Many philosophers today are pursuing a program according to
which the notion of a “normative reason” is the most fundamental normative notion. Thus,
these philosophers aim to analyse such notions as ‘right’ and ‘wrong’, the
notions of the various sorts of “value”, and so on, in terms of the kinds of
attitudes that we have “normative reason” to have, or the kinds of actions that we
have “normative reason” to perform.
In my view, this program suffers from a serious flaw: it
fails to recognize quite how profoundly context-sensitive the term ‘reason’ is. In consequence, this program is doomed to
waste a lot of time debating various pseudo-problems, which arise from failing
to recognize that in different contexts, the term ‘reason’ expresses different concepts.
Of course, philosophers have long known that the term ‘reason’
is used in many ways: for example, there are pure explanatory reasons (‘The reason for the bridge’s collapse was that its girders were suffering from metal fatigue’) and motivating reasons
(‘Martin’s reason for flying to Ireland was to go to his aunt’s funeral’), as
well as normative reasons (‘There are
reasons to be sceptical of this philosophical program’).
In fact, I am inclined to think that all of these uses of the
term ‘reason’ have some link with explanation.
Even the purely normative reasons are in a way explanatory: a normative reason
for a certain action or attitude is something that goes some way towards
explaining why that action or attitude has some positive normative status (like being an action that it is rational for one to perform, or an
attitude that it is appropriate to
have, or the like).
Still, even if we restrict our attention to statements about
“normative reasons”, it seems to me that the term ‘reason’ expresses
different concepts in different contexts. As I said, all normative reasons for
an action or attitude go some way towards explaining why that action or
attitude has some positive normative status; but there are many different sorts
of “positive normative status”.
For instance, suppose that you have a reasonable belief in proposition
that is in fact false. E.g., suppose that you reasonably believe that the man
approaching you is an enemy soldier who is determined to kill you, and that the
only way to defend yourself is to shoot him. In fact, however, the man is entirely
innocent and poses no threat of any kind. Do you “have a reason” to shoot him?
Well, if you shoot him, your act is perfectly rational, and indeed excusable; your
reasonable belief counts as an excuse in the eyes of the criminal law. But in
another sense, your act is not right; anyone
else who knew that the man approaching you posed no threat would be under an obligation to try to stop you from shooting
him if they could. So perhaps you don’t really “have
a reason” to shoot him?
The best way to handle such cases, in my view, is to say that your act of
shooting the man has one positive normative status (it is rational and
excusable), but lacks another (it is not objectively right). So in one sense, you
have “normative reason” to shoot (there is something that explains why your act
is rational and excusable); but in another equally legitimate sense, you do not
have “normative reason” for shooting (there is nothing that goes any way
towards explaining why your act was objectively right). It would be an
unprofitable pseudo-problem to worry about whether you really have reason to shoot him.
Yet another pseudo-problem that philosophers will get easily sucked
into is to worry about what your reason
is in these cases. We might think that in this case your reason for shooting
the man is the fact that it is reasonable for you to believe that the man
is attacking you. But of course, if we asked you to say what your reason
is, you would say that your reason is the fact that the man is attacking you.
Moreover, if the man really were attacking you, it would seem perfectly
true to say that your “normative reason” for shooting him consisted of
the fact that he was attacking you.
But surely it would be a sort of “double counting” to say that both the fact that the man was attacking you and
the fact that it was reasonable for you to believe that he was
attacking you are reasons for shooting him? In this way, it can start
to seem hard to say what your reason is.
This problem is easily dissolved if we say that “normative reasons”
of the sort that go towards explaining why an act is rational or excusable are
always facts about what beliefs or other attitudes it is reasonable for the agent to have, while “normative
reasons” of the sort that go towards explaining why an act is objectively right very often include
other sorts of facts about the external world.
It is worth noting that there is compelling linguistic
evidence that the term ‘ought’ is context-sensitive in exactly the same way. Suppose
that you are on top of a tower tracking someone who is making his way through a
maze on the ground. You might say, ‘He has no way of knowing it, but he ought
to turn to Left at this point’. But you might also say, ‘Since all the evidence that
he has had so far supports going Right, he ought to turn Right
(and not Left) at this point.’ Both your statements seem perfectly true, when
taken in their intended sense. But it surely can’t be true in this case that he
ought both to turn Left and not to turn Left. So, it seems, we must distinguish
between “objective” and “subjective” senses of ‘ought’.
As I have argued, the same point is true of ‘reasons’ as
well. So philosophers who take all their linguistic intuitions about the term ‘reasons’
to concern one and the same concept will be led into unprofitable pseudo-problems.
This is in my view a serious flaw with the “reasons” program.
Wow, I agree with a whole lot of that, but I don’t think I agree with the main thrust. I doubt there are so many different concepts of ‘reason’. There could be one, context-sensitive concept; that seems much more likely. (Compare: there aren’t lots of different concepts of ‘wealthy’, but only one, context sensitive concept.)
