John Gibbons. In both the practical and theoretical realms, there are, or might be,
two kinds of reasons. If there's a fire in the basement of your hotel, even if you ...
Guidance By Reasons John Gibbons In both the practical and theoretical realms, there are, or might be, two kinds of reasons. If there’s a fire in the basement of your hotel, even if you have no evidence of a fire or lots of evidence that there’s no fire, you have an objective reason to jump out the window into the canal below.1 These are sometimes called normative reasons. If what you’re genuinely or normatively required to do is determined by these objective reasons, then you really ought to jump out the window. Of course, in the story as told, you have no subjective or motivating reasons to jump. And this doesn’t just mean that you don’t have the kind of reasons that will cause you to jump. You don’t have the kind of reasons that would rationalize or make sense of jumping. What’s missing in the story are the kinds of reasons that would make it reasonable to jump. And people who like normative reasons are perfectly happy to point this out. You ought to jump out the window even though that would be completely irrational. So I take it as fairly obvious that if objective reasons are normative in the sense that they determine what you’re genuinely required to do, then subjective reasons and rationality are not normative in that sense. And you don’t need to look at the details of Detachment, or really anything else, to see this.2 All you need to look at is the story that gives you the idea of what an objective reason is. There are many such stories, but here’s what they have in common. You ought to φ when you’re justified in believing that you shouldn’t. You ought to φ when φ-ing would be irrational. And neither the fact that φ-ing is irrational, nor the things that make it irrational (the evidence and so on) count against φing, where that just means that they don’t give you a normative reason to avoid it. And neither the rationality, nor the things that make for the rationality, of some alternative count in favor of the alternative, where that just means that the facts about rationality are irrelevant to the determination of what you’re genuinely required to do. So I don’t think you need anything other than the stories to see that if objective reasons are normative, then rationality is not. So there is a bit of terminological nonneutrality in calling objective reasons normative reasons. If we think there’s a real question of what the person in the story is genuinely required to do, then we must choose between the objective and subjective reasons. I think this is a hard question, and we should not be happy with easy answers from either side. If you start with objective reasons, you can show that rationality isn’t normative. If you start with subjective reasons, you can show that it is. This is easy. The hard question is where we should start, and this brings with it the question of how to decide. If the question of what you’re genuinely required to do simply were the question of what would be best, then I do think that irrationally saving your life would be far better than going down in flames while keeping your mental house in order. I just don’t think it’s obvious that that’s what the question comes to. If there were a norm that told you to be irrational when that would be for the best, it would just be an accident when you act in accord with the norm. You can’t be irrational for that reason in any sense. There seems to be some sense in which this norm could never get a grip on us or guide us in the way we might think that norms are supposed to grip us and guide us. 1
Parfit, (1997). Cf. Kolodny (2005) where he seems to think that you need a great deal in order to get from the normativity of objective reasons to the conclusion that you have no reason to be reasonable. 2
2 You get the same thing in the theoretical case. Here are some sentences that look on the surface as though they could conceivably have at least one interpretation on which they have something to do with when you’re genuinely required to believe things. We assume always and everywhere that the question of p comes up and it matters to you. (T) Necessarily, for all p, you ought to believe that p only if p. (J) Necessarily, for all p, you ought to believe that p iff you’re justified in believing that p. (K) Necessarily, for all p, you ought to believe that p only if you would thereby know that p. If we are able to read these as making claims about what you ought to believe, it’s clear that they make competing claims. Take a case of justified, false belief, like the case of the missing keys where you’re justified in believing that your keys are where you left them, but, unknown to you, someone snuck in and stole them. In this case, (T) entails that it’s not the case that you ought to believe that p. And (J) entails that you ought to believe that p. So it looks like you have to choose between them. If you reject (T) on its normative interpretation, even if you’re perfectly willing to accept it on sixteen other interpretations, then it looks like you’re saying that when it comes to belief, truth is not genuinely required. And if it’s not required, then it looks like it’s optional. But if you really believe that it’s optional, what could be wrong with thinking that you don’t have it? So what could be wrong with believing the following conjunction? I believe it’s raining, but it’s not. The problem with believing that Moore-paradoxical thing is that it’s irrational, incoherent, or maybe even absurd. So it looks as though rejecting the truth norm on belief might be mildly problematic. This seems to go for justification as well. If you reject (J) on its normative reading, it looks like you’re saying that justification is optional. So what could be wrong with believing the following sort of thing? The store is open, but I have no reason to believe that. This looks Moore-paradoxical to me. And as a subjectivist, it looks akratic. You do one thing, believe that p, while thinking you have better reason to do something else instead, either withhold or deny. So rejecting (J) will make some people uncomfortable. If you’re an objectivist, you’ve inured yourself to the idea that there’s nothing wrong with being an idiot. So if you find out, perhaps on the authority of your psychiatrist, that you really do believe the store is open, all the while fully recognizing that you have no evidence for this, then you see yourself as irrational. You just don’t think there’s anything wrong with that. But things are not so comfortable for the subjectivist. If rejecting (T) does sanction believing Moore-paradoxical things, and if that’s irrational, then it looks as though subjectivism is problematic by the subjectivist’s own lights. And if, in the theoretical case, we think of being reasonable in terms of trying to get to the truth, then it looks like the subjectivist is saying that you have to try your best to live up to a norm you know you don’t have to live up to. And that should look problematic by anyone’s lights.
