I’ve been recently reading Gibbard’s and Schroeder’s new books (both brilliant) on expressivism. I’ve started to wonder whether there is a connection between expressivism and the so-called buck-passing views. I’m sure others have thought about this before, but, if there is such a connection, it seems quite surprising to me. So, I thought I’d quickly lay out why I think there is this connection and then give some options for what should be done as a result.
In Reconciling Our Aims, Gibbard says that to think of some act as wrong is to plan to have moral attitudes of resentment and blame towards it. In Being For, Schroeder argues that, in order to solve the negation problem and to provide a complete semantics for our normative language, the expressivist should say that to think of some act as wrong is to be for (this is to have a positive attitude towards) blaming people for doing the act.
Now, I was thinking about the claims (and thoughts) of the sort:
(i) I blame Ann for breaking the promise because it is wrong to break a promise [and]
(ii) My reason for blaming Ann for breaking the promise is that to do so is wrong.
I’m not sure how Gibbard or Schroeder’s expressivist would account for these claims (Schroeder mentions something like this issue in the end under section embedding under normative predicates). It seems to me that they would have to account for these claims by claiming that I’m saying and thinking here that I blame Ann for breaking the promise because I plan to blame breaking promises and that my reason for blaming Ann for breaking the promise is that I am planning to blame people for breaking promises. And, at least for me, these claims and thougths do not make a lot of sense.
You might think that this is not a problem for expressivism in general but rather only for the specific proposals that connect the term wrong to blaming. But, I do think the worry generalises to a wider class of views. This are the ones according to which moral claims of the form ‘x is F’ express a plan or being for some attitude that is related to the term F. In all such cases, I presume, it makes sense to say that I have the relevant attitude towards x because x is F (or that x is F is my reason for having the attitude).
Now, here are some alternatives to how we should react to this issue:
1) There is a reading for (i) and (ii) under Gibbard’s and Schroeder’s models that makes good sense of them. I would be interested to hear what they would be like.
2) This is just a problem for expressivist views that account for moral claims in terms of having plans or being for some attitudes. There are other forms of expressivism that avoid this problem.
3) Despite appearances, (i) and (ii) do not make sense. It does not make sense that the wrongness of an act gives reasons for blame. If there are reasons for blame, they are provided by the more basic, wrong-making features of the act. As a buck-passer, I am generally symphatetic to this reply. But, I wonder whether buck-passing view should be the result of semantic commitments of expressivism (although arguing for buck-passing from expressivism would be interesting).
4) (i) and (ii) make sense but expressivism fails to make sense of them and thereby this is one downside of expressivism.
I’m quite uncertain about what to think here, but I would be interested in hearing what you others make of this.
Hi Jussi,
Nice post!
First thoughts on how to pursue option #1:
Couldn’t we take (i) and (ii) as communicating (roughly) the fact that our plan to blame her for the token act is justified because it is one part of a more general plan to blame all acts of a certain type? We might use those sentences to communicate the fact, for example, that we are blaming her for what she did because of the type of thing she did, not just because *she* did it.
In a more reason-giving vien, I might go on to appeal to rule-utilitarian motivations in order to explicate why the plan to blame the type is justified in a way that the plan to blame the token is not so obviously justified — spelling that out might take some doing, but it is not obviously impossible, I do not think.
Does that help?
If so, you might think that this is a better statement for you to worry about:
(iii) I blame promise breakers because promise-breaking is wrong.
To deal with cases like that, couldn’t the expressivist point to the fact that (roughly) to think acts of type X wrong is to plan to blame people who perform acts of type X, but it *also* involves planning to react in certain ways to others’ reactions to acts of that type. For example to disapprove of people who approve of acts of that type, and to approve of those who discourage acts of that type. I am thinking, e.g., of Blackburn in “Ruling Passions”.
If that is right, then (iii) could still have substance.
What do you think?
Hi Brad,
thanks. Interesting comment. I had thought about the type-token thing and I’m a bit sceptical even if I cannot put my finger on the issue. In Gibbardian, to say that ‘promise-breaking is wrong’ is roughly to express a contingency plan to blame the agent in every token situation of promise-breaking. I’m not sure I see how it could justify to blame a person for the token-promise breaking that I already plan to blame her in this situation and others in each of the other tokens of similar cases. That seems like a wrong kind of a reason.
