Sometimes we believe we ought to do things. Sometimes we then do them. I'd love to know how the normative status of normative judgments (which I'm taking to be beliefs about what ought to be done all things considered) are related to the normative status of the things we do. I think that this is right: if your belief that you ought to Φ is justified, Φ-ing is justified. (If you ought to believe that you ought to Φ, you really ought to Φ.) I've written up a short little piece attacking a view (a.k.a., 'The View') that uses some principles I like but uses them for nefarious purposes (attacks on epistemic purism, attacks on views of the ontology of practical reasons that identify them with states of the world or worldly facts). I've attacked The View before (in 'The Myth of the False, Justified Belief' (here)), but my argument rested on intuitions about the moral significance of facts that an agent is non-culpably ignorant of that some people think are dodgy. (Fwiw, I've found much better rhetoric to use to get people to have the right intuitions than I used in that paper.) It can't be that facts you're non-culpably ignorant of determine what your obligation is, if you fail to take account of them, that's just bad luck. Or something like that. I'll try something different here and try to hit The View where it hurts. (Because I know the targets and we seem to be on reasonably friendly terms, I'm a bit more glib than I would be otherwise. Since they seem to be rather glib in attacking the views I cherish, I hope they'll forgive me as it's clearly not intended to be disrespectful.)
JUSTIFIED NORMATIVE JUDGMENT
Should your conscience be your guide? I really can’t say because we might never have met. If you are conscientious, you will often make the move from a belief about what ought to be done to an action rationalized (in part) by the belief:
(The Move) SBSOΦ
S Φ’s
I can’t say that you should have made The Move if I don’t know the specifics. It’s not obviously irrational to make The Move. Indeed, it seems often that it would be irrational not to. Still, I think this principle can’t be true:
(Con) SBSOΦ → SOΦ
The best people make The Move, but so do the worst. For counterexamples to Con, see any movie about Nazis (but don't see Valkyrie).
If you think there must be some sort of normative relation between practical judgments and the actions and intentions those judgments rationalize, perhaps it is captured by something like this:
(W-Con) SO(SBOΦ → SOΦ)
While all kinds of people make The Move, only those who act rightly conform to W-Con. Conscientious Hitlers are no threat to W-Con, only Con.
W-Con seems trivial. Gibbons (2009: 172) seems to agree. Note that W-Con permits what he calls ‘strengthened detachment’. With W-Con, we can infer (2) from (1):
(1) SOBSOΦ
(2) SOΦ
When should someone believe they ought to Φ? He suggests that someone who considers whether to Φ, gives the matter serious attention, has the intellectual skills to knowingly deduce that she ought to Φ, and has sufficient evidence ought to judge that she ought to Φ. It’s hard to say what sufficient evidence amounts to, but two claims are intuitively plausible:
(SE1) S has sufficient evidence for her belief that p if S has precisely the same evidence as someone who knows p.
(SE2) S ought to believe p if S has given the matter sufficient attention and has sufficient evidence to believe p.
Rather than bicker about whether evidence consists of knowledge or something else, I’ll just say now that nothing here turns on whether we opt for the right view of evidence or some alternative to E=K. Evidence can be pretty much whatever you want it to be, provided that you agree that even if p is part of S’s evidence, p isn’t the evidence that justifies S’s believing p when that belief is an inferential belief. Combine (SE1), (SE2), and W-Con, and the result is that you are permitted to do whatever it is that your epistemic counterparts are permitted to do. Indeed, you and your epistemic counterpart often ought to do the same things, provided that you ought to believe that you ought to do the same things.
This seems to be the view defended by Fantl and McGrath (forthcoming: 85). They say that if you have knowledge-level justification for believing p, it is permissible to treat p as a reason for action as well as belief. Suppose you have knowledge-level justification for believing that you ought to Φ. It would then be proper for you to treat that you ought to Φ as a reason for action, say, when you judge that you should Ψ because you know that Ψ-ing is the only means by which you could Φ. Suppose you have knowledge-level justification for believing p where you kow or justifiably believe you ought to Φ if p. It would then be proper for you to treat p as a reason for action and, arguably, proper to Φ accordingly. Like Fantl, McGrath, and Gibbons, I have a hard time seeing how W-Con could be false. Whatever it is that neutralizes the threats to the normative standing of the normative judgment should take care of whatever normative obstacles stand in the way of plowing ahead and acting on the normative judgment. In spite of the view’s intuitive plausibility, there is something wrong with:
(The View) W-Con, (SE1), and (SE2)
Those who defend The View believe that someone can have sufficient evidence to believe p even if ~p. Here is what Fantl and McGrath say on the matter:
… it is highly plausible that if two subjects have all the same very strong evidence for my glass contains gin, believe that proposition on the basis of this evidence, and then act on the belief in reaching to take a drink, those two subjects are equally justified in their actions and equally justified in treating what they each did as a reason, even if one of them, the unlucky one, has cleverly disguised petrol in his glass rather than gin. Notice that if we asked the unlucky fellow why he did such a thing, he might reply with indignation: ‘well, it was the perfectly rational thing to do; I had every reason to think the glass contained gin; why in the world should I think that someone would be going around putting petrol in cocktail glasses!?’ Here the unlucky subject is not providing an excuse for his action or treating what he did as a reason; he is defending it as the action that made the most sense for him to do and the proposition that made most sense to treat as a reason (forthcoming: 141).
It seems Fantl and McGrath are moved by a kind of anti-luck intuition. If someone shouldn’t Φ, there have to be reasons in light of which the subject shouldn’t Φ and should do something else instead, but they cannot be reasons that are inaccessible to the subject. Why? Perhaps it’s because the best that you can ever do is to do what the rational or reasonable person would do in your shoes. While there’s something to be said for the view that says (i) you should do the best you can and (ii) it cannot be that you should have done more or done otherwise than the best you could given your circumstances, I don’t think that this can be what motivates The View.
