A recently influential idea in the
philosophy of normativity is reasons primitivism. Reasons primitivists
hold that we can give no account of what it is for some consideration to be a
(normative) reason. At most we can say that reasons are considerations that count
in favour (or against) some response, or that when there is reason to do
something, there is something to be said
for doing that thing. In this post (which is intended in the spirit of Dave and
Dave’s request for half-baked ideas…) I want to raise a worry about this view.
While reasons are considerations which
count in favour, it does not seem that all considerations that count in favour
are reasons. For example, it seems that the fact that being tall makes it easier to see bands at concerts counts in favour of being tall. But I don't think we should say that this is a reason to be tall. Reasons are the sorts of things which can justify or make rational. But being tall is not the kind of thing which can be rational or justified. Similarly, if I would enjoy the film, then that seems to be something to be said for seeing the film. This seems true even if it is not possible for me to see the film. By contrast, if it's not possible for me to see the film, there seems to be no reason for me to do so.
These examples suggest that there are
certain necessary conditions on reasons which do not apply to favouring as
such. For instance:
R
is a reason for S to A only if A-ing is a type of action or attitude.
R
is a reason for S to A only if S can A.
Others – including some reasons
primitivists – have suggested further conditions on reasons. For example, some
suggest:
R
is a reason for S to A only if S can A for the reason that R.
R
is a reason for S to A only if S can know that R is a reason to A.
Again, neither of these conditions apply to
favouring. Thus to borrow an example from Mark Schroeder, if Nate loves
successful surprise parties but hates unsuccessful surprise parties, then the
fact that there is a surprise party at home counts in favour of Nate’s going
home. But it is not a reason for which Nate could go home. Nor could Nate know
that this fact is a reason for him to go home.
Is it legitimate for reasons-primitivists
to endorse conditions of this sort? On the face of it, such conditions appear
puzzling. In general, necessary connections require some sort of explanation. A
standard way to explain a necessary connection is by appealing to the nature of
one of the notions or properties involved. But reasons primitivists don’t seem
to be able to do this. There is nothing in the notion of a favourer which
supports these conditions. But reasons primitivists claim that this is all
there is to the notion of a reason. So it is unclear how reasons primitivists
can endorse these conditions. However, if reasons primitivists cannot endorse
conditions of this sort, that seems to undercut the plausibility of the view,
insofar as at least some such conditions seem highly plausible.
I’m no friend of reasons primitivism, but notice that this seems to be an analog of Ought Implies Can. I believe (following Michael Stocker and Walter Sinnott-Armstrong) that OIC is only a pragmatic principle–and I’d say the same thing about RIC. (Roughly, there normally isn’t any conversational point in telling someone that they ought/have a reason to do something that they can’t). Reasons primitivists could say the same thing.
I expect many people’s intuitions about RIC aren’t as clear cut as yours, which may be influenced by your theory of what reasons are.
I’m not sure that I see the worry – reasons primitivists presumably don’t want to say both that the reason-providing relation is unanalysable and also that it can be analysed in terms of counting-in-favour-of. Rather, they want to say only that the reason-providing relation is unanalysable. The “counting in favour” talk is not supposed to be an analysis of the reason-providing relation, but is instead only supposed to do something to *indicate* which (other) unanalysable thing it is we’re talking about. If that’s right, then I can’t see why they should worry if “is a reason to” and “counts in favour of” are not extensionally equivalent. I’m being moved by a more general thought here, which is that any claim to the effect that X is unanalysable can’t possibly face counterexamples (for better or worse).
I’m less sure about Steve Finlay’s worry above. Isn’t the pragmatic story for RIC less plausible than it is for OIC? (Reasons stand at the intersection of justification and explanation, and all that.) Regardless, I still think it would be interesting to know if reasons primitivists were forced into holding the pragmatic explanation of RIC.
I am not clear yet why they can’t capture the truth in your assumptions by looking at a broad range of attitudes. For example, when you say that, “it seems that the fact that being tall makes it easier to see bands at concerts counts in favour of being tall,” they can agree and say you are pointing to a reason to be glad you are tall if you are and to wish you were tall if you are not. Seems to me that you need to face this kind of response to solidify the intuitions that are supposed to ground the objection.