3 Still, the idea that you ought to be reasonable doesn’t look like a crazy idea, and I don’t think we should take its negation as the default position. I’ve argued earlier that you can’t derive our subjective requirements from our objective requirements, or explain the subjective in terms of the objective. Despite the initial appeal of this picture, if you start with objective reasons, you’ll never get the normativity of rationality. The question is where to start. For our purposes, the question is whether there’s anything to be said for starting with subjective reasons, the kind of things that make things reasonable. What, aside from making things reasonable, do these things have going for them? What I call the natural reaction is a very common reaction among the epistemologists, though it doesn’t seem nearly as common among the ethicists. This is the idea that the objective norms can’t grip us or guide us because they’re inaccessible to us in some sense. This paper is part of chapter six of a book I’ve been working on. This is the part where I try to figure out what the natural reaction comes to, what it commits us to, and whether it can give us some leverage against the objectivist. I think it does come to something. It might give us some leverage. And I go on to argue later that it does not commit us to very bad things like luminosity and internalism about justification. The Natural Reaction The objectivists think that you ought to believe when the keys are there and withhold or deny when they’re not. I think there’s something intuitive about the natural reaction to this idea, but it’s not at all obvious exactly how we should put it. We want to say that in the relevant cases, these alleged requirements are simply beyond our reach. If we did manage to comply with the objectivist’s norm, that would just be an accident. The objectivist’s “ought” is not the kind of “ought” that can get a grip on us in those cases. So they’re not the kinds of things for which we can be held responsible. So they’re not really the kinds of things that can be required of us. So whatever the objectivist’s “ought” means, it’s not really expressing the idea of a requirement. I think that the main idea behind the natural reaction is that justification is a normative notion. It’s not merely an evaluative notion. So if you’re trying to explain the significance or importance of being reasonable, value, whether intrinsic or derived, is the wrong place to look. When you’re thinking about value or what would be for the best, you’re just thinking about the outcomes. How you get there doesn’t matter. Whether you get there for reasons doesn’t matter. And if you do get there for reasons, it doesn’t matter if they’re good or bad. If there’s a fire in the basement, jumping out your window would be for the best. If the fact about the value of jumping can’t get a grip on you because you have no way of knowing about the fire, this is unfortunate. But it doesn’t make jumping any less valuable. If you don’t jump out your window, we won’t hold you responsible. And the fact that you had no way of knowing about the fire is obviously relevant in some way or another when what we’re after is a genuinely normative assessment of the action. On the objectivist’s picture, we might not always blame you when you get things wrong. We are, after all, just measly little humans. So when the evidence goes one way, and the facts go the other way, you have an excuse for being reasonable and not doing what you really should have done. This is the most that we can expect from the likes of us. According to the subjectivist, on the other hand, you don’t need an excuse for being reasonable. Being reasonable is itself something that’s required, not the next best thing when you can’t fulfill your real requirements. If we can’t reasonably expect you to jump, or we can’t reasonably hold you responsible for not jumping, this doesn’t mean
4 you have an excuse. It means that you were never genuinely required in the first place. So taking the natural reaction seriously amounts to the idea that there’s some substantive constraint on genuinely normative notions, a constraint that the objectivist’s alleged requirements, like facts about The Good, simply don’t meet. The problem, of course, is to say what that constraint is. We might try putting the natural reaction in terms of the notion of a reason. What could be wrong with believing, out of the blue, that there’s an even number of blades of grass in Washington Square Park if there is an even number? You have no reason to think there’s an even number. And thinking you don’t know about the keys is worse than thinking there’s an even number of blades of grass. You have no reason to think there is an even number, but at least you have no reason to think there isn’t. But in the case of the missing keys, you have every reason to think the keys are in the dining room and every reason to think that you know. If what you’re required to do is determined by what you have most reason to do, and you have no reason to withhold, that couldn’t possibly be required. The problem with putting the natural reaction in terms of the notion of a reason is that there are supposed to be two kinds of reasons. There’s the kind of reason that epistemologists tend to be most familiar with, and these are subjective, motivating reasons. These are the kinds of things that make things reasonable. But there’s also the kind of reason that the ethicists like. These are objective, normative reasons, and they, let’s just say, do something else. On the face of it, it seems that objectivists might be more comfortable, or anyway ought to be more comfortable, with the objective, normative reasons. So isn’t there something sneaky and underhanded about using the idea of subjective reasons in arguing against objectivism? And doesn’t the objectivist have an easy answer? The fact that p is false is a reason not to believe it. It’s not one of those silly old motivating reasons. It’s an objective, normative reason not to believe it. Perhaps there’s another way of putting the natural reaction. You imagine someone in the case of the missing keys with lots of evidence that the keys are in the dining room and no evidence that they’re not. And some part of you wants to say that they can’t withhold judgment. They can’t do anything but believe. And it’s the same part of you that wants to say that you can’t believe that you’re riding a pink elephant, no matter how much I’m willing to pay for that false belief. And once we’ve said that you can’t, we trot out The Principle, which says that ought implies can. The good thing about The Principle is that pretty much everyone accepts it. It is a philosophical claim, so there must be some dissenters somewhere, but generally speaking, it’s pretty safe to rely on. The bad thing about The Principle is what explains the good thing. The bad thing about The Principle is that it’s open to so many interpretations. So The Principle is going to be compatible with just about any theory on some interpretation or another. So there might not, strictly speaking, be any one thing that nearly everyone accepts. And once you try to rely on a particular interpretation of The Principle, you no longer get it for free. On certain value-based conceptions of objective, normative reasons,3 if going to the movie would be valuable, then you have an objective, normative reason to go see the movie. If that would be the most valuable use of your time, then that’s what you have most objective, normative reason to do. So that’s what you ought to do. The fact that you’re justified in believing that the movie is no good does not detract from the value of 3
Raz, (2005).
5 seeing it. So that doesn’t count as a reason not to go. And the fact that the movie is sold out or that it’s not playing in your town does not detract from the value of seeing it. So those aren’t reasons not to go either. So you still ought to go see it. But of course ought still implies can. What’s valuable is going to see the movie, and it’s logically possible for you to go see the movie. And in a possible world in which tickets are available and you have different evidence, it’s even possible for you to go to the movie for reasons. If your interpretation of The Principle is compatible with this theory, then on your interpretation, The Principle doesn’t really amount to anything. It can only be used against people who require you to do the logically impossible. It can’t be used against those who merely require you to go see a movie you have no interest in seeing when the only thing standing in your way is a security guard. And it can’t be used against those who require you to believe when the keys are there and withhold when they’re not. So putting the natural reaction in terms of The Principle requires a particular interpretation of it, just as putting the reaction in terms of reasons requires a particular interpretation of them. Williamson is rightly suspicious of one familiar and traditional conclusion that people occasionally draw from what I’m calling the natural reaction. If we can’t, in some serious sense, proportion our beliefs to the evidence or knowledge, or if we can’t always proportion our beliefs to the evidence, then maybe what we’re really required to do is to proportion our beliefs to what we think our evidence is, or to what we think we know. It’s not our knowledge about the world that ought to guide our beliefs. It’s our beliefs about knowledge that ought to guide our beliefs. This seems to have the consequence that even in the good case, it’s not your knowledge of where the keys are that justifies your further conclusions. It’s your second-order beliefs about knowledge that justifies those beliefs. The practical analogue of this is to trade in the objective rule that says that you ought to maximize value for the more subjective rule that says that you ought to maximize expected value.4 Following the objective rule is a matter of first-order beliefs about the good causing the appropriate behavior. But if you take expected value as your goal, it looks as though following the subjective rule will be a matter of your secondorder beliefs about your beliefs about the good causing the appropriate behavior. If the subjective counterpart to an objective norm makes reference to beliefs about X where the objective version makes reference to X itself, then it looks like you’re simply trading in beliefs about X for beliefs about beliefs about X. If our beliefs about our beliefs are more reliable than our beliefs about X, this may seem like a step in the right direction. But if we’re not really infallible about our beliefs, you’re always going to be able to generate the same kind of puzzle case where the facts, in this case the first-order beliefs, go one way while the (second-order) beliefs go the other way. So the familiar and traditional retreat to the inner won’t really solve the problem. And in many cases, the connection between X and the relevant standards will be obvious, but the connection between beliefs about X and the relevant standards will be obscure. So the step that doesn’t really help will end up looking like a step in the wrong direction. Suppose that A has more expected value, but B has more value. What should you do? It’s just built into the story that doing B would be better. So why should you do something else instead? If expected value were a kind of value, maybe that kind of value could outweigh the value of B. But expected value is not a kind of value. There has to be 4
Jackson (1991).