Here’s a suggestion: all of these interanalysable concepts are relative, always, to some probability distribution. (This is a suggestion for how they are context sensitive.) Sometimes it’s the agent’s credence function; sometimes it’s the evaluator’s; sometimes there is, or anyway our talk is committed to there being, an objective probability function. If that’s how the context sensitivity works, then maybe it’s no obstacle to the ‘reasons program’.
Jamie — I was using ‘concept’ so that the nature of a concept is sufficient to determine its reference / semantic value. So if two occurrences of a term have different semantic values, they express different concepts. Still, the two occurrences of the term may have the same meaning — especially if we think of the meaning as a Kaplan-style character, or function from contexts to contents. So, the word ‘wealthy’ has a single meaning but expresses many concepts – wealthy by the standards of group A, wealthy by the standards of group B, etc.
Your suggestion that statements involving ‘reason’ are implicitly relativized to probability distributions is very interesting, but I think there’s a snag. Surely, the agent’s actual credence function won’t determine any concept of a ‘reason’ if this credence function is grossly irrational? Perhaps you think that this is ruled out by your assumption that the agent’s credence function can always be represented by a probability distribution. But surely it’s possible for an irrational agent to have a probabilistically incoherent set of beliefs? If that’s right, then it is plausible to claim that every agent “has a reason” to have a credence function that can be represented by a probability distribution. But your account seems to trivialize that claim in a worrying way.
By the way, what about the supremely accurate “probability distribution” that assigns probability 1 to every true proposition in the algebra, and probability 0 to every false one? Then we’d have a maximally objective notion of ‘reason’. I wonder if you’d be happy with that? (I wouldn’t object myself.)
1. OK, many concepts, then, in the sense that there have been as many concepts expressed by the word “I” as there have been speakers of English.
2. Hmmm, I don’t see the problem. Suppose someone’s credences are not represented by any probability function. Then when we’re looking around for a probability function to plug into the empty slot in the reason concept, we’ll have to look elsewhere – maybe nearby, by finding a probability function that approximates her credences? Maybe we pick a subset of her credences and find a probability function that represents them? I might use my own credences, assuming they are coherent! Could be any of those in different contexts, I believe. But why is there supposed to be a problem here?
3. Yes, the extremist probability distribution that assigns 1 to each truth is a perfectly good one for reason-determining purposes, I agree.
By the way, PEA Soup’s little final verification step before your comment is posted test has convinced me that I am, in fact, a robot.
There’s too many things to say about this but here’s a beginning.
Say that you were brain-washed to believe that the people on the street are enemy soldiers out there to shoot you. You might be brainwashed to believe in the appropriate evidence too so that the belief is reasonable given what other things you believe. Given these beliefs you judge that you have a reason to shoot the pedestrians. If you don’t, you’ll die. And, so you start shooting them (I would anyway). That action seems rational at this point. It coheres with what you think you have reason to do. The act also seems excusable. Given what you believe and given that you didn’t have a choice to believe otherwise, this is what you would do if things went well cognitively speaking. You shouldn’t be blamed.
Now, there must be a perfect explanation for why your action is both rational and excusable. It’s essentially something to do with brainwashing. Is that a normative reason in any sense? No.
In that case, you don’t have any reason to shoot people assuming that there is no way that there could be enemy soldiers in the environment.
Things stand differently if you are in a dangerous environment. There you really might have good reasons to shoot people who are not enemy soldiers but look like ones. Chances are that you die otherwise and survival seems to be a good reason-provider. In that case you don’t have to refer to beliefs about enemies as reasons when you can talk about chances of getting killed.
So, I’m not sure what they pseudo-problem is. Also, I don’t see how this gives you the conclusion. Even if normative reasons were ambigious and reason ascriptions context sensitive, I don’t see how it follows that some sense of reasons cannot help to account for values, right, and wrong. This is true especially in the case that the talk about value, right, and wrong seem to be ambiguous and context-sensitive in the very same way.
Ralph – I’m puzzled by your post, because I don’t see why any of this constitutes any kind of argument that reasons are not the basic normative notion. It just looks like an argument that philosophers have to be careful to distinguish which sense of ‘reason’ they mean to be saying is the basic normative notion.
And some of us, you know, do take care to distinguish these different senses of ‘reason’ from one another, and to ask explicitly how they are related. For example, I discuss the main two senses of ‘reason’ that your post focuses on – the objective and subjective senses – in ‘Having Reasons’, and I discuss how they are related to one another in ‘The Scope of Instrumental Reason’ and in ‘Means-End Coherence, Stringency, and Subjective Reasons’, among other places. My paper, ‘Reasons and Agent-Neutrality’, is about another important distinction between different relations that might be associated with the word ‘reason’, and focuses specifically on how they are related to one another, and which one we should take to be basic.
The fact that there are lots of senses of ‘reason’ that people sometimes (or often) get confused about doesn’t suggest to me that we shouldn’t do any philosophy about any of them – it suggests to me that there is a lot of philosophy to be done about them.