I think there are two ways we can go here. One is to say that you do have a reason to be tall but deny that all reasons must justify or make rational (especially the connection between reasons and rationality assumed here is controversial). The second is to say that we should not say that being able to see bands counts in favour of being tall. You give an argument for this too: things that count in favour of are things that justify and make rational.
I should also add that didn’t Scanlon say that reasons are only for things that are sensitive to judgments. And, he agreed with Alex that the talk about favouring only serves an illuminating purpose. I think there is an explanation for this condition too.
I strongly disagree about surprise party case in which I think there is a reason to go home.
Interesting post, Jonathan. I don’t see why the reasons primitivist must accept that there are no non-trivial necessary truths about reasons. When we think of other views where some kind of thing is taken to be metaphysically basic or some concept is taken to be unanalyzable (in terms of necessary and sufficient conditions, that is), we don’t see philosophers accepting a ban on on non-trivial necessary truths. For example: Williamson (2000) claims knowledge is basic and that the concept is unanalyzable, but also makes a whole lot of interesting claims about it (e.g. necessarily, for X to be knowledge, X must be a safe belief).
Thanks for the comments everyone!
Steve – you’re right, of course, that RIC might turn out to be false. I tend to find Peter Vranas and Bart Streumer’s arguments for it (and against the pragmatic story) fairly convincing. Even if I’m wrong about that though it’d be interesting (as Alex says) if reasons primitivists were committed to denying RIC (and the other conditions on reasons). Commitments of this sort look like costs of the view. For instance, one common response to the wrong kind of reason problem is to appeal to the idea that reasons have to be the kinds of things one can respond to (the third condition I mention in the post). If reasons primitivists can’t appeal to this condition, they can’t avail themselves of that response to the problem.
Alex – you’re right that the ‘counts in favour’ talk isn’t typically put forward as an analysis. But the objection isn’t simply that reasons primitivists face counter-examples. The worry is that they are unable to explain conditions on reasons which seem plausible and which some might hope to do work (as in the wrong kind of reason problem, as I say in response to Steve).
Brad – Sorry, I’m not quite sure I follow. The response you suggest seems to concede my point that not all considerations which count in favour of A-ing are reasons to A. I think that’s all that’s needed to raise the question of what explains conditions on reasons which don’t apply to favourers as such.
Jussi – there’s quite a few points there! I think the first response is covered by some of the points above. The second response might be the way to go but at the moment I’m not seeing why the notion of favouring has to be bound up with notions of rationality and justification. (Whereas it seems very plausible that there are connections between these notions and reasons). You’re right about Scanlon – but my question is why reasons primitivists get to assume those sorts of conditions. I’d be interested to hear your explanation.
Daniel – you’re right that primitivists about F don’t have to accept bans on necessary connections involving F. But the suggestion in my post is only that necessary connections (tend to?) require explanation. Such explanations don’t have to appeal to the nature of F. They could instead appeal to other properties/concepts involved in the connection. For instance, if I remember correctly, Williamson explains why knowledge entails belief by appealing to the nature of belief as ‘would be knowledge’. (I’m not sure if an explanation of this sort is available for the connection you cite). I don’t yet see how explanations of this kind could be provided for the connections I cite but perhaps some are available.
Ok. So it sounds like you will grant that “favoring features” do always provide reasons but that they are sometimes reasons to act or intend and sometimes reasons to have other attitudes (say to wish or be glad). You are calling the former but not the later ‘reasons’ but that seems to be beside the point.
You can then ask why principles (perhaps the ones you list) govern reasons to act or intend but not reasons to have other attitudes towards the relevant actions. Perhaps the reason-primitivist could try to explain this by discussing the kinds of attitudes in question? The idea would be that we can explain why the principles govern reasons to intend but not reasons to be glad by appeal to the difference between intending and being glad. Just an idea – not claiming this will work out.
Note that pragmatics can come into the story in different ways. One story (which you might be assuming) is that the problematic reasons-claims are strictly true, but infelicitous to say.
