Sat 21 Nov 2009
‘Have’ and the reason relation
Posted by John Turri under Uncategorized
[24] Comments
Some people think that reasons are propositions. When you form a belief or perform an action, they say, your belief is based on a proposition, not another belief, an experience, a desire, or anything else mental. The basis of belief and action is one or more propositions. Call this view “abstractionism” about reasons (where ‘reasons’ picks out the basis of belief or action).
Here’s an argument against abstractionism:
- You can have reasons.
- You can’t have propositions
- So reasons aren’t propositions.
The argument is valid. 1 is obvious. But taking a cue from Mark Schroeder’s work, someone might object to 2, on the following grounds.
‘X has Y’ means ‘X stands in salient relation R to Y’ (or vice versa). For example, ‘I have a father’ means that someone stands in the father relation to me, not that I possess a father. And while we can’t possess propositions, we can of course stand in relations to propositions. So we can understand ‘I have a reason’ to mean that a proposition stands in the reason relation to me. Perhaps 2 seems true because we’re thinking of ‘have’ to mean ‘possess’. But really, it should be understood as naming the reason relation, in which case, 2 is false.
I don’t think this works, partly because I don’t think ‘have’ can be understood in the relevant way.
In general, when context serves to assign Y to X, via some salient relation R, it will sound fine to say ‘X has Y’. That’s because by uttering ‘X has Y’ in the context, we end up expressing the proposition that X stands in relation R to Y, that is, R(x,y). Depending on what relation is salient, this works fine where ‘X’ names a person and ‘Y’ a proposition. But it doesn’t seem to work when the salient relation is the reason relation. This spells trouble for the abstractionist’s response. [Ed. note: As comes out in an exchange with Brad C. below, I don't think this paragraph is essential to my point here. This is good, because it's not clear that the "in general" point is true.]
We can contrive contexts where it sounds okay to say ‘I have the proposition <Sarah published a book>’. Suppose you and I are playing a strange game. The game is to see who can think about their assigned proposition for the longest time. You get assigned <Barack published a book>, and I get assigned <Sarah published a book>. A third party asks, “Who’s supposed to be thinking about Sarah?” I respond, “I am. I have the proposition <Sarah published a book>.” This sounds fine.
Given that we can generate this effect in the context of our strange game, if the abstractionist proposal under consideration were correct, then we should expect it to sound okay to say ‘I have the proposition <Sarah published a book>’ when it’s salient that this proposition is my reason for believing something. But that doesn’t happen. Consider: I say, “Sarah published something.” Everyone asks what my reason is for thinking this. I respond, “Sarah published a book.” Everyone believes me. It still sounds ridiculous for me to say, “So I have the proposition <Sarah published a book>.” It would likewise sound ridiculous for others to say of me, “He has the proposition <Sarah published a book>.” I don’t even understand what that’s supposed to mean. (It would be fine, though, for others to say, “He has the belief that Sarah published a book.”)
Consider a different case. Barack and Joe both believe that Sarah published something. We ask them what their reasons are for thinking this. Barack says, “Sarah published a book.” Joe says, “Sarah published a journal article.” We all believe that they’re being sincere. Now someone asks, “So, which of these two gentleman has the proposition <Sarah published a book>?” The question is unintelligible. But why should it be unintelligible, if the abstractionist proposal currently under consideration were true? If it were true, we should understand the question to be ‘which of these gentleman stands in the reason relation to <Sarah published a book>?’. But we don’t. (It would be intelligible, though, for someone to ask, “Which of these two gentleman believes that Sarah published a book?”)
[Update: Here it might be worth repeating something I said in the comment thread below, since it helps further clarify the trouble I'm trying to articulate for abstractionism. "My point about the abstractionist proposal is this. Even when the reason relation is salient — we’ve just been asked what our reason is for believing Sarah published something, and we respond by saying ‘She published a book’ or whatever — it still sounds awful to say ‘I have the proposition [Sarah published a book]‘. I can think of other contexts where it makes sense to say ‘I have the proposition [Q]‘ — I understand what relation it’s picking out. (This is the point of my “strange game” example.) But it makes no sense in the context of the reason relation. Why would that be, I wonder, if the reason relation relates us to the proposition in question, and that’s very salient in the context? Why would ‘have’ have such a hard time picking out the reason relation? Why can’t we hear it that way?]