Confusions between the various ways in which the word ‘reason’ can be used don’t arise due to any philosophical theory that in some sense of ‘reason’, reasons are the basic normative notion. These confusions arise whenever we talk uncarefully about reasons, no matter what philosophical theory we have about them.
There are two ways to avoid such confusions: to be careful, or to just not talk about reasons at all. But not talking about reasons doesn’t make them go away.
And the fact that there are multiple senses of ‘reason’ doesn’t suggest that we have no way of figuring out which sense we are using at any given time, any more than the multiple senses of ‘bank’ suggest that we have no way of figuring out which sense we are using at any given time. Just as whether we are interested in financial transactions clues us into which sense of ‘bank’ we are interested in, there are earmarks to clue us into which sense of ‘reason’ we are interested in – if we’re only careful enough to observe what they are.
I agree, of course, that philosophers have confused themselves about many things by talking about reasons. But that simply doesn’t look anything like an argument that any particular theory about reasons is false, as far as I can tell.
Ralph.
You write:
It is worth noting that there is compelling linguistic evidence that the term ‘ought’ is context-sensitive in exactly the same way.
I’m not sure how this cuts against thinking that reasons are a or the basic normative notion. If one sort of normative notion tracks the context sensitivity of another normative notion, you could think (and I do think) it counts in favor of seeing them as at least systematically related if not possibly definable in terms of one another.
I agree with Mark that it’s hard to identify an argument against reasons being the basic notion here. But if there’s any resources for arguing for this in Ralph’s post, it’s in the suggestion that normative reasons are a subclass of explanations (a view I’ve also defended). Plausibly, normative reasons are to be distinguished by other kinds of reasons in that they are explanations relative to particularly NORMATIVE questions. And presumably what makes a question normative is that it involves normative concepts: concepts that must therefore be more fundamental than the concept of a normative reason. I’ve argued myself that this has to be the concept of value (and can’t be OUGHT), since reasons have to be weighed against each other and aren’t verdictive. Of course, this turns the Buck-Passing theory of value on its head.
Following up on Mark vR: I do, as a matter of fact, think that the objective and subjective senses of ‘ought’ derive from the fact that each is analyzed in essentially the same way in terms of the objective and subjective senses of ‘reason’, respectively.
In fact, it turns out that this can solve some puzzles about how the objective and subjective senses of ‘ought’ could be analyzed in terms of one another. I won’t get into it, here, but though what you objectively ought to do depends on what is the case, and what you subjectively ought to do depends on what you believe, it doesn’t work to say that what you subjectively ought to do is what you objectively ought to do in the world as you believe it to be. Most saliently, it fails due to what my colleague Jake Ross calls the ‘3 envelope problem’, which arises due to incomplete information. But there is no such problem for a similar interconnection between objective and subjective reasons.
I shouldn’t have made it sound as though I was giving an argument against the claim that SOME notion expressible by the term ‘reason’ is the basic normative notion. I wasn’t. (In fact, I find Steve’s argument against this claim quite appealing, although I’d favour the view that “normative reasons” are explanatory in relation to a question involving some normative notion, such as a question about what attitude or action is correct or appropriate or rational or the like.)
I was really only criticizing those who suppose that there is a unique concept expressed by “normative reason”. In describing the “reasons” program, I said that its proponents thought that THE notion of a “normative reason” was the basic normative notion. So Mark Schroeder isn’t a proponent of the reasons program as I was understanding it in my post, since he doesn’t think that there is such a thing as THE notion of a “normative reason”.
Some more specific responses:
Jussi – I will just deny your claims about your “brainwashing” example. On the one hand, if brainwashing really makes it rational for you to shoot the pedestrians, then I will say that something about your total mental state (your experiences and apparent memories, the beliefs and other attitudes that it is reasonable for you to have, etc.) really does count as a “reason” (in one sense of the term) for you to shoot them. On the other hand, if in this case you do not in any sense have a “reason” to shoot them, then I deny that it is rational for you to shoot them. (Of course, if you have gone completely insane that is another kind of “excuse”, and so the mere fact that an action is excusable does not all by itself imply that there was any sort of reason for you to do it; it is only excuses of the sort that involve rationality that imply that there was some sort of reason.)
I also agree with Mark Schroeder’s rejection of this attempted definition: “what you subjectively ought to do is what you objectively ought to do in the world as you believe it to be”. But my objection is simply that if you have an irrational belief about what you objectively ought to do, then this definition results in unacceptable “bootstrapping”. Even if we amend the definition so that your beliefs about what you objectively ought to do aren’t included in “the way you believe the world to be”, we still get counterintuitive results in cases like Jussi’s brainwashing case, I think.
Ralph,
I agree that there is sense of reason which makes it the case that I have reason to shoot people in this case and that is related to the set of mental states I have. This sense of the reason is what people refer to as motivating or explanatory reason. But, do I really have normative reason in some sense of normative reason? A good normative reason? I’m unfamiliar with that sense of normative reasons.
Also, I thought that the point was that whatever explains rationality of the action is the normative reason in the other sense. If it is the brainwashing that is supposed to make the action rational (as you write yourself), then by your lights brainwashing should be the normative reason in the other sense and not the set of mental states that result from the brainwashing. This is what I want to resist. Explanations for rationality are wrong kinds of things to be normative reasons.