In The Myth of the FJB, I argued that there can be facts we are non-culpably ignorant of that determines what our overall moral obligation is. In other words, factual ignorance and mistaken belief excuses wrongdoing, they does not obviate the need to try to justify engaging a course of action that is wrongful. These cases show that there’s a reading of ‘sufficient evidence’ on which (SE1) is trivial and a reading on which (SE2) is perfectly harmless, but these are different readings. If ‘sufficient evidence’ just means evidence that ensures that it is permissible to believe, (SE2) is trivial but (SE1) is (arguably) mistaken. If ‘sufficient evidence’ just means evidence that ensures that we satisfy whatever evidential requirements there are on permissible belief, (SE1) is trivial but (SE2) is (arguably) mistaken. This reading leaves open the possibility that the right to believe turns on more than just the evidence. The view I would defend is that only beliefs that can properly figure in deliberation are permissibly held and since any false belief will pass off a non-reason as if it were a reason, no false beliefs can justifiably be held. If someone has a false belief that leads them to engage in wrongdoing but has the same evidence as someone who acts rightly, the evidence not ensure that the agent conforms to the norms governing action or belief but it might provide an excuse for acting on beliefs that oughtn’t be included in deliberation. The problem with this line of attack was that it appealed to intuitions about the duties you have in light of fact
s inaccessible to the agent. Intuitions about such cases seem to be the sort of thing that move unaffiliated graduate students, not your opponents who think we should drink petrol with ice and wedges of lime.
Let’s try a different line of attack. Coop thinks it’s his lucky day. He’s found a quarter on the ground and heads off to buy a gumball. He passes a vending machine and sees that trapped inside are an infant, a puppy, and a kitten. On the machine, a sign says that anything in the machine could be had for a quarter. He knows that it would be best if the infant were freed, second best if the puppy were set free, third best if the kitten were set free, and that the worst state of affairs involves his buying that gumball he wants. Coop judges:
(3) Cooper ought to bring it about that the infant is freed.
It seems that someone could have just Coop’s evidence and know that (3) is true. So, it follows from The View that (3) is true. It is, however, perfectly consistent with everything I’ve just said that the mechanism that would allow Coop to free the infant is broken. Perhaps the only buttons that work are the buttons that free the puppy or free the kitten. Now, consider:
(OIC): If S ought to Φ, S can Φ.
According to (OIC), what Coop ought to do is something other than what (3) says.
Here’s an argument against The View. It implies that your obligation can be what you cannot do. Obligations aren’t like that. Thus, either epistemic obligations are not fixed entirely by facts about an individual’s evidence or facts about what ought to be done are not fixed by facts about epistemic obligation. Having considered the arguments for W-Con, I find them convincing. When there’s a decisive case to be made against Φ-ing, there’s not a decisive case to be made for settling the issue by judging that you should Φ. As Fantl and McGrath observe, “When trying to determine what is true – that is, in forming beliefs – we draw conclusions from the reasons we have. The same goes for trying to decide what to do. Here, too, we draw conclusions about what to do – we form intentions – from the reasons we have. We bring reasons into our reasoning knowing that we might draw all sorts of conclusions from them along the way, some practical and some theoretical” (forthcoming: 78). We draw conclusions about what to do and believe on the basis of the very same reasons. One intuitive motivation for The View is the thought that the reasonable and conscientious agent cannot fail to discover and then do what should be done. If we were to say otherwise, then it often is a matter of dumb luck that we managed to meet our epistemic and moral obligations. It turns out that that cannot be the rationale for The View, not if it is incompatible with (OIC). [Zimmerman notes that (OIC) causes trouble for Con, but doesn't discuss W-Con. His view differs from W-Con (in part) because he imports an objective element into his view that The View doesn't and cannot given the arguments that motivate it.]
Someone plausibly suggested that the subject ought to try to release the infant rather than, say, bring it about that some critter is freed from the vending machine. It’s true that this costs a quarter and does no good at all, but it’s the right thing to do. Someone could say that, but look what that does to The View. In some nearby possible world our vending machine is working properly and some counterpart of Coop’s has saved the infant. It surely seems possible that he knows that when he pushes the relevant button the baby will be released. So, he has sufficient evidence to believe this. By (SE1), Coop has sufficient evidence to believe that by pushing the button the baby will be released. By (SE2), that is what he ought to believe. By W-Con, that is what he ought to do—release the baby by trying to do so. But, it’s not the case that that’s what he ought to do. He can’t do it.
It seems that those who love The View have to say that our initial supposition was mistaken. Coop’s counterpart did not know that pushing the button causes the baby to be released. Why didn’t he know? Because an epistemic counterpart could have just his evidence for believing this proposition and be mistaken where this mistaken belief would rationalize an action that the agent cannot perform. So, S knows p only if S has no epistemic counterparts in any possible world who falsely believe p and whose belief that p would rationalize forming the intention to perform an action the agent cannot perform under those circumstances. If that’s right, is there anything we can know about the external world? Given The View, all it took to show that Coop’s counterpart did not have knowledge was to find some counterpart of Coop’s counterpart (i.e., Coop) who had the same evidence but had a mistaken belief. Now it is starting to look like The View is committed to the view that we have sufficient to believe p only when our evidence for believing p entails p. That’s worse than my view! My view rules out false, justified beliefs that figure in deliberation. The View rules out the possibility of justified belief based on fallible grounds. So, The View rules out false, justified beliefs and justified, true beliefs based on even the best inductive grounds. It’s a bad view, much worse than mine.
I haven’t found the passage where those who defend The View say that they accept (OIC). So, I haven’t found the passage that shows that The View is incompatible with some view they cherish, only that The View is incompatible with something that strikes many of us as being obviously true. They are well within their rights to say that ignorance subverts obligation, but impotence does not. If they were to go this route, they have to say that someone who doesn’t know p can Φ even if it is physically impossible for someone who does not realize that p is true to Φ. Suppose people can get things out of our very sturdy vending machine either by putting a quarter in its slot or using the right combination to open a combination lock. Coop has no idea what the combination is. Someone could say that Coop isn’t obliged to open the lock to release the infant because he does not know that the combination for that lock is 12-34-56. They can say that he’s obliged to release the infant, the puppy, and the kitten by pulling the vending machine apart with his bare hands even though this is something he cannot do. People can say all sorts of things they shouldn’t.