[Later update: Tomorrow's headline: Philosopher changes his mind.]
John,
Why are you going with a contextually assigned relation as opposed to one contributed by the sentence? In the father example, it sure looks like the word ‘father’ in “has a father” determines the relation (father/child) we are talking about.
But maybe I’m missing your point.
Hi Mark,
It seems like context still matters.
Just think what might be expressed if that sentence were uttered (shudder to think) at a slave auction. In that case, ‘have’ might name the master/slave relation, and ‘father’ contributes nothing to picking out the relation.
Another example: we’re wondering what faith someone belongs to. We ask, “Do you have a rabbi?” Response, “No. I have a father.”
Hi John,
Interesting post.
You wrote: “In general, when context serves to assign Y to X, via some salient relation R, it will sound fine to say ‘X has Y’. That’s because by uttering ‘X has Y’ in the context, we end up expressing the proposition that X stands in relation R to Y, that is, R(x,y)”
I am having trouble seeing that this holds true in general. Assume premise P entails conclusion C; the entailment relation holds between P and C. But uttering “P has C” does not express this relation. There seem to be lots of other counterexamples. For example, Giles is my father, but it makes little sense to utter “I have Giles” – I see no reason to think it would express the claim that we are in the father-son relation.
On the other hand, it might well make sense to say “I have a father” or “P has a consequence”, which suggests that the relevant sentence in the reason case would not mention the *individual thing* that is the reason, but the fact that there is some reason. “Jane has a reason to X” rather than “Jane has R to X”. And then the abstractionists seem to have no special problem.
But maybe I am missing your point!
Interesting stuff! Hereby a nitpick and a constructive idea both of which concern putative equivocations:
NITPICK: I think you mischaracterize the argument as “obviously valid” given the very objection you consider. The objection is that there is an equivocation on ‘can have.’ In (1) ‘can have’ may be spelled out as ‘stand in an appropriate relation to’ or the like. But that is not the sense of ‘can have’ that makes (2) seem right. If (2) was stated in terms of this sense of ‘can have’ it would be false.
The fallacy of equivocation is a fallacy. So, if it isn’t obvious that there isn’t an equivocation, it isn’t obvious that the argument is valid.
IDEA: Ordinary talk may equivocate on ‘reason.’ So, let’s distinguish between de facto reasons for p from doxastic reasons. The former are, roughly, propositions that stand in some appropriate relation to p. The latter are, roughly, propositions S stand in some appropriate relation to that stand in an appropriate relation to p. (This requires qualification – consider cases in which S reasonably, but mistakenly, takes q to stand in an appropriate relation to p).
Given such a distinction we may say that q *is* a reason for S to believe that p (or, in the realm of reason for action, to phi) even though S has no clue as to whether q and has not considered q. However, we may also say that S does not *have* (q as) a reason for believing that p (or, action theoretically, to phi).
Maybe de facto reasons don’t deserve the honorific ‘reasons’ although this is the vernacular term. To wit: There have always been good reasons to cut down on Co2 omissions – although we haven’t realized it until recently. Interestingly, people will often refer to de facto reasons (roughly, the worldly fact, the true propositions) when asked what reasons they have. To wit: Q: What is your reason for thinking that S published something. A: She published a book. So, there is definitely some conceptual slippin’ n’ slidin’ goin’ on.
Best,
Mikkel
PS: If phrases like ‘q is a good reason for her to believe that p although she does not have (q as) a reason for believing that p’ ring too wrong, we could call de facto reasons ‘grounds.’ Whichever way we go, terminological explicitness *may* save the day here.
I don’t like the view that reasons are propositions but then again I am suspicious of this argument too. First, note that your argument, if it works, also rules out the view that reasons are facts. The only view of reasons it would be compatible with I can think of is that reasons are beliefs.
Like many others, I tend to think of the talk of reasons in terms of the talk of counting in favour of. So, for me, 1 says:
1*) There are considerations that can count in favour of your beliefs.
Now, let us think that reasons are propositions. In this case, from 2, we would get:
2*) There are propositions that can count in favour of your beliefs.
This sounds fine so nothing like 3) follows. So, I do think the relational view is right. Consider also the argument:
1**) I can have a father
2**) I cannot have a human being
3**) Therefore fathers aren’t human beings.