Jussi — I didn’t say that brainwashing was a normative reason for shooting people. I said that in some possible cases of “brainwashing”, some of the mental states that the agent acquired as a result of such brainwashing constituted such a reason.
I don’t know why you say that “Explanations for rationality are wrong kinds of things to be normative reasons”. You give no argument for this claim. I would have thought that whenever a belief is rational (or as some epistemologists would say, “justified”), even if the belief is false, the believer must have some normative reasons for this belief, and it is precisely those reasons that explain why the belief is rational. Not even the most hard-line externalist in epistemology (Tim Williamson, say) would dispute this claim. Why do you dispute it?
Are you really “unfamiliar” with a sense of “normative reason” according to which an agent who acts on a rational false belief has some “normative reason” for the action? As I’ve just argued, any such agent has some normative reason for his false belief. So why aren’t the reasons for this belief also among the reasons for this action? Surely there’s nothing strange about saying that the man in the maze “has a reason” for turning Right (even though this is in a sense a misleading reason, and really the man would be better off turning Left).
I don’t dispute that. What I had in mind was the in the brainwashing case, the rationality of the belief that I should do such and such seems to be a matter of only coherence between the mental states the agent has. If this is what it takes for rationality I’m not sure it gets you to justification or good reasons for having the belief. I thought justification or good normative reasons for beliefs would have to have something to do with considerations that make the belief more likely to be true. And, mere internal coherence – rationality – for which there is an odd causal explanation doesn’t seem to give us that. So, I don’t see the immediate connection from rationality to having a reason for the belief. That’s my reason for disputing that connection.
I am. Many Germans in the Nazi Germany probably had good reasons to believe that Jews were to be killed as an inferior race. That’s what all the available evidence they were allowed to see pointed towards. Did they have any normative reasons to exterminate them? No. We can point to belief desire pairs they had that causally explain their actions and maybe weakly rationalise it. This gets us to motivating/explanatory reasons but not to good reasons.
Jussi — I do not accept that it was rational for anyone in Nazi Germany to believe that the Jews were inferior. Moreover, even if it were rational for them to believe this, that would not make it rational for them to believe that they ought to kill all European Jews. In short, I deny that it was rational for those Germans to attempt to kill all the Jews; so I also deny that they had any normative reasons for killing them. It is only when we rationally act on a rational false belief that the normative reasons for our belief also count as normative reasons for our action.
Suppose, however, that we had compelling reasons to believe that an approaching armada of spaceships was an invasion force of aliens who would kill us all unless we killed them first. Then it might be rational for us to launch a genocidal attack. But in that case there is a reason for us to attack, isn’t there – even if in fact the aliens are entirely peaceful and the evidence that we have had is tragically misleading?
By the way, I didn’t say that rational or justified belief just consists of coherence among one’s beliefs. It is crucial that one’s experiences, apparent memories, and immediate intuitions should support one’s beliefs as well. But surely your beliefs could still be rational or justified even if you are a brain in a vat, or are in some other way being radically deceived. Even in this case, there may still be a sense in which your beliefs are more “likely to be true” if they are more rational than if they are not rational; but it is a highly controversial question in epistemology what exactly that sense is — we surely don’t need to get into that question here.
Jussi, Ralph – I think you’re somewhat at cross-purposes, here. Ralph is trying to distinguish the subjective nomative sense of ‘reason’ from the objective normative sense; Jussi says there is no such ‘normative’ sense, only a motivating sense.
Well, in a sense, Jussi’s right – most uses of the phrase, ‘normative reason’ in moral philosophy are intended to pick out the objective sense. Following those uses, he finds it weird when Ralph calls subjective reasons ‘normative’.
But Ralph is also right – claims about subjective reasons are normative, not just descriptive. Compare the two senses of ‘ought’ – the senses in which Bernie, who rationally believes his glass to contain gin and tonic, ought to take a sip, and the sense in which he ought not to. The former sense isn’t just a description about what he is going to be motivated to do – it constitutes some kind of evaluation of him, saying that it would be rational of him to take a sip.
Ralph is also right that the objective/subjective distinction cuts across both practical and epistemic reasons, but that in epistemology we are most interested in what we have subjective reason to do. Duh – we know that what we have objective epistemic reason to do is to believe the truth; the problem is how to go about figuring out what that is.
So another immodest plug for ‘Having Reasons’, in which I do my best to clear up as much about this very question as I can, in the most careful way I know how.
I’m in complete agreement with Ralph that much contemporary metaethics errs in taking “normative reasons” to constitute a single context-independent class. But whereas the instance of this discussed here so far is the objective/subjective distinction, I think the divisions are much more fundamental. (In fact, I’m not completely convinced that there really are ‘subjective’ reasons). Instead, I’ve argued, “normative reason” talk is relativized to particular ends. Considerations only count as “reasons” in relation to the pursuit of particular ends.