REFERENCES
Fantl, J. and M. McGrath. forthcoming. Knowledge in an Uncertain World. New York: Oxford University Press.
Gibbons, J. 2009. You Gotta Do What You Gotta Do. Nous 43: 157-77.
____. Forthcoming. Things That Make Things Reasonable. Philosophy and Phenomenological Research.
Zimmerman, M. 2008. Living with Uncertainty. New York: Cambridge University Press.
’ stand for saving the infant. Let ‘Ψ’ stand for saving the puppy. Given (OIC), it cannot be that Coop is obliged Φ. He’s obliged to do the next best thing, Ψ. Given that he’s obliged to Ψ rather than Φ, he’s obliged to refrain from judging that he should Φ. Weaken The View, and it still faces essentially the same problems with (OIC) that the view faces.] [Another footnote: Fantl and McGrath (2002: 77) say that S justifiably believes p only if S is rational to act as if p. It might seem that they reject (SE1) and (SE2). They do. But, they do so on the grounds that it is possible for subjects with the same evidence for believing p to face practical situations where the stakes of acting on p differ. These differences are differences that their subjects are aware of. In my cases, the differences between the situations Coop and his epistemic counterparts face are not differences that they are aware of. Fantl and McGrath’s remarks concerning cases of facts that the subject is not cognizant of that have some practical significance make it pretty clear that they think that such facts do not affect the normative standing of the agent’s actions, attitudes, or the reasoning that connects them.]
WARNING: There's something wrong with the comments box. My comments keep getting eaten. Save your comments before you try to post them. Here's the second part of the comment I've been trying to post in response to Matt:
It seems to me incoherent to me say that it’s proper to treat _that I am permitted to go do Boston_ as a reason for judging that you should go there when you can’t permissibly act on that premise. Note that if you accept (1) and (2), the case is a counterexample to the principle that p is justifiably acted upon (for some p-dependent choice) when justifiably believed. But, it seems then that in light of the reasons that the agent was aware of the agent ought to act against her own judgment about the thing to do in spite of the fact that it’s either not the case that she should revise her views about the thing to believe or positively oughtn’t revise her beliefs about the thing to believe. Again, more weirdness. This is an ought grounded in reasons she’s aware of. If you’re going to say that she shouldn’t act on her judgment because of what the reasons she’s aware of demand of her, why is she to believe she should act that way anyway?
Suppose you press our agent and say that she should have gone to Austin. She defends herself and says that she gave the matter serious consideration. The reasons to go to Austin didn’t seem any more important than those that demanded that she be in Boston. She notes that lots of reasonable people in her shoes make the same choice she did, etc… I would say “Here the unlucky subject is defending it as the action that made the most sense for her to do and the proposition that made most sense to treat as a reason and she defends her judgment as the judgment that made most sense for her to make”. If her judgment was unreasonable, I could see saying that neither the judgment nor the action was excusable. But, if her judgment was unreasonable, her judgment wasn’t justified. If the judgment about the thing to do is reasonable, that ensures that the action was reasonable. Don't you say in the petrol/gin case that the subject isn't just offering some excuse because they can show, "the unlucky subject is not providing an excuse for his action or treating what he did as a reason; he is defending it as the action that made the most sense for him to do and the proposition that made most sense to treat as a reason."
I submit that any case where you justifiably judge that A-ing is the thing to do is a case where some subject who does A can successfully defend it as "the action that made the most sense for him to do" and the reasons that provide the justification for the normative judgment make the propositional contents of those normative judgments the ones that are reasonable to include in deliberation.
This is interesting and I agree with what you say. I do have small nitpicky point about this response:
“Coop’s counterpart did not know that pushing the button causes the baby to be released. Why didn’t he know? Because an epistemic counterpart could have just his evidence for believing this proposition and be mistaken where this mistaken belief would rationalize an action that the agent cannot perform. So, S knows p only if S has no epistemic counterparts in any possible world who falsely believe p and whose belief that p would rationalize forming the intention to perform an action the agent cannot perform under those circumstances.”
I don’t think that anyone who takes this reliabilist line has to take it to such extremes. One doesn’t need be able to distinguish p from not-p in all possible worlds and in all possible circumstances. Rather, I thought the idea is that one has to be able to tell the difference between p and not-p in resembling circumstances in which one could easily be in. What are the circumstances one could easily be in is supposed to depend on what the actual world is like. Are there actually both many machines with broken mechanisms and ones that work around? I presume there are. This is why S would not know as he couldn’t tell the difference. But, this wouldn’t threaten our ability to know lots of things as for many things we can reliably classify things in circumstances we could easily be in (even if we could not do this in all worlds).
Hey Jussi,
That seems exactly right. In fact, I hope that’s how people react. There’s no way that this could be true:
(K) S knows p only if S has no epistemic counterparts in any possible world who falsely believe p and whose belief that p would rationalize forming the intention to perform an action the agent cannot perform under those circumstances.
But, it seems there’s an argument from The View to (K). We ought to reject (K) and the view along with it.
(1) In the actual world, @, Coop cannot release the infant and falsely believes on very good evidence that when he puts in his quarter and pushes the button to release the infant the infant will be released.
(2) In some counterfactual scenario, w1, Coop has just the same evidence for his relevant beliefs as he does in @, he can release the infant, and (assume) knows that that when he puts in his quarter and pushes the button to release the infant the infant will be released.
According to (SE1), Coop has sufficient evidence to believe that he ought to release the infant by pushing the button in both @ and w1. According to (SE2), Coop ought to believe that he ought to release the infant by pushing the button in both @ and w1. According to W-Con, he ought to release the infant in both @ and w1. According to OIC, it’s not the case that he ought to release the infant in @. It’s seems like a pretty good reductio of The View and this sort of reasoning is what leads from The View to (K).