6 something in your story to explain the significance or importance of expected value. And I take it as obvious that the notion of value simply cannot do this. If B is more valuable than A, then B is more valuable than A. And nothing you add to the antecedent of the tautology will keep the consequent from following. The natural reaction to objectivism is not that the alleged objective requirements can’t ever get a grip on us in some important sense. It’s that they can’t always get a grip on us. If we put the natural reaction in terms of the notion of a reason or in terms of The Principle, we’re committed to some particular interpretation of these things. And if we take the reaction as a serious objection to the objectivist, we’re committed to the idea that there’s some kind of “ought” that isn’t subject to the same difficulty. There has to be something that can always get a grip on us in the relevant sense. So it seems to matter a lot exactly how we put the reaction and exactly what the reaction commits us to. If the reaction commits us to something as implausible as the idea that we’re infallible about our beliefs, then as intuitive as it may seem, it’s really just based on a mistake, and we should do our best to get used to the idea that we’re required to be unreasonable on a daily basis. Here’s my plan. I think that the idea behind the natural reaction is that justification is a normative notion, and that there’s an important difference between the normative and the evaluative. There are two different ideas behind this distinction. On the one hand, genuinely normative reasons and requirements must be capable of guiding us in some important sense. And on the other hand, these reasons and requirements must be accessible to us in some important sense. The alleged requirements to jump out the window or to withhold when the keys aren’t there can’t guide us or get a grip on us because they’re inaccessible to us. So genuine requirements, or genuinely normative notions, are subject to a certain kind of epistemic constraint that facts about The Good are not. Here’s one way, but not the only way, of putting the epistemic constraint. It could be that φ-ing would be a very good thing even if you have no way of knowing that φ-ing would be a good thing. This is jumping out your window when you have no way of knowing about the fire. But you can’t be genuinely required to φ if you have no way of knowing that you ought to φ. Or to put it another way, if you ought to φ, then you’re in a position to know that you ought to φ. The condition of being required to do something is a luminous condition. Now maybe there’s nothing wrong with that. Maybe Williamson is just wrong when he argues that there are no interesting luminous conditions.5 I think that this way of putting the epistemic constraint on genuine requirements, or at least the requirements of rationality, is what leads to internalism in epistemology. And maybe there’s nothing wrong with that either. But I think that there’s another way of putting the epistemic constraint on genuine requirements that doesn’t have either of these consequences. This epistemic constraint is better motivated because you can derive it from the fundamental normative notion, the notion of doing things for reasons. And on the proper understanding of doing things for reasons, you can’t derive the luminosity constraint from that notion. So rather than arguing against luminosity or internalism, I will primarily be concerned with arguing in favor of my preferred interpretation of the natural reaction. On that interpretation, the alleged requirement to withhold judgment in the case of the 5
Williamson (2000): ch. 4
7 missing keys is not a genuine requirement because it doesn’t meet the epistemic constraint. So you don’t have the right kind of reason to withhold, and in the most important sense, you can’t. This interpretation of the reaction commits you to two very different things. The natural reaction, on pretty much any interpretation, commits you to the idea that genuine reasons, or reasons that generate genuine requirements are motivating reasons. They’re things that make things reasonable. While this may be the standard view among the epistemologists, the ethicists have their own reasons not to take motivating reasons seriously. So for the most part, this chapter is devoted to the defense of motivating reasons. All of this is only intended to move those who have what I’m calling the natural reaction, and not everyone does. So at the end of the chapter, I’ll argue that the other familiar conception of reasons and requirements, a conception that doesn’t include any serious epistemic constraints on normativity, simply doesn’t work in the theoretical case. It turns out to be surprisingly difficult to come up with a sensible picture of reasons for belief using only objective reasons. And these difficulties do not depend on any allegiance to the idea that we ought to be reasonable. But my interpretation of the natural reaction also commits you to a particular account of what the epistemic constraint comes to. So in the next chapter, I’ll argue in favor of my preferred constraint, and show that it doesn’t commit you to things like internalism and luminosity. Even if you have no problem with internalism and luminosity, it’s important to see that you can’t get them from the natural reaction alone. But more importantly, I think that only an externalist version of subjectivism can solve the main puzzle of the book. But before we get to that, let’s start with the fundamental normative notion: doing things for reasons. Two Kinds of Guidance There’s an avalanche coming down the mountain, about to hit a ridge. If it goes to the left, it will destroy the village. If it goes to the right, it won’t. It would be for the best if the avalanche went to the right. But we don’t think the avalanche has a reason to go to the right. It doesn’t even have an objective, normative reason to go to the right. And we don’t think that it has an excuse when it goes to the left, even though it had no way of knowing that going to the left would destroy the village. People, like other natural disasters, are a source of consequences, and these consequences can be evaluated in a way that is independent of their source. The fact that our minds are the way they are puts us in a position to do far more damage than any other animal. The badness of these consequences may depend on their effects on minds. But the fact that they originated in a mind is irrelevant. But people have things that avalanches don’t. We have reasons to do things, believe things, and care about things. Sometimes we have most reason to do, believe, or care. So we are often required to do, believe, or care. Evaluative notions apply straightforwardly to avalanches. But genuinely normative notions do not. It’s extremely hard to ignore the obvious suggestion that the fact that our minds are they way they are not only puts us in a position to do some damage. That’s what makes us subject to reasons and requirements. If reasons and requirements apply to some things but not others, there must be some explanation of this fact. If all we meant when we said that S has most objective, normative reason to φ were that S’s φ-ing would be for the best, then the avalanche would have an objective normative reason to go to the right. After all, that would be for the
8 best. But I think everyone would have the natural reaction to this. Even if we were to call the evaluative fact a reason, it just can’t get a grip on the avalanche. And here’s one way of thinking about what that means. Even if the avalanche has a reason to go right, and even if it’s possible for the avalanche to go right, it can’t go right for that reason. If this is why avalanches don’t have reasons, then it looks as though a necessary condition on having a reason is the ability to do things for that reason. And this looks like a general constraint on the notion of a reason. It’s not a reason to act if you can’t act on it, and it’s not a reason to believe if you can’t believe on the basis of it. I think that both people who like motivating reasons and people who like objective reasons ought to be willing to accept this constraint on some interpretation or another. But the two different conceptions of reasons lead quite naturally to two different conceptions of what it is to do something for a reason, and it’s important to see both the similarities and differences between these two conceptions. Start with motivating reasons. These are the kind of things that make things reasonable. But they’re also the kind of thing that gets us to do the things they rationalize. If you believe that q on the basis of your beliefs that p and that if p then q, the rationality of the conclusion is determined by the rationality of the beliefs it’s based on. And believing q for those reasons is simply a matter of the reasons themselves causing the belief. Unfortunately, not just any old kind of causation will do. The beliefs must cause the conclusion in the appropriate way, and it’s just a sad fact of life that no one has an account of which way is the appropriate way. But I think most people do have the general idea, even in the absence of an analysis. Suppose the beliefs that p and that if p then q lead, by a process of free association, to your imagining that not-q. You start daydreaming about not-q and things turn scary. The thought of q’s being false is so disturbing to you that this leads, by way of wishful thinking, to the belief that q. There is such a thing as bad reasoning. But cases like this don’t look like reasoning at all. The beliefs caused the conclusion they rationalize, but they didn’t cause it in the right way. And the basic idea is that this basis doesn’t make sense of believing that q from the agent’s point of view. Perhaps in some sense the beliefs themselves make sense of believing that q. But you don’t believe because they make sense of believing. Even though free association involves causation, and it even involves causation by content, it’s not the kind of causation by content that constitutes believing on the basis of reasons. Assuming that we have some handle on this notion, we have some idea of what it is to believe for a reason, at least when the kind of reason we’re talking about is a motivating reason. What about the other kind of reason? The fact that there’s a fire in the basement gives you an objective, normative reason to jump out the window. And the fact that you’ll need milk tomorrow can give you an objective, normative reason to go to the store today. What is it to do something for one of these reasons? The most natural suggestion here is that if you act on the basis of the fact that p, you must know that p.6 And your knowledge that p must cause you to do what the fact that p is a normative reason to do. Of course, not just any old kind of causation will do. So it’s not as though this team is any better off when it comes to relying on the notion of causally related in the appropriate way. The crucial difference is not in the kind of causation. The difference is that with motivating reasons, the reasons themselves do the causing, and with normative reasons, your knowledge of the reasons does the causing.