The nicest example of this I have: an episode of the Amazing Race, when an African-American father/daughter team visited the location in Africa from which the slaves were shipped off to America. The daughter asks, “Do you think the slave traders told them where they were going?” and the father replies, “No – what reason did they have to tell them?” I presume that he would not deny, if challenged, that there were MORAL reasons. But as I interpret this case, he’s assessing reasons as explananda for the question: “Why would it be good for the achievement of the slave traders’ ends to tell the slaves where they were going?”
The standard objection here will be that he’s invoking motivating rather than normative reasons. I don’t think this is right (I agree with Dancy that ‘motivating reasons’ strictly speaking are just normative reasons that motivate). It seems to me quite proper for the father to allow that the slave traders might have had some reasons (relative to their ends) of which they were unaware (and which therefore could not be ‘motivating’ reasons). I take him to be denying even that.
In response to Jussi, this provides resources for explaining BAD normative reasons. There really may have been normative reasons, relative to Hitler’s goal of exterminating the Jews, to use gas chambers. (They can’t be merely motivating reasons, because he may have been unaware of them). But I would claim that these are BAD reasons. (“Bad” here expresses the ends that I am committed to, while they are “reasons” in relation only to Hitler’s ends, so there is something funny about the contextual dynamics here, but I think it’s something found in ordinary discourse).
Shameless plug: I’ve argued for all this in “The Reasons that Matter”, AJP 2006.
I posted before I read Mark S’s last, but evidently we at USC are given to shamelessness and immodesty in our plugging!
OK, Ralph, your clarification of the “Reasons Program,” and what you says to follow up on it make me think I don’t disagree as much as I thought. And I think you’re right to argue that even “subjective” reasons are normative, even if (as Mark S says) the term ‘normative reason’ has been appropriated by some to mean something like objective reason. (I’m actually worried that some uses of ‘normative reason’ are so hyper-objective that no agent could act on them, hence the move to ideal advisor views to cash out the idea of an objective reason.)
The one thing I would add about subjective reasons is just that I think that within the class of subjective reasons there are different grades (if you will)relativized to different bases of evidence or grounds, and that in many conversational contexts we manage to make one or the other of these salient. What a person has a reason to do is then evaluated relative to those grounds and is roughly what it would make sense for a person to do given awareness of those grounds. But without going fully objective we might in another conversation describe the same case differently depending on which grounds were salient in that context. I’ve got a bit of this idea playing a role in an old paper of mine written in response to Robert Johnson’s (rightly) highly regarded Conditional Fallacy paper from the Phil Quarterly.
I think though that if we think of context playing a role to highlight the grounds on which we make judgements about reasons it is probably misleading to think that there are distincive senses of reason at work here. For, so far as I can tell, once we fix on the grounds we are using what seems to me the same notion of what it makes sense to believe or do given those grounds.
Ralph,
I think I agree with you about actual Germans at the time. There was far too much counter-evidence to the claims of Germans. But, it seems to be that we can create a thought-experiment in which ordinary Germans only had evidence for the mistaken belief about races and their moral status and what should be done to them. Thinking of South before and slaves gets you close. In that case it looks like we want to say that their beliefs about reasons were rational but lacked support of normative reasons that could explain the rationality.
In the alien case I agree that there is a good normative reason. We should adopt the long term policy of responging to threats because that improves our chances of survival. We don’t have to say that the reason is the innocent aliens. Their appearance just is a derivative, normatively empty reason we have to take into account in applying our policy that’s supported by the basic reasons.
I guess I would want to say that in the BIV cases the contents of our beliefs are different and most of our beliefs are true about what they are about even in that scenario.
About the subjective reasons and oughts I would like to go the Kolodny way and say that they are not really normative but only normative-by-the-lights-of-the-agent kind of way. I’m not sure what is meant by normativity if it’s not something to do objective reasons – that which is really favoured and not only has the appearance of it. This goes for Steve’s Hitler case. I can’t get myself to believe that he had normative reasons to do what he did. But, given Hitler’s values, from his perspective it looks as if he has reasons to do things he did. I do accept that there is a sense of reasons we use to refer to this but why say that it is a normative sense?
Ralph suggested that there might be a plurality of normative statuses that normative reasons could explain,
Mark vR suggests that what varies is the scope of grounds, but that there may be just one notion of a reason, related to “what it makes sense to believe or do given those grounds”.
It seems that we can have both: relative to some fixed grounds, we can e.g. have many different (overall) normative statuses (say, “doing x is supported by at least one reason”, “doing y is reasonable”, “doing z is what one ought to do”) – it may well be that doing x is not reasonable, and that one ought not do it, but that there is at least something that speaks in favour of it.
So given fixed grounds, “giving a reason”, and “making reasonable” or “making required” seem to be different relations.
(I’m echoing here Dancy’s distinction between reasons and ought-making).
(I would add that what gives a reason may be a single consideration, whereas what makes something reasonable or required seems to be the entire set of grounds).