I think we’re on basically the same page epistemologically. For roughly the reasons you cite, I wouldn’t accept both (SE1) and (SE2). But, people keep saying that it’s crazy to give those up and that (W-Con) is perfectly harmless. I’m supposed to give up on the externalist view that I like that dispenses with (SE2). Well, maybe I should, but that’s not because The View is better than my view.
Hi Clayton,
Interesting problem. I have to prep for class so I can’t read the whole thing carefully, but do you address the suggestion that we reject W-Con in favor of this:
(W-Con-Intend) SO(SBOΦ → SO(Intend-to-Φ))
Hey Brad,
I think that could be a move in the right direction, sure. I actually need to do some prep for class as well, but let me say that there are two things about the move from (W-Con) to (W-Con-Intend) that worry me. Both worries might be minor, dialectical worries.
First, the arguments and intuitions that are supposed to motivate (W-Con) need to be re-examined if we’re going to replace (W-Con) with (W-Con-Intend). Second, there are worries about the relationship between reasons that bear on whether to intend and reasons that bear on whether to act.
I’ll come back to this later (I hope), but consider @ and w1 from my response to Jussi. Suppose in w1, Coop ought to release the infant by putting in the quarter and pressing the button. Suppose in @, that’s not what Coop ought to do, so he ought to do something else. If we say that he he ought to believe that he ought to intend to release the baby in @ and w1, but what he ought to do differs, we get the strange result that he ought to intend to do what he oughtn’t do in @. (This seems to violate a principle I like, which is that reasons that bear on whether to A and whether to intend to A are the same reasons.) This can be avoided by pursuing the he ought to try to act as he does in @ in w1, but given the epistemological commitments of The View, this still seems to generate some unattractive epistemological consequences.
So, I don’t hate (W-Con-Intend), but I don’t know if those who defend The View could retreat to this.
Clayton,
ah I see. I think I was slightly confused about the dialectic (perhaps I am still). Someone might though also try to save (W-con), (S1) and (S2) by being an externalist about justification as well. That might make (2) in your response false. So, they might think that Coop doesn’t have the same evidence either in @ and the counter-part world. In fact he has less evidence in our world, given that whatever information he has here does not make it any likelier that he will be pushing a button that works given how many failing machines there are around. This is why he is not in the same state as someone who knows (and whose evidence is sufficient to tell the working machines apart because there are no failing machines in that world). This would explain why he does not have sufficient evidence and why he should believe what he believes. This would avoid the absurd conclusion even if (S1) and (S2) might still be true. I don’t know how plausible or implausible externalism about evidence is but at least it might save the View.
Hi Clayton,
Interesting. I will have to think about your worries…
For what it is worth: I was thinking, in part, of a more general view that reasons are best understood as reasons for judgment sensitive attitudes, and that thus, strictly speaking, there are no reasons to act. By extension, there is, strictly speaking, never an act that we ought to perform.
However, on this neo-Scanlonian view, there *are* reasons to intend to perform an act of a certain type and there are types of acts of which we ought to intend to perform a token — & our ordinary talk of reasons for actions should be taken as referring to these reasons.
This *seems* to eliminate your second worry…but maybe those who hold the View you are after would not like that position…
Brad
Hey Brad and Jussi,
Have to run to a different class, but in response to Jussi.
Good point. I happen to be a kind of externalist about evidence (e.g., I think that evidence consists of only true p’s and any p we know non-inferentially) and I know that there are those who are even more ‘externalist’ than I (e.g., Williamson) who don’t restrict evidence in this way to that which is known without inference. It makes it tough to evaluate their views (as evidenced by the exchanges in the OUP volume on Williamson (where I think he wins just about every exchange)), but I think it’s independently plausible to think that this externalist maneuver won’t help The View that much if we say this:
(*) A subject’s justification for believing p won’t involve p itself if that belief is only inferentially justified.
So, while we might not be able to duplicate our subjects’ evidence in @ and w1, I think we can duplicate the evidence that serves as the basis for these subjects’ beliefs.
Hi Clayton,
This is a nice worry. Matt and I had a conversation about this, and I think we generally agree on what to say in response. We have three basic points. (Sorry, this is a bit lengthy.)
1. If you’re using us as representatives of The View, we resist that label, and not just for the reasons you point out in the footnote. Rather, we’re suspicious of W-Con and the inference from (1) to (2) (and we explicitly say things to this effect in the first appendix of the book). If one opts for fallibilism (or, rather, denies factivity) about knowledge-level justification, then we don’t see the appeal of special pleading for ‘oughts’. If you can have knowledge-level justification for falsehoods, then you can have
knowledge-level justification for false propositions about what you ought to do.
2. What McGrath and I don’t like is that only true propositions or facts can be reasons Coop has. But this view is not the source of the problem in the Coop case. For here are some true propositions/facts which Coop knows: an infant’s life is worth more than a puppy’s, a kitten’s, my having a gumball, or my keeping the quarter; and so if I save the infant, I’ll be promoting more good than I would if I save the puppy,the kitten, etc. These facts seem to support saving the infant. Does this mean Coop ought to save the infant? Everyone should have some stance on the answer to this question. What options are there for those who want to deny that Coop ought to save the infant? Here are two: 1)perhaps they can say that these aren’t really reasons Coop has to save the infant (not because they’re false, of course: they’re all true). Maybe this is simply because the action can’t be done (and one can’t have reasons to perform an action one can’t perform). 2) Or perhaps they can say that these are reasons Coop has to save the infant, but the
fact that the action can’t be performed defeats them and so prevents the reasons from making it the case that Coop ought to save the infant. Notice, though, that McGrath and I can make these very sorts of moves about whatever false propositions appear to support saving the infant
(such as saving the infant is the best thing I can do or, assuming oughts can be reasons, I ought to save the infant). We can say that these falsehoods either (a) fail to be reasons Coop has to save the infant, because that action is unavailable to him, and a consideration can be a reason for doing something only if that action can be done, or (b) are reasons for Coop but are defeated by the fact that the action is
unavailable. In this way, we, too, can avoid having to say that Coop ought to save the infant. These false considerations, we could say, are rather reasons only for Coop to try to save the infant or to decide to save the infant or to take steps toward saving the infant, and not for him to save the infant.