6
Hyman (1999).
9 But why do you have to know? Isn’t belief enough? It doesn’t look that way. Your epistemic enemies play a trick on you. They plant last month’s advertising section in your newspaper, thereby getting you to believe that there’s a sale on milk at the store. Just by chance, it turns out that there is a sale on milk. You go to the store, and there’s a sale. But do you go to the store because there’s a sale? I don’t think so. But if you want to be on the safe side, you can say that in order to act on the basis of the fact that p, you have to either know or believe that p. And this seems like a natural thing to say if we’re talking about reason-giving explanations. I ask why she went to the store, and you tell me that she went because she was out of milk. But here’s what happened. As a result of an elaborate plot on the part of a mad scientist, the lack of milk in her fridge led to a perfectly planned explosion, which launched her all the way to the store. This is a causal explanation of her going to the store in terms of her lack of milk. But it’s not a reasongiving explanation. And it’s not a reason-giving explanation because it doesn’t involve the kind of causation by content that makes sense of the action from the agent’s point of view. And in order to get the right kind of causation by content you need the right kind of states with that content. So there are two important differences and one important similarity between the two conceptions of doing things for reasons. Here are the differences. On the one hand, doing something for a motivating reason is a matter of rational causation by the reasons themselves. And doing something for a normative reason is a matter of rational causation by the knowledge of the reasons. On the other hand, motivating reasons are capable of moving you whenever you have them. And normative reasons are capable of moving you whenever you know about them. But both conceptions of doing things for reasons rely on the same notion of rational causation. If you know that jumping out the window will save your life, but you’re too terrified to think straight, you might have no inclination at all to jump the ten feet into the lake below. But this isn’t supposed to keep the fact that jumping will save you from being a normative reason to jump. On the objectivist’s conception of doing things for reasons, it’s not about what the future fact itself can cause. It’s about what it makes sense to do when you know the fact. One thing that distinguishes us from avalanches is the ability to do things for reasons. And both sides agree that doing things for reasons involves rational causation. This doesn’t mean that we always do what we have most reason to do, even when we know about the reasons. The mere fact that it doesn’t move you doesn’t by itself show that it’s not a reason. You might just be being unreasonable.7 And this is true on either conception of reasons. But both sides can agree that good reasons will always move a reasonable person. They just disagree on the proper interpretation of this idea. So consider the following sentence. (Reasons) (S)(φ) If S is rational and has most good reason to φ, S will be inclined to φ for that reason. I think this should be acceptable to both sides on some interpretation or another. We’re looking for a test to distinguish the normative from the evaluative. Anything can be good or bad, but only things that can happen for reasons can be required. The two conceptions of reasons lead quite naturally to two conceptions of doing things for reasons. Both conceptions involve the notion of rational causation, the kind of 7
Korsgaard (1986).
10 causation by content that makes sense of the thing from the agent’s point of view. Motivating reasons are the states that move you, and normative reasons are the contents of the beliefs that move you, at least when those beliefs are true or constitute knowledge. Let’s say that motivating reasons guide you directly, and normative reasons guide you indirectly. They guide you by way of your knowledge of them. Here’s the question to ask. Which conception of doing things for reasons provides a better test for distinguishing the normative from the evaluative? The two interpretations of (Reasons) lead to two different tests for genuine normativity. Assume that there’s something (positive) to be said for S’s φ-ing. To see whether that positive something is genuinely normative or merely evaluative, we ask if that positive something would always move a reasonable person, or whether a reasonable person would always act for that reason. The two conceptions of doing things for reasons lead to two different interpretations of the test, which I neutrally label (N) and (E). Consider the following open sentences. (N) (S)(φ) If S is rational, and S’s φ-ing is X, then the fact that S’s φ-ing is X will cause S to φ (or try to φ). (E) (S)(φ) If S is rational, and S knows that S’s φ-ing is X, then that knowledge will cause S to φ (or try to φ). The difference between the two conceptions of guidance shows up in the difference between (N) and (E), and this shows up at the end. In (N), it’s X itself that moves you, and in (E), it’s your knowledge of X that moves you. The difference between the two conceptions of when your reasons are capable of moving you shows up in the middle. Motivating reasons can move you whenever you have them, and normative reasons can move you whenever you know about them. The similarity between the two conceptions shows up in the first part, the part where we assume that S is rational. Whatever it is that moves you, it must do so by way of rational causation. Assuming that S is rational does not rely on any particular conception of doing things for reasons. Assuming that S is rational guarantees that she’ll be inclined to do what it makes sense to do. And both conceptions rely on that notion. According to the (N) test for normativity, if some substitution instance of (N) is true, then the thing that goes in for “X” is capable of directly guiding us. And it’s not just that it can guide us in some circumstances or another. It can guide us whenever it’s there. And that means it will guide us as long as we’re being reasonable. And this is the sense in which it can always get a grip on us, not that it always will. If we distinguish the normative from the evaluative in terms of direct, as opposed to indirect guidance, then the idea that normative things must be capable of guiding us is best understood as the idea that normative things must be capable of directly guiding us. So satisfying the open sentence is a necessary condition for genuine normativity. According to the (E) test for normativity, the thing that goes in for X is capable of indirectly guiding us, and it will guide us whenever we know about it, as long as we’re being reasonable. Which is the better test? If we’re looking for a test to distinguish the normative from the evaluative, we want facts about The Good to fail that test. Can facts about The Good guide us directly? In the case of the bad directions, the directions say to turn right, but turning left will be for the best, and of course the reasonable person would follow the directions. If the goodness of the option did get you to turn left in the case as described, it seems that it would do so against your will. And if it could always do this, it seems that it would be by