Jussi – Kolodny’s view, if I recall correctly, is a view about subjective reasons. It is the view that they are analyzable in terms of objective reasons, and in a certain way: not what you believe to be objective reasons, but what the objective reasons are in the world as you take it to be. It’s my view, too.
Moreover, here is what Ralph might mean by ‘normative’ – you know, that stuff that metaethicists get interested in, and wonder how it fits into the natural world, or whether it requires positing a sui generis realm of values or seriously revising our picture of how the semantics of the language we use in that area works. Subjective reasons are obviously normative in that sense – they’re certainly not on the other side of the normative/descriptive divide.
Of course, many philosophers have the theory that what makes things normative in that sense is a relationship to objective reasons. I share that theory and have exploited it in print. But on the view under consideration, subjective reasons are analyzable in terms of objective reasons, so no problems, there. And the word ‘normative’ has sometimes been used in print, usually by people following Broome’s usage, I believe, to refer stipulatively to whether something involves objective reasons. It is also sometimes used, along with ‘evaluative’, to exclusively and exhaustively divide the realm of things that interest metaethicists – I’m not sure that I understand exactly how this classification works, but it’s one to watch out for. In any case, I think it was clear that Ralph had something like my very general use in mind, as do many philosophers who use that term.
In any case, it doesn’t really matter for Ralph’s central point: if English speakers can succeed in talking about subjective reasons by saying things like ‘Jon has a reason to do such-and-such’, then that’s all Ralph needs, for his claim that ‘reason’ can be used in different ways, over and above distinguishing explanatory and motivating reasons. I think a wealth of examples show that ‘reason’ can indeed be used in precisely this way in English, and several have come up, here.
1. I think that Jussi’s latest comment shows that Mark Schroeder’s initial irenic suggestion that Jussi and I are just talking past each other is in fact mistaken. Jussi really is a supporter of the “reasons” program that I was complaining about. I agree with what Mark says in reply to Jussi here (although actually I think that Jussi is correct in his interpretation of Niko Kolodny – which means that I am committed to disagreeing with Kolodny on this point as well …).
2. I agree with Steve that in addition to the distinction between “objective” and “subjective” senses of ‘reason’, there are also different kinds of reasons, such as moral reasons, or the reasons that are relative to various specific goals or aims. Still, I doubt that all of these different kinds of reason correspond to different senses of the term ‘reason’. Compare: there are different kinds of knowledge, such as perceptual knowledge and mathematical knowledge, but we wouldn’t want to conclude that there are many different senses of the term ‘knowledge’. To defend the view that there are different senses, we would need a clear pair of cases where the truth value of a sentence differs depending on which sense of ‘reason’ is in play. Steve’s example from Amazing Race is more simply explained in terms of the familiar phenomenon of restricted quantification: in this context, the speaker is simply quantifying over a restricted domain – roughly, over the reasons that one might expect the slave traders to recognize, or something like that.
Moreover, I am pretty sure that I disagree with Steve’s idea that all reasons are relativized to specific ends. Surely for every end E, we can ask, “What reason do you have to aim at end E?” The sense of ‘reason’ that appears in this question surely can’t itself be relativized to any specific end …?
3. Mark van Roojen is quite right that one view that we should consider is the view that different occurrences of ‘reason’ or ‘ought’ may express the relativization of the very same concept to different bodies of evidence or information. So, on this view, the very same relational concept ‘x is a reason for y in relation to body of information I’ is part of what is expressed by all these occurrences of the term ‘reason’. What I said in my post is compatible with this view.
In fact, however, I don’t accept this view. My view is closer to Arto’s idea that in fact we have both sorts of contextual variation – i.e., both variation in the sort of normative status that the “reason” in question is explanatory of, and variation in the body of evidence or information that this normative status depends on. But it would take too long for me to explain why I hold this view. (Roughly, it is because of the way in which I believe the two sorts of normative status will come apart in cases where one suffers from error or ignorance about the purely normative propositions.)
Mark,
A comment about this:
“Kolodny’s view, if I recall correctly, is a view about subjective reasons. It is the view that they are analyzable in terms of objective reasons, and in a certain way: not what you believe to be objective reasons, but what the objective reasons are in the world as you take it to be. It’s my view, too.”
I’m not sure this is right. The view is mainly about rationality. And, more precisely the idea is that rationality only has apparent normativity – not real normativity, whatever that is. Rational requirements only feel normative from the inside. The rationality he is interested in is then the requirements between beliefs about reasons and the reactions to having beliefs about reasons (the right intentions to act).
The charge of irrationality in this case is criticism for not reacting to reasons as they seem to us. It’s true that he does grant that there is a different charge of *objective irrationality* which is being insensitive to objective reasons.
Here are good quotes:
“When we tell someone, in the register of advice, rather than that of appraisal, that he ought to rationally have attitude A, or that it would be irrational of him not to have it… [w]e are making the descriptive, psychological claim that he believes that he has conclusive reason for that attitude. We are telling him, as we might put it, that *from his point of view*, or *as it seems to him* he has conclusive reason to have the attitude (Kolodny, Why be Rational?, 557).”