3. Here’s something we think is pretty solid: Coop ought to decide to save the infant. If justified falsehoods can be reasons Coop has, as
McGrath and I think, this result can be secured. For Coop has a very strong reason to decide to save the infant, viz.: saving the infant is something I can do and better than anything else I can do. But, if false propositions can’t be reasons then, because this is false it’s not a reason. The closest true proposition in the vicinity is that saving
the infant is something Coop *probably* can do and it is better than the other options. The danger, though, is this: once you deprive Coop of “false” reasons the “true” reasons that remain might not be strong enough to ground the
right “oughts”.
Hey Jeremy,
Thanks, you’ve given me a lot to think about. I’ll need to respond in shifts.
I probably didn’t make this sufficiently clear in the post, but there’s The View and there’s The View-light. I took you and Matt to be defending something in the neighborhood of the view, in part, because of remarks like this concerning the gin/petrol example, “… it is highly plausible that if two subjects have all the same very strong evidence for my glass contains gin, believe that proposition on the basis of this evidence, and then act on the belief in reaching to take a drink, those two subjects are equally justified in their actions and equally justified in treating what they each did as a reason, even if one of them, the unlucky one, has cleverly disguised petrol in his glass rather than gin.”
I took you to be saying that if p is justifiably believed (w/knowledge-level justification), p is justifiably acted upon where that means that there’s no practical obligation to refrain from acting on p. Rather than endorse:
(W-Con) SO(SBSOΦ → SOΦ)
Here’s a possible counterexample to (W-Con) the view I thought you were committing yourselves to avoids. SBOΦ but Φ-ing is permissible and refraining from Φ-ing is permissible. Provided that someone can be obliged to believe that they ought to Φ, such cases are counterexamples to (W-Con).
Something safer avoids that view:
(WeakWide) SO(SO~Φ → S~BSOΦ)
The idea is that if S believes she ought to Φ and it’s not the case that S ought to refrain from such a belief, it’s not the case that she ought to refrain from Φ-ing. The gin/petrol passage suggested that you thought that if it’s permissible to believe that you ought to Φ, it’s permissible to treat _that you ought to Φ_ as a reason for action (e.g., when reasoning to the necessary means, sufficient means, etc…). I can’t tell whether you want to give up (WeakWide) as well, but I suspect that given a fallibilist view of justified belief, (WeakWide) will generate problems structurally similar to (W-Con). No special pleading for ‘oughts’, but special pleading for ‘mays’?
This touches on an issue that I’m interested in. Some seem resistant to the idea that justifiably/permissibly believing p entails justifiably/permissibly treating p as a reason for action for some p-dependent choice because the _kind_ of justifications differ. As I read your remarks here:
“When trying to determine what is true – that is, in forming beliefs – we draw conclusions from the reasons we have. The same goes for trying to decide what to do. Here, too, we draw conclusions about what to do – we form intentions – from the reasons we have. We bring reasons into our reasoning knowing that we might draw all sorts of conclusions from them along the way, some practical and some theoretical” (forthcoming: 78)”
I thought the view you were recommending was something stronger than the view that says it is mere epistemic propriety at issue. When you justifiably believe p and take p to count as a reason in favor of Φ-ing, it is neither epistemically nor practically improper/impermissible/unwarranted/wrong/unjustified to do so. Anyway, it’s hard for me to assess the view, in part, because I thought you were committing to something like this and figured that this sort of view naturally supports (WeakWide) given some further assumptions about reasons and oughts.
In your remarks you said, “If you can have knowledge-level justification for falsehoods, then you can have knowledge-level justification for false propositions about what you ought to do.” That’s an interesting move. I had two questions here. First, would you be willing to say both:
(i) If you can have knowledge-level justification for falsehoods, then you can have knowledge-level justification for false propositions about what you ought to do.
(ii) If you can have knowledge-level justification for falsehoods, then you can have knowledge-level justification for false propositions about what you ought to believe (including beliefs about what you ought to do).
I guess if you accept both (i) and (ii), you could still hold onto certain wide-scope principles linking permissible belief about what ought to be done with permissible doings in accordance with belief.
Second, when you have the false, justified belief: [Φ-ing is either permissible or mandatory], can you justifiably Φ and and justifiably treat this (false) proposition as a reason for action (e.g., in reasoning towards the necessary means to Φ-ing)?
_________________
Concerning (2).
“What McGrath and I don’t like is that only true propositions or facts can be reasons Coop has.”
Yes. I, predictably, love the opposite view, but I wanted to push things in the OIC direction to avoid sinking back into those debates. I suspect (but I could be wrong) that there’s very little that I’m going to say that is going to move you to the view that only a true proposition/fact/actual state of affairs can constitute a (normative) reason. I think that like you I would love to know what someone can say about what happens to the reasons when you can’t. One idea is that the reasons aren’t reasons when they aren’t reasons you can act on. This brings me to another question about your view that I’ve been dying to know the answer to.
You and Matt give a really nice rational for the principle that knowledge-level justification for believing p gives you justification for treating p as a reason. I noticed (and note in a footnote in some other work) that there are two ways you can go here. You can say:
(JB1) If S’s belief that p is justified, p is a reason that S can justifiably include in S’s deliberations.
(JB2) If S’s belief that p is justified, S can justifiably include p in S’s deliberations because whether or not p is a genuine reason, this won’t affect the normative standing of S’s treating it as such.
It seems that if you go for (JB2), we can say that there’s the question about what the reasons are and a separate question about when we’re entitled to use them as reasons for the purpose of justifying various things and the right to treat something as if it is a reason does not entail that it is a reason. No special pleading. Reasons are like currency. You can permissibly treat something as currency that isn’t provided that you justifiably believe that it is the genuine article.