2**) sounds true to me just like 2) does. But 3**) must be false. Yet, there’s no (not game-involving) context where fathers would be salient and in which 2**) would become ok to say. So, something must go wrong with the test you propose in the end.
Hi Mikkel,
One good nitpick deserves another: I did say the argument was “valid,” and I did say that 1 is “obviously true,” but I don’t think I said the argument was obviously valid.
But I take your point. We could understand the objection I consider either of two ways. One way grants that ‘can have’ is used univocally, and so the argument is valid, but unsound because 2 is false on this reading. This is the way that I understood it. Alternatively, we could understand it to grant that 2 is true, but only in a sense of ‘can have’ different from 1, in which case the argument is invalid because equivocal.
I agree that ordinary talk is very equivocal on ‘reason’, and that’s always something to be on guard against. Do you think I’m explicit enough when I say that I’m interested in “the basis of belief and action,” or alternatively, “the reasons we base our beliefs and actions on”? That is, I’m interested in one of the relata of the basing relation.
Brad,
Good points. I think the ‘in general’ claim is too ambitious. Re-reading the original post, I’m pretty sure I could just eliminate that paragraph without loss, which is good to know.
I think we can build up the context in such a way that it sounds perfectly fine for you to say ‘I have Giles’, thereby picking out the ‘father’ relation. Suppose you and I don’t know who our fathers are. We’re going to learn today. We’ve each been told that our father is either James or Giles. You’re presented with a slip of paper, which tells you who your father is. You read it. “Well?” I ask. You turn to me and say, “I have Giles.” This is fine.
Now, my point about the abstractionist proposal is this. Even when the reason relation is salient — we’ve just been asked what our reason is for believing Sarah published something, and respond by saying ‘She published a book’ or whatever — it still sounds awful to say ‘I have the proposition [Sarah published a book]‘. I can think of other contexts where it makes sense to say ‘I have the proposition [Q]‘ — I understand what relation it’s picking out. (This is the point of my “strange game” example.) But it makes no sense in the context of the reason relation.
Why would that be, I wonder, if the reason relation relates us to the proposition in question, and that’s very salient in the context? Why would ‘have’ have such a hard time picking out the reason relation? Why can’t we hear it that way?
Jussi Suikkanen submitted a comment, but for some reason I can’t to approve it. At some point perhaps it’ll get through. Here’s what he wrote (2009/11/21 at 8:23pm):
John,
I’m thinking that there might be further ambiguities that your response picks up on, since it is going to be part and parcel of the view of ‘have’ that has been proposed that there are at least two ways it functions — one as in ‘have a father’ in normal contexts, and the other as in ‘have a pocketknife’. Context might decide between those two, but then once we realize we have the ‘have’ of the first sort, we don’t need context to do further work, we let the compliment (‘a father’, ‘a child’, ‘a reason’) to do the work of picking out the relevant relation.
Hi John,
Ok – now I think your case is much stronger. I wonder weather abstractionists could respond as follows (just playing devil’s advocate here):
First, ordinary language users do not use ‘proposition’ to talk about propositions and that is one good reason the sentences you are focusing on sound odd.
Second, instead of using ‘proposition’ ordinary language users often use ‘fact that’. For example, someone might say that their belief that 4 is not prime is based on the fact that 1, 2, and 4 are its factors, and thereby express the claim that their belief is based on the proposition that 1, 2, and 4 are its factors.
Third, if the second point is granted, then the abstractionist could point to a context in which I am not sure about your reason – was it the fact that you were scared or the fact that he was angry. I might pass a slip asking “What reason did you have for doing it?” and I could pass back “It was the fact that he was angry”.
Jussi,
My view is that reasons are mental states, so I’d be glad if the argument were compatible only with that view. (As it turns out, the argument is compatible with another view, a different one I’ve been working on, but which doesn’t appear in the literature yet. If you want to know more, shoot me an email.)
2** is false. There is of course the despicable institution of human slavery. A master has slaves, possesses slaves, owns slaves. “I have four people I’d like to sell at the auction” is intelligible, even when fathers are salient, etc.
Maybe we sometimes also use “the thought that” or “my thought that”?