“The (seeming) normative force of the ‘ought’ of rationality derives from (seeming) reason, the reason that the subject believes he has”.
Getting back to the original post, we can ask what then explains the rationality of our actions and intentions. If actions and intentions are reactions to beliefs about reasons, the explanation of rationality needs to go back to our beliefs about reasons. What explains the rationality of shooting the apparent enemy soldier in Ralph’s case? On this view this is rational because it fits what the agent beliefs he has reason to do.
I agree that we often say in this situation that the agent’s reason was his believing that the enemy was about to shoot him. I have no problems with this. But, I don’t think that the ordinary usage extends to the term ‘normative reason’. Good reasons is a better term in normal language. However, I do think it is an open question whether we think that false beliefs are good reasons. And, in the Kolodny account we seem to be able to have a descriptive account of rationality that uses only explanatory reasons.
Am I part of the reasons program? I’m not sure. I do think that normative reasons can be used to account for many other normative and evaluative notions. Probably not everything. I am open to persuaded to think that reasons are less useful than what I think in making sense of things.
I agree with Mark S’s first post that if we are clear about what we mean by normative reasons, then it isn’t much of an argument against the project that reasons are used in other senses.
I started also to think about an analogical argument that resembles Ralph’s original post:
“Many physicists today are pursuing a program according to which the notion of a “force” is the most fundamental physical notion. Thus, these philosophers aim to analyse such notions as ‘accelaration’ and ‘weight’, the notions of the various sorts of “friction”, and so on, in terms of ‘forces’.
In my view, this program suffers from a serious flaw: it fails to recognize quite how profoundly context-sensitive the term ‘force’ is.”
And, then we could go on to give various examples where ordinary language uses the term ‘force’ in different contexts in different ways that physicists use it. That wouldn’t be much of an argument. I wonder why the reasons argument is supposed to be different?
Jussi – when metaethicists divide the world into the normative and the descriptive, in order to explain what their subject matter is supposed to be, would you say that rationality goes on the normative side, or the descriptive side? Hint: one of these answers exhibits a dim grasp of the sort of thing that metaethicists do.
My point is: if speakers of English can say things using the word ‘reason’ to talk about things that you would appeal to Kolodny in order to explain, but Kolodny is talking about rationality, and rationality is distinct from reasons but also falls on the normative side of the normative/descriptive divide, then there are two things on the normative side of the normative/descriptive divide that we can use the word ‘reason’ to talk about. Which, if you recall, was all that Ralph said, when he said that even restricting to normative senses of ‘reason’, there are still multiple senses in play. So where does this go wrong?
I take it that that is a substantial disagreement in the metaethical study of rationality that is hotly debated. I’ve had a number of most interesting discussions with prominent metaethicists about the question whether rationality is normative or descriptive. I’ve talked to people who take either side of the question and as far as I know these people are at the top of game so to say. I really find this tone quite patronising.
Jussi – here’s me: ‘rationality is one of the things studied by metaethicists, and some people use ‘normative’ to mean – you know – the kind of thing studied by metaethicists. In that sense, Ralph’s claim about there being two normative senses of ‘reason’ is obviously right’. And here’s you: ‘no, mark; it’s controversial whether rationality is one of the things studied by metaethicists. In fact, the question of whether it is or not is one of the hotly debated issues studied by metaethicists.’ I feel like something has gone wrong, somewhere; I just can’t put my finger on it.
Ralph,
Thanks for engaging with my suggestions. You write,
“I am pretty sure that I disagree with Steve’s idea that all reasons are relativized to specific ends. Surely for every end E, we can ask, “What reason do you have to aim at end E?” The sense of ‘reason’ that appears in this question surely can’t itself be relativized to any specific end …?”
Although I don’t think it’s what you’re getting at, let me first observe that (a) we may not have reasons for our ultimate ends (per Hume), and (b) we may have reasons to aim at an end that are reasons in a different sense from the reasons we have to pursue the means to those ends (i.e. aiming at the end may promote a different end).
What I think you mean, however, is that when you ASK someone for their reasons, you’re not assuming something relativized to any particular ends — so here at least “reasons” can’t be so relativized. What you might mean, however, is “In doing X, what considerations did you take to be indicative that X-ing would promote SOME end of yours?”
Admittedly, the semantic value of “reasons” here is not filled out by any SPECIFIC end. But it’s still arguably the case that whatever one’s reasons may have been, they were reasons FOR THAT PERSON in relation to some specific end. We can similarly talk about “explanations” as a class, regardless of their explananda [I misused this word in my previous post.]
My view no doubt looks unnecessarily messy. The back story here is that I’m trying to meet two desiderata: (a) that I accommodate the reasons judgements that we ordinarily make, and (b) that reasons turn out to be entities that really can justify and motivate (and really exist). I don’t believe that any simpler account of reasons can pull this off, but that’s too much to argue here.