_____
I will try to come back and say more about (3), but this has been immensely helpful and rewarding (to me), so thanks again. These issues have been bouncing around in my head for months now and working through your manuscript has been a whole lot of fun.
John’s view, though, totally wrong. I mean, the arguments for it are clever and original, but you just have to have a much crazier view than he’s willing to sign on for if you plan on wielding (W-Con).
Hi Clayton,
I’ll jump in here. Your response to Jeremy brings up a number of interesting questions. I will consider several of these (apologies for length).
1. What is the relationship between talk about reasons and talk about appropriately treating something as a reason?
Jeremy and I tend to talk about having a reason rather than appropriately treating something as a reason. But you are right to ask about the relation between the two, and between these two and epistemically qualified statuses like p’s being warranted enough to be a reason you have to PHI. We are sympathetic to the following claims:
i. If p is a reason you have to PHI, you are appropriate to treat p as a reason to PHI.
ii. If p is warranted enough to be a reason you have to PHI, you are epistemically appropriate to use p as a reason to PHI.
We would not want to drop ‘epistemically’ in (ii). Suppose you are grading and you know a student likes Neil Diamond. You cannot stand Neil Diamond or those who like his music. Is it appropriate to treat the fact that the student likes that music as a reason to flunk him? That sounds wrong, but for reasons irrelevant to the epistemic status of the fact for you. So, we would not accept JB2 in your last comment.
2. Do we have to include a qualification ‘epistemically’ or ‘warranted enough’ in any proposed connection between epistemic and practical justification?
In the book, we do affirm some relationships between knowledge and action without those qualifications. We say:
If you know R is a conclusive reason there is for you to PHI, then you have a conclusive reason to PHI, viz. R.
If you know that there is most reason for you to PHI, then you should PHI.
But we do not claim these hold for justified belief, because of worries about act-availability. Those worries would also spoil the corresponding wide-scope rational requirement to be such that if you believe R is a conclusive reason there is for you to PHI, then PHI. We are sympathetic to a similar requirement involving intending to PHI, as some Brad suggests. However, in some rather wild cases even the intentions might not be available, and so perhaps we need to retreat to a requirement like this: be such that if you believe R is a conclusive reason there is for you to PHI, then you don’t intend not to PHI.
3. If you are justified in believing that you ought to PHI, and justified belief is all you need to make something warranted enough to be a reason to PHI, how could it fail to be true that you ought to PHI?
Suppose we think of oughts as grounded in reasons. What you ought to do is what you have most undefeated reason to do, or something like that. Here is the answer to the question of (3), then: it could be that you ought to PSI, rather than PHI, because your reasons to PSI beat out, perhaps just barely, your reasons to PHI, but you are nonetheless justified in thinking your reasons favor PHI-ing, and so justified in thinking you ought to PHI. You might be justified in this false belief because you miscalculated the reasons or relied on the normally reliable word of someone else who knew your reasons, etc.
4. Finally, a clarification about the gin/petrol case. We discuss this case as part of an argument that one can have something as a reason even if it is false. We take it to be plausible that the guy in this case is perfectly reasonable/rational/justified in reaching for a drink, despite the fact that it contains not gin but very well-disguised petrol. If you can only have true reasons, it gets harder to find sufficient reasons to ground this reasonableness. One can try turning to the likes of I think it’s gin or it looks like gin, but there are two problems here. One is that these look like weaker justifications than this is gin. But it seems the guy with petrol in his glass is exactly as reasonable to reach for a drink as the guy with gin. (Mark Schroeder makes these and related points in his really nice paper, “Having Reasons.”) The other is that it is pretty clear that what motivates the guy to reach for a drink is not any consideration about what he is thinking or even about the appearances, but simply the consideration that, mmm, here is a glass of gin! So, his motivating reason – the consideration that moves him to reach – is a falsehood. And since the very same linguistic oddities that are used to raise doubts about whether normative reasons-possessed can be falsehoods are found in the case of motivating reasons, too, this is encouragement to think that those oddities do not show us that normative reasons-possessed must be true. (The statement ‘The reason for which he reached for the glass was that it contained gin’ sounds no better and no worse to our ears than ‘He had a reason to reach for the glass, namely that it contained gin’.)
Matt,
Thanks for your extensive comments, you and Jeremy have given lots to think about. Just to clarify a little, if I’m reading you correctly you think that it’s possible for situations to arise where:
(i) S’s belief that p is justified and false;
(ii) S is epistemically permitted to treat p as a reason for action (for choices that S knows to be p-dependent);
(iii) S is _not_ practically permitted to act on p (i.e., do that which S ought to do if p were true where S knows this is what S ought to do if p);
(iv) (i)-(iii) can hold if ‘p’ is the proposition that the agent is permitted to A.
I think part of what makes this confusing is that I’d think if (i)-(iv) could be satisfied, the gin/petrol case would be the case where these conditions are satisfied. But, you seem to insist that he’s permitted/justified to drink the petrol (share the petrol with friends?) and that permission/justification seems practical. Suppose our agent reasons as follows:
I’ve promised A I’d get us a gin drink. This is a gin drink and I’m permitted to drink it. So, we should drink this one
As our drinker and his partner head to the hospital, I say that he should be excused for giving the petrol to his partner because he reasonably thought that it was gin. I’m happy to say that and no more. So, I’m happy to say that his relevant normative judgment (i.e., that it’s permissible to drink the stuff he shared with A) is false. (I’m not going to say that it’s justified/permissible/consistent with discharging his epistemic duties, etc…). You say that’s too weak–what he did was permissible. But, then his normative belief isn’t justified and false. It’s epistemically justified and true.
Hi Clayton,
Are you wondering whether it is ad hoc for us to distinguish the two cases in the way we do – the gin/petrol case and the case in which you have a false but justified belief about what you are permitted to do?
Let me spell out the thinking about the two cases, and maybe the reasoning won’t seem as ad hoc. Concerning the gin/petrol case, yes, we think that the person has every reason to drink and has no good reason not to drink, and so since drinking is best supported by his reasons, he should or is at least permitted to drink.