So again: I might pass a slip asking “What reason did you have for doing it?” and I could pass back “My thought that he was angry”.
Brad,
I’m glad to hear I’m making progress!
I agree that ‘proposition’ talk is uncommon (not unheard of, though). I should switch, to give abstractionists a fairer shot. Instead of ‘fact’, would the equally natural ‘claim’ work?
‘claim’ might be preferable, because the other major view about the nature of reasons (aside from the view that they’re propositions, and the view that they’re mental states) is the view that they’re facts. If I put ‘fact’ talk in the abstractionist’s mouth here, it might hamstring them at another point in the dialectic.
On your third point, is this a case where you think it would at some point make sense to say, “I have the claim/fact that he was angry”? I’m still having trouble imagining that.
Hmm, I’m a bit suspicious of Mark’s account of ‘has’ as well. But why should I think that (2) is true in the first place? True, talk of `having propositions’ does sound a bit weird. But perhaps that’s simply because ‘has’ has a familiar, fixed meaning in the context of claims about reasons and mental states (which may or may not be the same), but not in the case of propositions. (No great surprise this, given that `proposition’ is a throughly technical term.) If so, there need not be any great obstacle to extending our usage of ‘has’ so that it is appropriate to speak of having propositions whenever one has the corresponding reasons (if this is what one’s theory of reasons requires).
Also, your argument seems to assume that we can substitute co-referential terms into the context of claims about what ‘has’ what. But if ‘has’ means something like ’stands in salient relation R to’, this won’t always be the case (depending on what R is). Doesn’t that provide one with a further out here?
Mark,
Suppose all the ambiguity has been done away with, and we let the complement ‘reason’ help set the relevant relation picked out by ‘have’. Once the reason relation gets loaded into ‘have’ this way, why does it sound unintelligible to say, ‘I have the proposition that Sarah published a book’? Why don’t we hear this as picking out the reason relation?
Am I wrong to think that this should be intelligible, if the proposal were true? Am I wrong to think that we should expect to hear it that way, if the proposal were true?
Brad,
I’d say ‘my thought that he was angry’ is more apt to pick out the belief that he was angry, rather than the proposition [he was angry]. If someone says ‘I have the thought that he’s angry’, we understand her to mean that she believes he’s angry.
John,
I think that the proposal would be that ‘have’ in ‘have a father’ has done all of its work when together with ‘father’ it predicates standing in the child/father relation to someone. Similarly then for having a reason.
If someone says to you, “I have something, but you don’t know what it is,” you might answer with “What?” Now they say, “a father.” I suspect you can complain that you have been tricked into thinking they were making a different sort of claim than you thought they were. But you can’t complain that in the sense used to answer the question their answer is still false unless they have their father right there ready to hand preferably confined or on a leash.
Thanks to everyone for their thoughts. I think I just figured out a way of framing a rejection of 2, which I’m comfortable with. Perhaps someone will tell me that I shouldn’t be!
******************************
We might say, “I have, as my father, James.” We also might say, “I have, as my reason, the proposition that Sarah published a book.” These sound perfectly natural, I think.
Once I say, “I have, as my father, James,” I could then ask you, “What about you?” You could then respond, “I have Charles.” This sounds fine. (I don’t know your father’s name. ‘James’ isn’t my father’s name. Let’s just pretend.)
Once I say, “I have, as my reason, the proposition that Sarah published a book,” I could then ask you, “What about you?” You could then respond, “I have the proposition that Sarah published an article.” Do you think this sounds okay?
If that sounds okay, then I think the abstractionist can comfortably deny 2. If 2 means, “You can’t have, as reasons, propositions,” it just begs the question against the view that reasons are propositions. And if it means something else, it’s of dubious relevance.
So now I think Mark S. is effectively right (not about what reasons are, but about 2 in the argument). Once the ‘have, as an F’ locution occurred to me (and I noticed how natural it can be to elide the ‘as an F’), it all seemed to fall into place — one of those “Now I’ve got it” moments. Talk about “standing in the reason relation” isn’t so natural, and that might be what threw me off.
Am I letting them off the hook too easily? Or does this sound basically right?
I’m coming to the party late, but (1) sounds fine and (2) sounds horrible (to my ear):
(1) I have, as my father, James.
(2) I have, as my reason, the proposition that Sarah published a book.