Jussi:
I sympathize with your resistance to calling my “bad” reasons “normative”. The problem here, I think, is that we metaethicists have invented a defective concept in the “normative”. (Note that nonphilosophers never call anything a “normative reason”). Calling something “normative” suggests endorsement of it, but this isn’t what I intended. These reasons are “normative” in that they stand not in a motivational relationship to an agent, but in a relationship which (for want of a better characterization) if cognized can ‘rationally’ motivate an agent.
Now that Mark S. has taught me some html, I’ll provide a link to that AJP paper:
The Reasons that Matter
Jussi — As I said, I agree with your interpretation of Niko Kolodny. But I disagree with the view that you correctly ascribe to him.
In my view, the mere fact that you believe that you have an objective reason for shooting the man who is approaching is neither necessary nor sufficient for its being rational for you to shoot.
In general, the view that ‘rationality’ is not a genuinely normative notion seems an outrageously false view to me, but it is certainly a view that some extremely clever people (e.g. Kolodny) have defended. The argument that Mark has just given against this view seems fine to me. But clearly this argument won’t impress the proponent of the view that ‘rationality’ is not a normative concept (although I dare say that we could build on Mark’s argument to make it a more dialectically effective weapon against this view). I haven’t really given any arguments against this view in this thread here on PEA Soup; I may try to post something about this in future.
Steve — OK. Let me try this. I don’t think that it’s a conceptual truth that it couldn’t be the case that for every end E that you have, you have a reason for having E. (I.e., even if Hume is right, I don’t think he’s right as a matter of conceptual truth.) Moreover, is it really plausible that statements about our reasons for belief are relativized to specific ends? And wouldn’t it be a very unfortunate view if there were no systematic connection between the meaning of ‘reasons’ in statements about reasons for belief and statements about reasons for action (and for that matter, statements about reasons to feel, such as reasons to feel worried or reasons to feel guilty, etc.)? So I remain pretty sceptical about the view that “reasons” statements are implicitly relativized to ends.
P.S. On Jussi’s analogy with physics — like most other philosophers, I believe that there are some basic differences between the methods of philosophy and the methods of physics, and that these differences oblige philosophers to worry more about language than physicists have to.
Philosophers rely on “intuitions” more than physicists; and in at least some cases, these intuitions are just linguistic intuitions about the truth or falsity of certain sentences in certain contexts — in which case we need to be constantly aware of the danger that these linguistic intuitions relate to different concepts. Moreover, philosophers are trying to develop accounts or explanations that answer certain questions that can be phrased using the ordinary-language senses of terms like ‘knowledge’, ‘right’, ‘wrong’, and so on; and so it remains important for philosophers to continue using the ordinary-language senses of certain crucial terms.
Ralph,
I agree that it’s possible to have a reason for every end that you have (that’s what coherentist justification gives you). I meant to say that it’s possible that we don’t have reasons for our ultimate ends (per Hume), not that it’s not possible that we do have reasons for them.
Regarding reasons for belief: I confess I haven’t given those as much thought as I ought. Here’s a response, shooting from the hip (at 2:30am — I’ll probably regret it after some sleep):
I’d distinguish between (the concept of) reasons to BELIEVE that p, and (the concept of) reasons why it’s PROBABLY THE CASE that p (i.e. evidence). I mention the latter because I think that this is what people commonly mean by “reasons for belief”, and I don’t think that they are normative reasons in the sense that reasons to act are (they’re not explanations in relation to an end). The former are explanations of why believing that p would promote certain ends. Among these reasons, the EPISTEMIC ones are those that explain why your believing that p would probably realize the end of your having true beliefs about p. (To count as epistemic, it would have to be a constitutive rather than instrumental means). Observe, therefore, that the reasons why it’s probably the case that p and the epistemic reasons to believe that p coincide: two different concepts with the same extension. Therefore it’s not really an error to call evidence a reason to believe.
I’m sure others will point out to me why this won’t work.
So, that something is studied by metaethicists makes that very thing by definition ‘normative’. I like that.
I agree that the belief that you have objective reason to shoot is neither necessary or sufficient for rationality. Rationality judgments must be more holistic than that. The rationality of actions is assessed in terms of sets of normative and non-normative beliefs the agent holds and whether the action is appropriate with the set.
In your examples, on the necessity side, I’d like to point to the fact that you are using rationality in the over-all sense. For over-all rationality the belief about the reason may not be sufficient as one is forming beliefs in an irrational way. But, shooting may be rational to a degree as far as the belief goes. In the uncertainty case, surely you have some degree of belief to having the reason. Wouldn’t that be enough to rationalise the action?
Here’s another worry. Let’s grant that rationality is normative and that (even false) beliefs and desires of the agent explain the normative status of rational action. Let’s also call the belief/desire pair reasons. Have we established that these reasons are normative? Only if we assume that everything that explains the normative status of something must be normative itself. I’m not sure this is true. A broken nose can explain why hitting something was morally wrong. This is an explanation for the normative status of the action but it doesn’t make a broken nose normative.
I also like the fact that it is impossible for a *metaethicist* to argue that something is *not normative*. Doing so would make that thing a subject of metaethicists investigation and thereby normative.