11 means of a magic power. I don’t know if anyone in the history of philosophy has ever believed this, but I think it’s safe to say that it’s not generally taken as a serious option these days. But suppose that The Good did have this magic power. You get to Elm Street, decide to turn right, but some mysterious and irresistible force gets either your hands or the steering wheel moving in the opposite direction. You may have a reason to turn left, and you may turn left, but you don’t turn left for that reason. And this is true regardless of which conception of doing things for reasons that you’re working with. We might say that it’s true on the neutral conception of doing things for reasons. All you need to deliver this result is what the two different conceptions have in common, the right kind of rational causation. So even if The Good did have magic powers, it’s clear that The Good cannot guide us directly. So facts about The Good fail the (N) test for normativity, just as they should. Do facts about The Good pass the (E) test? This is much more complicated. If you knew that going to the store today would be better than going tomorrow, this could lead by way of the right kind of rational causation to your going to the store today. But that doesn’t mean that it passes the (E) test. This is all very controversial, and what you say here depends primarily on whether you accept the Humean Theory of Motivation. According to that theory, no belief at all, regardless of its content, could ever rationalize, justify, or make sense of an action. In addition to the belief that φ-ing will be for the best, you also need the desire to do what’s best. If that desire is rationally optional, S could be perfectly reasonable, know that her φ-ing would be for the best, and not be the least bit inclined to φ. If that’s all true, then facts about The Good will not pass the (E) test. Unfortunately, it doesn’t look as though anything passes the (E) test given these assumptions. According to the Humean theory, no belief at all, regardless of its content, could ever rationalize, justify, or make sense of an action. So merely assuming that S is rational and knows that her φ-ing is X, will never guarantee that she’ll be inclined to φ, and this goes for any X. And even if something could pass the (E) test, I take it as obvious that the fact that turning left will get you to your friend’s house will not pass the (E) test, given our current set of assumptions. So the standard examples of normative reasons will turn out not to be normative after all. Well, maybe we should weaken the (E) test. Rather than saying that the knowledge that φ-ing is X must move a reasonable person, why not say that the knowledge could move a reasonable person, perhaps in conjunction with rationally permissible desires? The problem, of course, is that facts about The Good pass the revised test. If you know that going today will be better, this could get you to go. But if facts about The Good pass the test, that means that the test is no good. Any account of the distinction between the normative and the evaluative should have The Good land squarely on the evaluative side. So there doesn’t seem to be any way at all of distinguishing the normative from the evaluative using the notion of indirect guidance if the Humean theory is true. But suppose for a moment that the Humean theory is false. Some beliefs, all by themselves, can make sense of doing things. It certainly wouldn’t follow that just any old belief could make sense of doing things. But in the practical case, beliefs about The Good are among the most plausible candidates. If you think your φ-ing would be for the best, then you see your φ-ing in a positive light. If the Humean theory is false, it seems that this would make sense of φ-ing. So if the Humean theory is false, it seems that facts
12 about The Good will pass the (E) test, and that means that the test is no good. So if this is the stand you take on the controversial issues, it seems that you must reject the (E) test for normativity. But this doesn’t carry over to the theoretical realm. Suppose I offer you enough money to form the false belief that you’re riding a pink elephant. If you need the money, you could quite sensibly believe that it would be for the best if you believed that irrational thing. Wanting to be irrational and thinking it would be for the best if you were irrational are quite different from thinking that you ought to be irrational. But I assume that if you’re rational, you will not believe that irrational thing. Since you have no evidence for the belief, you won’t believe on the basis of that. And since you’re being reasonable, the black arts of the mind will not be available to you. So maybe, on reflection, The Good fails the (E) test after all, at least in the theoretical case. So if this is your stand on the controversial issues, maybe you can accept the (E) test after all. Unfortunately, once again, it’s not at all clear what does pass the (E) test, given this set of assumptions. Let φ-ing be turning left on Elm Street. Suppose that S is rational and knows that her φ-ing will get her to her friends’ house. Does it follow from this alone that S will be inclined to φ? Presumably not. S may have no reason to go to her friends’ house. You need to add more to the story. But what do you need to add? If what you add amounts to the fact that turning left would be for the best, the standard examples will not pass the test, because we’re assuming that facts about The Good fail the test. If facts about The Good did pass the test, that would mean that the test was no good. But maybe what you add to the story amounts to the fact that turning left would be the most reasonable thing to do. We’re tempted to add the fact that S wants to go to her friends’ house. We don’t expect irrational desires to move reasonable people. But if we assume that S is rational and that she wants to visit her friends, we can safely conclude that this is a reasonable desire. And if S wants to visit her friends but quite sensibly wants something else more, we don’t expect the outweighed desire to move the reasonable person. And we’re assuming at the moment that the Humean theory is false. So if what S wants most is to visit her friends, but she knows that she shouldn’t, we can’t automatically assume that the desire will outweigh the belief. So the thing that goes in for X is not merely that it will get her to her friends’ house. And it’s not merely that it will satisfy some desire or another. It seems that what goes in for X is either the fact that φ-ing is the most reasonable thing to do or the facts in virtue of which φ-ing is the most reasonable thing to do. So maybe, given our current assumptions, rationality passes the (E) test. But it’s really not obvious that it does. Suppose that S is rational and knows that her φ-ing is the most reasonable thing to do. I think it follows that S will be inclined to φ. But the question is not only about what she’ll do. It’s also about the reasons for which she’ll do it. If the things that make it reasonable for her to φ also make her φ, then the reasons for which she φ’s are the ordinary, first-order motivating reasons that make it reasonable. Rationality only passes the (E) test if what really moves her is her knowledge of these facts, and not the facts themselves. The fact that φ-ing is the most reasonable thing to do plus the fact that S is rational seem to entail that S will be inclined to φ. So it seems that her knowledge of these facts is redundant. So it’s really not obvious that she’s moved by her knowledge of the things that make things reasonable rather than by those things themselves.