But another possibility is that there are pragmatic influences on how we interpret their semantic content. On a contrastivist story (like that of Justin Snedegar, for example) to say that R is a reason for S to A is always to say that it is a reason for S to A rather than to B, C, etc. instead. Pragmatics might make it natural to assume that the intended contrast-set consists entirely of actions within the agent’s control. In that case, on its default reading, a reasons claim will be false, and not simply infelicitous, if the agent lacks the relevant ability. I think a primitivist could adopt this story.
In general, the more interesting explanations provided by any theory, the better. However, I don’t see why we should think there is a particular demand to provide explanations of the sort being asked for here. Perhaps you will agree that a general theory of the normative and evaluative domains could be explanatorily weak in some areas and explanatorily stronger in other areas (e.g. if a buck-passing account of evaluative properties were successful for the reasons primitivist), and we could then compare whole theories to see how they fare. Or do you think there is something particularly bad about lacking explanations of necessary truths using the concept said to be primitive, above and beyond a general need for a theory to provide explanations? If so, I’d like to know why. I would have thought a primitivist in any domain will need to assent to some very general (necessary) truths, just in order to ensure that disputes with opponents are not merely verbal. This certainly seems so for debates about the nature of reasons, where there is a concern that disputes risk being merely verbal. One might even think that it is the very general truisms that all sides to these debates should agree to (which may or may not include reason implies can) which least require explanation. They may be brute.
Thanks again all!
Brad – yes, I’m happy to grant that if X counts in favour of Y then either X is a reason to Y (if Y is an action- or attitude-type) or X is a reason to have some attitude (e.g. desire, being glad, hope) towards Y. This isn’t to deny that the latter things are reasons. However, it is to grant that they are not reasons for the thing favoured – in effect, it’s to grant that there can’t be reasons for things like height, eye colour, or perceptual experiences. My question is then why there are no reasons for such things. And the worry is that reasons primitivists simply have to take it as a primitive fact that there can only be reasons for some things and not others. Now perhaps this wouldn’t be so bad in itself. But the more things that have to be taken as primitive the worse, I think.
Steve – yes, I think I was assuming the first story, which I thought was the Sinnott-Armstrong line (although I confess that it’s been a while). In any case, you’re right that the contrastivist story is another way to go. Thanks for pointing that out – I’ll need to think about it.
Daniel – Necessary connections are very strong claims. If it seems that the world couldn’t but be a certain way, it seems fair to ask why that is. Now perhaps – this probably depends on further issues about the nature of analysis or modality – some necessary connections are brute. But it still seems to me that the fewer necessary connections a theory posits, the better. Now since this is a comparative claim, you’re right that we’ll ultimately have to compare theories to figure out the upshot of my points in the post. Note though that, e.g. value primitivists don’t face the worry I raise here. Height, eye colour, and perceptual experiences can be good, just as actions and attitudes can. So value primitivists don’t face a task of explaining why only certain sorts of things can be good. That looks like one respect in which value primitivists have an advantage over reasons primitivists.
I’m not sure if I see the point about verbal disputes. I don’t think that saying that a necessary truth is a conceptual truth is to deny that this necessary truth requires explanation. Rather, it’s to provide an explanation. (Why are all bachelors unmarried? Because, given the concept of a bachelor, part of what it is to be a bachelor is to be unmarried)
If to say that a necessary truth is a conceptual truth is to provide an explanation (as you claim) then that’s all a reasons primitivist need do for any of the necessary truths about reasons she wishes to accept, no? (I hope things are not quite that simple; I’m not a reasons primitivist, so I’d actually quite like your arguments to succeed.)
Daniel – sorry, I should have put that more carefully. The thought was that one way to explain a necessary connection is to provide a conceptual analysis from which that necessary connection follows. But reasons primitivists don’t offer any kind of analysis of reasons, so they don’t seem well-placed to explain the necessary connections in that way.
Thanks for the clarification, Jonathan. I guess our discussion brings out that it is not obvious how reasons primitivism is best construed/defended (there is a metaphysical thesis and there is a conceptual thesis, and one might only accept one of them).