By the way, I think even if one’s reasons must be true, the subject would have still have good reasons to drink (or to give his friend a drink) and have no good reasons not to. Think about what the subject knows: this is very likely gin; I’d really like some gin; there is almost a zero chance this is petrol disguised as gin. Knowledge of those facts makes it the case that the subject has much stronger reasons to drink than not to. So, I think that even you, Clayton, could agree that the subject has much better reason to drink. I of course admit that there exists a reason not to drink – it’s petrol – but I’d insist that the subject doesn’t have or possess that reason, and so it can’t matter to what he is permitted to do. Maybe you want to reject that claim. But notice that whether only possessed reasons matter to permissibility is independent of whether possessed reasons have to be true.
So, I guess I think that if only possessed reasons can matter to permissibility, the guy with petrol in his glass is permitted to drink – irrespective of whether possessed reasons must be true. Clayton, do you agree about this much?
The point about normative beliefs is quite different. We think, on general fallibilist grounds, that it is possible to have justified false normative beliefs – beliefs about what one ought to do. How could this be so? We think that what you ought to do is a matter of what reasons you have, and your reasons could favor PSI-ing even if you justifiably think they favor PHI-ing. (It’s crucial that it’s the reasons you have or possess that matter to what you ought to do – not the facts beyond your ken.) That’s why you would think that you should PHI, even though you should PSI.
One worry here is why, on our view, *I ought to PHI* isn’t itself a reason you have to PHI. It’s false, but it is justified, and, on our view, its being justified makes it warranted enough to be a reason. Is it a reason? We don’t take a stand. We say only that it doesn’t defeat the reasons making it the case that you ought to PHI. So, perhaps it’s a defeated reason, or perhaps it is no reason at all. The idea that *I ought to PHI* — if it is a reason at all – is a different and derivative kind of reason isn’t very radical, especially if we think of oughts as grounded in the overall balance of the reasons a person has.
The two sorts of cases don’t seem to raise the same issues. The gin/petrol case really has nothing to do with normative beliefs. The person seems permitted to drink not because of the justified normative belief but because of the first-order reasons he has to drink.
Does all of this come closer to addressing your worries?
Okay, it’s still flipping out. I’ll try to break this up into chunks.
Hi Matt,
Thanks for taking the time to walk me through this, I’m sorry that I’m still either confused or not convinced or both. Part of me was worried that it was ad hoc, but if you want to say that the cases of false, justified normative judgment are different I can’t think of any principled reason to think you can’t say that.
As for the gin/petrol case, I don’t deny that you can get the verdict that the agent should drink the stuff in the glass given facts to go on, propositions to go on, etc… You could say that all reasons are facts but say that only facts that pass through an epistemic filter bear on whether someone should A. (In other places, I would argue that you shouldn’t drink the stuff in the glass or share it with others and that this verdict you can’t get unless you think reasons consist of facts, but that’s for another post/paper.)
Suppose you’re right and “only possessed reasons can matter to permissibility”. A case that you suggest shows that the justified belief/justified action link is severed is one like this:
“We think that what you ought to do is a matter of what reasons you have, and your reasons could favor PSI-ing even if you justifiably think they favor PHI-ing. (It’s crucial that it’s the reasons you have or possess that matter to what you ought to do – not the facts beyond your ken.) That’s why you would think that you should PHI, even though you should PSI.”
Let me try to make this concrete.
(i) Suppose you know that p is true and know p counts in favor of going to Austin.
(ii) Suppose you know q is true and know q counts in favor of going to Boston.
(iii) Suppose you believe falsely but with knowledge-level justification that neither of these moral reasons is more demanding than the other.
(iv) Suppose you know you made an old friend the following conditional promise: that you would go to Boston if duty permitted.
(v) Suppose further that your judgment about the comparative strength of the reasons you are aware of is way off—the reason to go to Austin is much stronger.
I take it that you want to say the following:
(1) It’s wrong in some practical sense for you to go to Boston, you should go to Austin.
(2) It’s epistemically permissible for you to treat that you are permitted to go to Boston as a reason for action when you reason from two things justifiably believed by you (one of which is known):
There’s no more reason to go to Austin than Boston, so I’m permitted to go to Boston. So, to keep the promise, I must go to Boston. I ought to go to Boston. I shall go to Boston. I ought to buy a ticket, etc…
It seems to me incoherent to me say that it’s proper to treat _that I am permitted to go do Boston_ as a reason for judging that you should go there when you can’t permissibly act on that premise. Note that if you accept (1) and (2), the case is a counterexample to the principle that p is justifiably acted upon (for some p-dependent choice) when justifiably believed. But, it seems then that in light of the reasons that the agent was aware of the agent ought to act against her own judgment about the thing to do in spite of the fact that it’s either not the case that she should revise her views about the thing to believe or positively oughtn’t revise her beliefs about the thing to believe. Again, more weirdness. This is an ought grounded in reasons she’s aware of. If you’re going to say that she shouldn’t act on her judgment because of what the reasons she’s aware of demand of her, why is she to believe she should act that way anyway?
Suppose you press our agent and say that she should have gone to Austin. She defends herself and says that she gave the matter serious consideration. The reasons to go to Austin didn’t seem any more important than those that demanded that she be in Boston. She notes that lots of reasonable people in her shoes make the same choice she did, etc… I would say “Here the unlucky subject is defending it as the action that made the most sense for her to do and the proposition that made most sense to treat as a reason and she defends her judgment as the judgment that made most sense for her to make”. If her judgment was unreasonable, I could see saying that neither the judgment nor the action was excusable. But, if her judgment was unreasonable, her judgment wasn’t justified. If the judgment about the thing to do is reasonable, that ensures that the action was reasonable. Don’t you say in the petrol/gin case that the subject isn’t just offering some excuse because they can show, “the unlucky subject is not providing an excuse for his action or treating what he did as a reason; he is defending it as the action that made the most sense for him to do and the proposition that made most sense to treat as a reason.”