13 In the practical case, the question comes down to something like this. Suppose that when you know that turning left will get you to your friends’ house, part of what makes it reasonable for you to turn left is your desire to visit your friends. Does the desire itself get you to turn left, or is it really your knowledge that you desire it, or your knowledge that the desire makes it reasonable? In the theoretical case, it comes to this. Suppose that your justified beliefs that p and that if p then q are what make it reasonable for you to believe that q. Do those beliefs themselves get you to believe that q, or is it really your knowledge that you believe them, or your knowledge that they’re justified, or your knowledge that they make it reasonable to believe that q? I’m a big fan of the primacy of the first-order, so my answer is clear. It’s the justifiers themselves that move you and not your knowledge of the justifiers. The problem with the move to the second-order is that an obvious regress looms. You believe that p and that if p then q. There’s an obvious rational connection between these and believing that q. Suppose we say that the mere existence of the rational connection is not enough, that you also have to know about the connection, and that it’s your knowledge of the connection that really moves you. Now we need to ask why the second-order belief can move you when the first-order beliefs can’t. You believe that believing that q is the most reasonable thing to do. There’s an obvious rational connection between this and believing that q. But if we’re saying that the existence of the connection is not enough and that you also have to know about the connection, then we need a third-order belief connecting the second-order belief to the conclusion. And it’s not merely that the third-order belief will be no better off than it’s lower-order cousins. It’s that the more complicated the beliefs get, the less likely they are to move you. So the third-order belief is probably worse off than its cousins. So given this particular set of assumptions about the controversial issues, it looks as though once again, nothing passes the (E) test, not even the facts of rationality. But if rationality is the only thing that passes the (E) test, then as we’ll see, the (N) test and the (E) test come to the same thing. The natural reaction commits us to the idea that there’s an important distinction between evaluative notions, presumably including The Good, and genuinely normative notions, presumably including some kind of reasons and some kind of requirements. It’s uncontroversial that there is such a distinction. But it’s somewhat controversial what the distinction comes to. If you think the distinction between the normative and the evaluative is explained in terms of the idea that genuinely normative things must be capable of guiding us, it looks as though the notion of indirect guidance simply won’t do. If you state your test in terms of indirect guidance, and you make your test so difficult that The Good can’t pass it, there’s no reason to think that anything will pass it. And it’s fairly clear that the standard examples of objective, normative reasons won’t pass it. If you make your test easy enough for The Good to pass, that means that the test is no good. If there is a distinction between the normative and the evaluative, The Good will fall squarely on the evaluative side. Getting a Grip Facts about The Good fail the (N) test, just as they should. But does anything pass the (N) test? Let’s turn from the obviously evaluative notion of what has most value to the apparently normative notion of what you have most reason to do. Reasons, it seems, must be capable of moving reasonable people. They must be capable of guiding our behavior. In some way or another, they must be capable of getting a grip on us. And this seems to be, in one way or another, what the distinction between the normative and
14 the evaluative comes to. If we think of these things in terms of direct, as opposed to indirect guidance, it seems that genuine reasons must satisfy (N). So consider the following sentence. (NR) (S)(φ) If S is rational and S has most (good) reason to φ, then the fact that S has most reason to φ will cause S to φ (or try to φ). If (NR) is not true on some conception of reasons, then according to the (N) test, those reasons, no matter what we decide to call them, are not genuinely normative. Since there are, or might be, two kinds of reasons, there are two ways of interpreting this sentence. On one interpretation, it’s about objective, so-called normative reasons, and on the other, it’s about subjective, motivating reasons. Let’s start with the allegedly normative reasons. (NNR) (S)(φ) If S is rational and S has most normative reason to φ, then the fact that S has most normative reason to φ will cause S to φ (or try to φ) I take it that (NNR) is obviously false. In the case of the bad directions, you have most normative reason to turn left, but reasonable people would turn right. We don’t need to assume that allegedly normative reasons are simply facts about The Good in order to see that they fail the (N) test. Rather, it’s the fact that both normative reasons and evaluative facts fail the test that suggests that S’s having most normative reason to φ simply amounts to S’s φ-ing being for the best. So what about motivating reasons? (NMR) (S)(φ) If S is rational and S has most motivating reason to φ, then the fact that S has most motivating reason to φ will cause S to φ (or try to φ) The claim that S has most motivating reason to φ means that S has most good motivating reason to φ. And motivating reasons aren’t just reasons that get you to φ. They’re also the sorts of reasons that determine the rationality of φ-ing. So I think it’s safe to assume that that if S has most good motivating reasons to φ, then φ-ing is the reasonable thing to do, and if S is rational, that’s what she’ll be inclined to do. So I take it that (NMR) is pretty obviously true, and we’ve finally found something that can guide us directly. The notion of direct guidance is not understood in terms of the notion of proximal causation or in terms of the notion of a causally sufficient condition. The idea is that we don’t need beliefs about these reasons in order for them to guide us. They can rationally cause our φ-ing without our having to be aware of them. Finding something that passes the (N) test seems to involve figuring out what people are rationally required to respond to or what we can always reasonably hold someone responsible for. According to (NMR), if you have most motivating reason to φ, but you’re not inclined to φ, then you’re being irrational. Given that we’re understanding motivating reasons in terms of things that make things reasonable, this may seem somehow trivial or definitional or something like that. So you might think that (NMR), all by itself, is no big deal. But if you think that satisfying the open sentence (N) is a necessary condition for genuine normativity, it shouldn’t matter how exciting the relevant substitution instance is. All that should matter is that it’s true. So we have a proposed test for genuine normativity. In some sense, good motivating reasons trivially pass this test. But this by itself is not a problem either for the reasons or the test. The question is whether the test is a good one, and we answer that by
15 looking at the open sentence, not the substitution instance. So compare (NMR) with (FNR). (FNR) (S)(φ) If S is fantastic and S has most normative reason to φ, then S will be inclined to φ. Let’s just define “fantastic” in such a way that being fantastic simply amounts to doing what you have most objective normative reason to do. If you don’t turn right in the case of the bad directions, it follows that you’re not being reasonable. And if you don’t turn left, it follows that you’re not being fantastic. So far, (NMR) and (FNR) are on a par. One thing we can say about motivating reasons is that they satisfy (N). And one thing we can say about normative reasons is that they satisfy (F). (F) (S)(φ) If S is fantastic and S’s φ-ing is X, then S will be inclined to φ. The difference is in the open sentences, not the substitution instances. And here’s one difference. If you’re not fantastic in the case of the bad directions, we just won’t hold it against you. And when I say we won’t hold it against you, I don’t mean that even though we think you deserve it, we’ll refrain from yelling at you because we don’t think the yelling will have good consequences. I mean that in the consequence-free zone of our heart of hearts we won’t hold it against you. And all that means is that we don’t really believe that you’re genuinely required to be fantastic. It would be nice if you were fantastic. We would prefer you to be fantastic. It would be fantastic if you were fantastic. And it might even be that you would be required to be fantastic if only things were completely different, e.g., if you were omniscient. But saying all of these evaluative things about being fantastic simply doesn’t approximate the idea that being fantastic is genuinely required of you. It would be nice and preferable if it stopped raining so that we could go to the park. And if things were completely different, and the weather were an agent and our friend or owed us some money, then maybe it would be required to do us a favor. But none of this comes anywhere near the idea that as things stand, the weather is genuinely required to get better. If you say the same thing about the weather and an action, you’re not saying anything normative about either. But why isn’t the weather required to get better, and why aren’t we required to be fantastic? The idea seems to be something like this. Even if it would be for the best if the weather turned nice, and even if we insist on calling that evaluative fact a reason, as things stand, the weather cannot get better for that reason. If the weather did get better, it would just be an accident that it did what it ought to do. It did what it should, but it didn’t do it because it should. Even if it would be for the best if you turned left, and even if we call that a reason, as things stand, you simply cannot turn left for that reason. You could turn left for some other reason, perhaps if your paranoid delusion gets you thinking that your friends are trying to trick you. But even though you do the right thing, that’s just an accident, and you just got lucky. You didn’t do the right thing for the right reason. We assume in (N) that S is rational. But we don’t assume this simply in order to make the subjectivists win. This came from the neutral conception of doing things for reasons, what the two different conceptions have in common, the relevant notion of rational causation. If your knowledge that there’s a fire in the basement of your hotel somehow gets you to try to bash your head against the wall, but you miss the wall and end up throwing yourself out the open window by mistake, you did what you had most normative reason to do. But you didn’t do it for that reason, even on the objectivist’s