Maybe it is just because the language is stilted. Events are pretty good candidates for causes and it sounds bad to say stuff like:
(3) It had, as its cause, the event that was Charles abdicating the throne.
Better:
(4) It had, as its cause, Charles’ abdication of the throne.
Maybe what the abstractionist should say is that (5) is true:
(5) I have, as my reason, that Sarah published a book.
Maybe they should then say that (5) isn’t true because of some fact/state of affairs in virtue of which it is true that Sarah published a book. Maybe that’s what arguments from error are supposed to show (not that I think that works. Like you (at some time slices) I don’t think reasons are propositions whether the reasons we’re talking about are normative or motivating. They are either all facts or they ought to divided up between facts and states of mind.)
John,
thanks. About this:
“2** is false. There is of course the despicable institution of human slavery. A master has slaves, possesses slaves, owns slaves. “I have four people I’d like to sell at the auction” is intelligible, even when fathers are salient, etc.”
My intuition is that this sentence is false and 2** true. Really, humans cannot be owned. True, there is an institution in which people think they own others but they are mistaken. They are playing a game in a way in which it might seem like it’s ok to say these things.
John,
sorry about leaving another comment (and sorry about these not getting through to the blog either). Anyway, I’m really intrigued by your great argument. I started to think of a following variation of it. Consider the following claim:
(*) One reason can be stronger than another.
This seems to me to be obviously true. Now, the main options there are, are that reasons are either propositions, facts, or beliefs. Test then these things with (*). You get:
(i) One proposition can be stronger than another.
(ii) One fact can be stronger than another.
(iii) One belief can be stronger than another.
It’s not obvious that any one of these makes sense. What are relative strengths of propositions, facts, or beliefs? (Maybe strength of belief could be something like certitude, but I don’t think that the claim that ‘one is more certain that p than q’ captures (*).
So, my version of the argument would be:
1. Reasons have strengths.
2. Propositions, facts, and beliefs do not have strengths.
3. Therefore, reasons are not propositions, facts, or beliefs.
Now, this seems like a valid argument. And, 1. and 2. seem true, so it seems sound as well. But, that cannot be right. So, do you think there is a difference between this argument and yours? Or, would you endorse this conclusion too?
My diagnosis is that we are equivocating with ‘reasons’. The term can either refer to the reason-relation or to the things that are reasons. The relations have strengths whereas the things related do not. That’s why 1 and 2 can be true but 3 false. The same seems to go for your argument.
Hi John,
We agree, then, on the nitpick and the dialectical set-up. We also agree, of course, on your follow-up nitpick – sorry for the misreading. I’m less sure whether your restriction to “a relata in the basing relation” captures what the abstractionists take themselves be in competition about.
Re: ‘I have, as my reason, (the proposition) that q.’
This response might actually be problematic for the abstractionist. For ‘I have, as my reason, (the proposition) that q and I don’t believe (the proposition) that q’ has a Moore paradoxical air.
If it is Moore paradoxical, it may be that the proposition can serve as a reason only if it is believed (or stands in some relevantly similar doxastic relation like presupposing). Just a thought.
Hey Mikkel,
“If it is Moore paradoxical, it may be that the proposition can serve as a reason only if it is believed (or stands in some relevantly similar doxastic relation like presupposing). Just a thought.”
I think that’s right, but I know some abstractionists will want to say that the mental states are enabling conditions without which the reason wouldn’t be had or in your possession. (I worry that they are helping themselves to something they have no right to in making this move because you might think that the difference between there being a reason to A and your having that reason to A is closed if you know the thing that is the reason exists/obtains. But, if you also know that it’s a necessary existent, it’s hard to see how the having/possessing could depend upon the right sort of contingencies (i.e., the contingent facts that determine whether p is true.))
Here’s a thought, along the lines of what John suggests above (comment 18).
There are different ways for one thing to ‘have’ another. One person may have another person as a father, or as a boss, or as an enemy; one number may have another as a successor, or as a square root, or as a divisor; and so on.
To ensure that John’s original argument doesn’t equivocate in its use of ‘have’, we can rewrite it to make explicit what sort of having is involved:
1. You can have reasons as reasons.
2. You can’t have propositions as reasons.
3. So reasons aren’t propositions.
But 2 now seems to beg the question.