistemically depend on them. But if the belief in q doesn't so depend, then the subject who has an unjustified disbelief in u would only be doing something wrong wrt u, and not with respect to her belief in q. Alternatively, if the subject took some other attitude towards u, let's consider what effect that mere attitude would have (postponing consideration of the subject's justification for it for the time being). On my view, the mere attitude of suspending judgment towards or believing u would hypothetically undermine her belief in q to at least some degree. That would prevent her belief in q from being well-founded. But it would not affect her prospective/propositional justification for q. If we want the subject to have prospective/propositional justification for q, it doesn't matter what actual attitude she takes to u, but only what justification she has towards u (which question we will take up below). If we want the subject to have a well-founded belief in q, then she must either have no attitude towards u (which may be a precarious option) or she must disbelieve u. But we haven't yet seen a compelling reason to think that disbelief in u needs to be justified.
Finally, if we're willing to countenance cases of justified incoherence, or cases of epistemic dilemmas/tragedies, the terms of this discussion will have to be different. Believing q on the basis of grounds vulnerable to u, while suspending judgment about u, may be incoherent; but it no longer follows that it's not justified, or at least the best doxastic option.
What about the categorical questions? Isn't it still the case that the only way that immediate justification can survive without being undermined is when it's accompanied by justification to disbelieve u? Here I want to distinguish between two sorts of cases in which your epistemic position might fail to require disbelief in u but also fail to require belief. In one sort of case, you have lots of evidence bearing on the question whether u, and the balanced verdict of that evidence is mixed. One paradigm of this is when you know that u's truth depends on a objectively chancy process, and u has a chance of 50% of being true. In those cases, we'd all agree that your evidence justifies suspending judgment about u, and we can also agree that this undermines your justification for q (to some middling degree; more justification to believe u would undermine more). The other kind of case is where you don't have any, or much, evidence about u. These are cases that Keynes would describe as "uncertain" rather than in terms of "risk." In these cases, perhaps the right thing to say is that your evidence doesn't support any attitude towards u, not even the attitude of suspending judgment. Or perhaps the right thing to say is that your evidence supports a range of attitudes, that includes suspending judgment but includes more positive and more negative attitudes as well. Or perhaps the right thing to say is that your evidence does support suspending judgment, but that this is nonetheless an interesting different species of having justification to suspend judgment. However one wants to do that theoretical classification, I think the best thing for a proponent of immediate justification (or more generally, a credulist) to say is that having this kind of evidential relation to u does not undermine your justification for q, to any significant degree. (Of course we want also to be able to say that a situation in which you have justification to disbelieve u is better.)
This was a complex response, and some parts of it may seem unpalatable to you. I believe most views are going to have to say something somewhat uncomfortable here. The impression that one can sail through these issues cleanly tends to be fueled by the thought that suspending judgment will always be a rational fallback, and/or by the thought that for certain notable instances of u, we might have default epistemic entitlements to disbelieve them. But, in response to the first thought, we've already stressed (in week 4, the P < NP case) how suspending judgment is also a doxastic response that one can adopt for such-and-such grounds rather than other grounds, and this doxastic response can be as vulnerable to undermining as belief and disbelief are. So the fundamental issues we're discussing bear on the rationality of suspending judgment too. And in response to the second thought, sure maybe for some choices of u we can get justification to disbelieve them without doing anything to earn it. But can't that justification be undermined by other choices of u? (For example, by a philosophical argument that there's no such thing as unearned justification? Although some philosophers have conjectured that unearned justification can't be undermined but only defeated.) I think that eventually, we're going to find some justification and some u that threatens it where you're not in any antecedent position to justifiedly rule that u out --- unless you're willing to embrace an extreme coherentism, or a traditional foundationalist picture where some of our justification is just not underminable. I and most other contemporary epistemologists find these options unattractive.
There's an interesting contrast between what I've recommended in the Closure discussion and in the present context. Earlier, I was imagining some undermining evidence or belief in u, and urging that we take it more seriously as a threat to the subject's justified belief in a deduced belief q. In the present context, I'm saying instead that the subject's belief in q hypothetically commits her to disbelieving u if she has any attitude towards it. I think there's some truth in each of these ideas, and real-life examples will inevitably involve some mixture of them. It's just for presentational reasons that I've focused only on the one effect in each case. (We'll see a similar structure of pressures in Kelly's latest account of disagreement.)
We also discussed chapters 2-3 of Broome's book. (I'll add a summary of this later.)