16 conception of doing things for reasons. All you need to prefer (N) to (F) as a test for normativity is the neutral conception of doing things for reasons, and the idea that avalanches don’t have reasons because they can’t do things for them in any sense at all. If you don’t turn right in the case of the bad directions, then you’re not being reasonable. Many of us will hold this against you in the consequence-free zone of our heart of hearts. We think that unlike being fantastic, you are genuinely required to be reasonable, even if it makes you late for dinner. We don’t have to think that you ought to be reasonable because keeping your mental house in order is intrinsically valuable. We might just think that you ought to be able to do the right things for the right reasons. And any conception of that will require rational causation. If you do sensibly turn right, you not only do what the subjectivist says that you should. You also do it for the right reasons. You don’t just luck into the reasonable option. If you’re overwrought or in a panic, maybe now we’ll say that you have an excuse for not doing what you should have done. But unlike the objectivist, we’re not requiring you to do something in circumstances in which your compliance with the norm could at best be just an accident. Since one and the same thing both generates the requirement and gets you to comply with it, we can always hold you responsible, not only for doing the reasonable thing, but for doing it for the right reasons. Suppose someone says that you ought to believe when the keys are there and withhold when they’re not, even though withholding judgment in those circumstances would be irrational. The natural reaction to this is not merely to say that you’re required to be reasonable. The natural reaction is to say that the alleged requirement to withhold isn’t a genuine requirement because the alleged requirement can’t get a grip on us. Since obviously evaluative things can get a grip on us in the sense of indirectly guiding us, the relevant notion of getting a grip must be understood in terms of direct guidance. So the objective “ought” may well be covert evaluative talk about what’s for the best, but it doesn’t express the idea of a genuine requirement. But if we take this reaction seriously, we are committed to the idea that there are some requirements that can always get a grip on us in the relevant sense. And it turns out that the requirements of rationality can always get a grip on us in the relevant sense. Presumably, this is not the only way of drawing the normative/evaluative distinction. Though I must admit that I don’t know of any other principled way of drawing it. But in any case, it seems like an option we should seriously consider. The idea is something like this. Only good reasons generate genuine requirements, and a necessary condition on being a good reason to φ is that there must be a rational route from the reasons to the φ-ing. So in order for it to be a genuine requirement, there must be a rational route from the reasons themselves, or from the things that generate the requirement, to the thing required. That’s what it takes to do the right thing for the right reasons. It’s not enough that we can sometimes respond to the alleged reasons. And it’s not enough that the alleged reasons would be able to indirectly guide us if only we were in a position to know about them. Genuine reasons, or reasons that generate genuine requirements to φ, are the kinds of reasons for which we can always hold people responsible, and they’re the kinds of reasons that can themselves rationally cause us to φ. Since this is supposed to be a picture according to which you’re required to be reasonable, it’s not the least bit surprising that on this picture, genuine reasons are the kinds of things that make things reasonable. Genuine reasons are motivating reasons. According to the other obvious option, these standards for genuine reasons are far too strict. According to the other option, it’s enough if the reasons, or the things that
17 generate the requirements, can indirectly guide us in some circumstances or another. So there doesn’t have to be a rational route from the reasons themselves to the φ-ing. It doesn’t even have to make sense for there to be a rational route from the reasons themselves to the φ-ing. It’s enough if in some other circumstances there’s a rational route from knowledge of the reasons to the φ-ing. Unlike the previous option with the stricter standards, this option does not have the consequence that normative reasons are normative in name only. I think it’s just obvious that the claim that S has most normative reason to φ just amounts to the claim that S’s φ-ing would be for the best. I think it’s just obvious that talk about what’s for the best is evaluative talk. And when you talk this way, it makes no difference whatsoever whether S is an agent, the weather, or an avalanche, and it makes no difference whether S’s φ-ing is an action or a natural disaster. So deep down, I think we should take it as a criterion of adequacy on an account of the distinction between the normative and the evaluative that it have the consequence that normative reasons are normative in name only. But it’s clear that the ethicists are not going to go along with any of this, at least, not those ethicists who think you’re required to be unreasonable on a daily basis. So the natural reaction is committed to something controversial. This might not be that controversial among the epistemologists. But it is controversial among the philosophers. But at least so far, it doesn’t look as though the natural reaction is committed to anything embarrassing. The idea behind the natural reaction is not just that you’re genuinely required to be reasonable. It’s that alleged requirements that tell you to be unreasonable aren’t genuine requirements because they can’t get the right kind of grip on us. Even if we did manage to be unreasonable just when they say that we should, we couldn’t possibly do that for the right reasons. Since it’s just part of the standard story about what normative reasons are that they don’t get that kind of a grip on us, we shouldn’t be worried about saying that normative reasons are normative in name only. I think we should be worried if we don’t say it. There’s a fire in the basement of your hotel, but you have no evidence of a fire and lots of evidence that there’s no fire. Of course it would be for the best if you were an idiot and jumped out the window for no reason whatsoever. In fact, that would be fantastic. But whichever conception of reasons you start with, you’re committed to the significance of rational causation, at least to the extent that you think it matters whether people can do the right things for the right reasons. So you shouldn’t be embarrassed to say that there’s a sense of “reason” that takes Reason seriously, and in that sense, you have no reason whatsoever to jump out the window; what you ought to do is determined by what you have most of this kind of reason to do; so of course you can’t be required to be unreasonable. And it’s not just that you shouldn’t be embarrassed. You should get up on your high horse and look down on those who disagree with you. “Be reasonable” is a categorical imperative that applies to all agents regardless of their contingent desires. I think that explaining the normative force of rationality is a little like explaining the wetness of water. The table is wet because it has water on it, and the streets are wet because they have water on them. But if someone asks why water is wet, you need a different kind of explanation, ultimately an explanation in terms of what it is to be wet. If someone asks about the normative force of a certain set of rules, they’re basically asking for an explanation that makes sense of following the rules. Sometimes we can give such an explanation, and sometimes we can’t. And when we can’t give such an explanation, sometimes it’s our fault. We just haven’t thought of it yet. But sometimes, it’s the rules’