I submit that any case where you justifiably judge that A-ing is the thing to do is a case where some subject who does A can successfully defend it as “the action that made the most sense for him to do” and the reasons that provide the justification for the normative judgment make the propositional contents of those normative judgments the ones that are reasonable to include in deliberation.
Hi Clayton,
In each of the gin/petrol case and the normative beliefs case, there is an act, A, that is intuitively the one favored by what are intuitively the reasons the subject has. In the gin/petrol case, A is drinking what’s in the glass. In the false normative belief case, A is the act that the subject falsely justifiedly believes she ought to do. The difference between the cases is this: in the gin/petrol case, we’re trying to save the result that the subject ought to do A, while in the false normative belief, we’re trying to save the result that it’s NOT the case that the subject ought to do A. We think that all views should aspire to save these results because, intuitively, the subject ought to drink what’s in the glass and, demonstrably, it’s not the case that the subject with a false justified beliefs ought to do A.
So, in the gin/petrol case, we’re trying to save the result that the subject ought to do what intuitively her reasons suggest, while in the normative belief case, we’re trying to save the result that it’s not the case that the subject ought to do what intuitively her reasons suggest. So, it’s easy to save the result that subject ought to drink what’s in the glass because whether one claims that the subject’s reasons can be false or not, all reasons the subject has favor drinking. (But of course this wasn’t the point of the case, anyway. It was to show that the subject can have false reasons — namely, that there’s gin in the glass.)
It’s less easy to save the other result. But this is complicated anyway, because it’s difficult to know what to say about the possibility of false justified beliefs about what one ought to do. We’re suspicious of any claim that they’re impossible, though we don’t really commit either way. But it should be clear that the cases are different, because the intuitions needed to be saved are different: 1) for standard fallibilist reasons, a subject can have a false justified belief that she ought to do A (no special pleading) and 2)it’s not the case that such a subject ought to do A.
How can we have both? If a subject’s reasons do not favor A-ing, but the subject has strong evidence that they do, then the subject might have a false justified belief that she should do A. Here the subject, as you point out, will be able to defend her A-ing. But that alone isn’t what makes A-ing justified. It’s not what makes drinking justified in the gin/petrol case. What makes drinking justified is that her reasons in fact heavily favor drinking — not merely that she can defend it.
Of course, you might say that as soon as she has strong (justifying) evidence that her reasons favor A-ing, that evidence is going to get folded into her reasons and automatically make A-ing justified. That’s what we’re suspicious of. It might be right. But we don’t really see why it would automatically be the case: maybe the reasons for not-A-ing are so strong that even in conjunction with the reasons for A-ing and the strong evidence that A-ing is justified, they win out.
Hey Jeremy,
Again, thanks for the time, patience, and another response. It’s given me a lot to think about. (Speaking just for myself, I quite enjoy reading through your responses and Matt’s, but I can imagine that you might someday in the very distant future get tired of this, so I apologize if I drag this out too long.)
Let’s focus on that case of a (n alleged) justified false belief about the comparative strength of two reasons the subject has. You write:
If a subject’s reasons do not favor A-ing, but the subject has strong evidence that they do, then the subject might have a false justified belief that she should do A. Here the subject, as you point out, will be able to defend her A-ing. But that alone isn’t what makes A-ing justified. It’s not what makes drinking justified in the gin/petrol case. What makes drinking justified is that her reasons in fact heavily favor drinking — not merely that she can defend it.
Good point. I think I agree with part of what you’re saying, but I worry about a view that says that the subject can successfully defend her A-ing when her A-ing isn’t justified.
It’s not that I want to say that _the defending_ is what makes A-ing justified (or the successful defending makes it justified), it’s that whatever it is that figures in the defense will do the justifying _if_ the defense is successful. So, if there really are reasons/evidence for a judgment about the strength of available reasons that bear on what to do, why wouldn’t _that_ be what determines whether a proposition about comparative strength is permissibly treated as a reason for action/inference as opposed to the facts about comparative strength that the subject doesn’t know and can be perfectly reasonable while failing to know them?
Of course, you might say that as soon as she has strong (justifying) evidence that her reasons favor A-ing, that evidence is going to get folded into her reasons and automatically make A-ing justified.
Some could go that route directly, but I think they could try to get there indirectly. Here’s my Austin/Boston example:
(i) Suppose you know that p is true and know p counts in favor of going to Austin.
(ii) Suppose you know q is true and know q counts in favor of going to Boston.
(iii) Suppose you believe falsely but with knowledge-level justification that neither of these moral reasons is more demanding than the other.
(iv) Suppose you know you made an old friend the following conditional promise: that you would go to Boston if duty permitted.
(v) Suppose further that your judgment about the comparative strength of the reasons you are aware of is way off—the reason to go to Austin is much stronger.
Why can’t someone say that whatever it is that makes (iii) true (e.g., the apparent strength of the reasons where she knows that these reasons are reasons where apparent strength is determined by intuitions/intellectual seemings), the subject has the justified belief that neither of the reasons to go to A or B is stronger. If she has knowledge-level justification for judgments about strength of reason and comparative strength of reasons, it seems she can justifiably treat (false?) propositions about comparative strength as reasons.
If this is the way things go, it’s going to be harder to describe cases where a subject has justified, false beliefs about comparative strength of reasons. If I’m permitted to treat the proposition that R1 is stronger than R2 as a reason for action (or that neither R1 nor R2 is stronger), can it nevertheless be the case that I’m obliged to act against the normative judgments I know to be entailed by these judgments about comparative strength?
[I know from earlier there was some skepticism about the idea of treating certain kinds of normative propositions as reasons for action, but I think we can have cases where even the propositional contents of verdictive judgments serve as reasons for action (e.g., that I ought to A gives me a reason to do what I know must be done to A and reasons that bear on whether to believe known consequences of the fact that I ought to A; it could be that I’ve made a promise to wear a funny hat whenever the reasons I’m considering are such that neither is stronger, so I can reason from the premise that R1 and R2 are equally stringent to the judgment that I should wear my funny hat, etc…]