18 fault. The rules really don’t make sense. But if someone asks you for an explanation of why making sense makes sense, you need a different kind of explanation. Doing things for reasons is what normativity is all about, and it’s one of the things that separates us from avalanches. So whenever anyone does anything for a reason, a certain set of standards are automatically in force, the standards that determine whether the reasons really do make sense of what they’re supposed to make sense of. So far, I’m only preaching to the choir. I don’t think that everyone will have what I’m calling the natural reaction to the idea that you ought to believe when the keys are there and irrationally withhold when they’re not. I just think that this will be the natural reaction among the epistemologists. I think this is a perfectly natural reaction to have, and I think that everyone should have it. But at the moment, I’m only asking if you should be worried about having it if you do. And I don’t think that the standard arguments people give for preferring normative reasons to motivating reasons give us anything to worry about. This is already too long, so I’ll skip my evaluation of the sorts of arguments the ethicists usually give for preferring objective reasons to subjective reasons. When you’re thinking about reasons, you’re thinking about the world. When we give advice, it makes sense to say… And so on. Trust me, those arguments don’t work. Objective Reasons to Believe So much for preaching to the choir. If you have what I’m calling the natural reaction to the idea that you ought to believe when the keys are there and withhold when they’re not, then at least so far, you have no reason to be embarrassed. The natural reaction is committed to the idea that genuine reasons are the kinds of reasons that make things reasonable, the kind that can move you directly, and many epistemologists will think that this is the only sensible way of talking about reasons. But the ethicists don’t think that, and there may be some epistemologists who don’t think that either. And it’s not as though there’s nothing to be said for the other side. If you believe in the case of the missing keys, then you’re wrong, and you’ve made a mistake. So it looks as though you’ve violated some norms, or you’ve done something you shouldn’t have done. And it seems that the kinds of things that make it the case that you should have done otherwise are going to be objective, normative reasons. And there’s the practical analogue of our puzzle, which might seem to lend some support to the argument from advice. You get to Elm Street, and you’re trying to figure out what you ought to do. Lucius tells you that the directions are wrong and that turning right will not get you to your friends’ house. You don’t reply by saying that this information is irrelevant because you were only trying to figure out the most reasonable course of action given your current state of information. It must matter to you whether your actions will have their intended consequences. If the consequences didn’t matter to you, you wouldn’t intend them. So if the only possible form of subjectivism is the one on which the external world is always and everywhere irrelevant, maybe objectivism is the best we can do. On the objectivists’ way of thinking about reasons, the choir’s standards for what counts as a genuine reason are far too strict. Of course, reasons must be capable of moving you, or guiding you, or getting a grip on you. It’s just that on the alternative understanding of reasons, indirect guidance is enough. So even in a situation in which you have no way of knowing that p, the fact that p can still be a normative reason to φ if
19 the knowledge or belief that p could rationally cause you to φ. The fire in the basement itself doesn’t have to be capable of rationally causing you to jump out the window in your current circumstances. It’s enough that if you were in other circumstances where you did know about the fire that knowledge could rationally cause you to jump out the window. It seems easy to dismiss these normative reasons as merely potential reasons, or possible reasons, or reasons you would have if only you were in that other situation where you know about the fire. And it seems easy to take the “ought” of advice as making evaluative claims about what would be a good thing to do. But until the subjectivists can give an account of why these things automatically matter, if they don’t matter by generating genuine requirements, maybe we shouldn’t be too embarrassed. But we shouldn’t get too comfortable either. So the objectivists will have their own interpretation of the idea that reasons must be capable of moving, gripping, and guiding us, and they’ll have their own interpretation of what it is to do something for a reason. If we think of the underlying idea in terms of indirect guidance, it’s natural to accept the following pretty weak necessary condition on what can be an objective normative reason to φ. (PW) The fact that p is a normative reason to φ only if the knowledge or belief that p could rationally cause someone to φ in some logically possible situation or another. This is supposed to come from the objectivist’s own picture of what it is to do something for a reason. If you act on the basis of the fact that p, you must know that p, or at the very least, you must believe that p. And unlike (E), the idea here is only that someone could act for that reason, not that any rational person would. So (PW) doesn’t rule out the fact that it’s chocolate from being a reason to eat it. Knowledge of that fact could reasonably lead someone (e.g., someone who wanted some chocolate) to eat it. We don’t think of leading to or rationally causing as leading to on its own. And (PW) doesn’t rule it in as a normative reason either. It’s just a necessary condition. (PW) is really pretty weak. And here we have a fundamental disanalogy between The Good and The True when it comes to normative reasons. As far as (PW) is concerned, the fact that φ-ing would be for the best may well be a normative reason to φ because knowledge of the evaluative fact could rationally cause a good person to φ. And that’s what it would be, on this picture of reasons, to φ for that reason. And again, (PW) doesn’t entail that it is a normative reason. It’s just a necessary condition. But the fact that p can’t be a normative reason to believe that p, because the knowledge that p can’t rationally cause the belief that p. It is the belief that p. And the main problem is not that it’s too close to cause it. It’s that it’s too close to rationalize it. So you can’t believe for that reason, even given this much more inclusive notion of doing something for a reason. Similarly, the fact that not-p can’t be a normative reason to believe that not-p. And the fact that not-p can’t be a normative reason to withhold judgment either. Your knowledge that not-p might be so disturbing that through some process of repression it gets you to give up the belief that not-p. But the knowledge that p is false can’t rationally get you to fail to believe both p and not-p. So it’s not a normative reason to withhold. So if you’re in a situation where you only have three options with respect to the proposition that p, where you have to believe, withhold, or deny, the fact about whether or not p is true cannot provide any normative reason to go one way rather than another.
20 Saying that you ought to withhold judgment because the keys aren’t there is like saying that you ought to believe you’re riding a pink elephant because I’m wiling to pay. The fact that I’m willing to pay means that there’s something good about believing. Call those things reasons if you must. But there’s no way to believe, withhold, or deny for those reasons because there’s no rational route from the knowledge of the normative reason to what they’re supposed to be reasons for. Neither the fact that believing that p will make you happy nor knowledge of that fact can rationally cause the belief that p. The fact that p is true is like this in one respect. In the objectivist’s sense of believing for a reason, you can’t believe that p on the basis of the fact that p. So the fact that p, like the fact that if you did believe p you’d thereby know that p, is not a normative reason to believe. But sometimes, like in an ordinary case of perceptual knowledge, we might say that you believe that p because p is true. This means that the fact that p itself is a partial cause of your belief. And that makes it more like a motivating reason than a normative reason. Motivating reasons are rational causes. Normative reasons are the contents of rational causes, at least when the relevant beliefs are true. And in an ordinary case when you go to the store because you’re out of milk, the fact about the milk is a normative reason, and your knowledge of that fact is a motivating reason. And if you know that p and that if p then q, these are excellent motivating reasons to conclude that q. Motivating reasons aren’t identified by their internality. They’re identified by their role in reasoning. So if you’re looking for a sense in which the fact that p or your knowledge that p could be reasons to believe, withhold, or deny, the only option is the notion of a reason according to which reasons are things that make things reasonable. On this picture, you believe the right thing for the right reason when the things that make you justified also make you believe. I have no problem with conclusive reasons. So sometimes, the things that put you in a position to know, e.g., that you can see that p, also make you believe. The problem with the alleged objective requirements isn’t that they can’t ever get a grip on you. It’s that they can’t always get a grip on you. The problem is that some things that keep you from knowing, like the fact that the keys are gone, can’t rationally cause withholding in those circumstances. But things that keep you from being justified can always rationally cause withholding in those circumstances. So on either notion of a reason, the fact that the keys aren’t where you left them is not a reason to withhold judgment. So it seems safe to give into temptation and say that in the case of the missing keys, you have no reason of any sort to withhold judgment. And since what you ought to do is determined by what you have most reason to do, withholding judgment can’t really be required.