But, of course, you are right that (iii) would remain problematic even if that works out. Thanks for it – it is a better way of putting the idea. Now, I wonder if we can push similar line to your Blackburn move. We could step to
(iv) I blame promise-breakers, disapprove of people who approve of these acts, approve of people who discourage acts of this type (and so on) because promise breaking is wrong.
And, it seems like we can use (iv) to generate the same problem.
Hi Jussi,
I agree with your last move, but I think that the further you push the sentence (towards something like iv) it becomes more plausible to take your option 3.
On the first issue, I do not think they need to say that sentences like (i) are usually used to express a view like this: the fact that you plan to blame all promise-breakings is itself a reason to blame this token (although might not a norm of structural practical rationality get them that?).
Is there some reason they cannot say that uses of sentences like your (i) usually express some view like this: My reason for planning to blame this token is the fact that there are overwhelming reasons to plan to blame all promise breakings.
Brad,
well, maybe it is the case that (i) expresses what you say according to your last proposal. But – that seems to be taking the option 3) instead of 1). The proposal seems to be that wrongness of the act consists of the fact that there are other, non-wrongness-provided reasons to plan to blame all acts of the type to which the token belongs. That this result comes about from considering expressivism seems quite interesting.
Jussi,
I am not sure that much follows. The proposed reading is just about what sentences like the one you mention are typically used to express.
For all that, couldn’t the expressivist also say that wrongness can be a reason (given a background commitment to being structurally rational)?
More generally: My proposal is about what some subset of sentences using the word ‘wrong’ are typically used to express. It need not be about what the “wrongness of the act consists of”.
Brad,
maybe. I phrased what I had in mind badly. Perhaps the thought is that when one says that ‘I blame … because … is wrong’ I typically express the thought that there are reasons to blame for … That still sounds like a buck-passing view of what a certain subset of wrongness claims typically express.
Of course this moved would not be required if, given rationality based consirations, plan to blame for a set of actions gives reason to blame for a token of that set. I still sceptical about this but it is a thought.
Hi, Jussi.
Thanks for the interesting question. I should clarify that nothing in Being For turns on the talk about blame, or even any other attitude. The main idea is a structural one, and I talk about blame in order to give an example to show what a view with that structure might be like. And I chose ‘blame’ in particular, simply because that was Gibbard’s view in Wise Choices, Apt Feelings, so since it was familiar, I could get away with saying less about it. So I don’t think the framework is essentially connected to fitting attitudes ideas, although they are a natural way of trying to implement it.
In general, and this is one of the main lessons that I try to draw in the book, I think that to assess your sentences, which includes expressions like ‘because’ and ‘the reason why… is that…’, we need to know not just the expressivist semantics for its constituent clauses; we also need to know the expressivist semantics for ‘because’ and for ‘the reason why… is that…’.
Now, I haven’t offered an expressivist semantics for those constructions, and I don’t know if there could be a satisfactory one (I’m not, after all, a proponent of expressivism). But I’m skeptical of whether the specific issue that you are raising will pose any deeper or more difficult problems than the general issue about whether normative clauses can figure as the complement of ‘because’. But without a real proposal for a semantics for ‘because’ on the table, I just think it’s hard to assess.
Compare, just for a reminder of how this dialectic works, the case of ‘not’. Ayer and Stevenson and Hare all stated broadly noncognitivist theories of atomic moral sentences of the form, ‘X is wrong’ or ‘X is good’. None of them told us how ‘not’ works (though Hare tried at least a couple times). In fact, there are reasons to think that no satisfactory story _could_ be told about how ‘not’ works. So in the light of unclarity about how ‘not’ worked, on these views, it was easy to generate intuitive false predictions about what sentences involving ‘not’ must mean. Yet as I show in Being For, it turns out that it is at least possible for there to be simple expressivist languages in which ‘not’ works satisfactorily, and these intuitive predictions are false. The reason the intuitive predictions are false, is because they are only based on a guess about what an expressivist semantics for ‘not’ would look like.
So now we have a new construction: ‘because’. We have some intuitive constraints on how it is supposed to work, and one of them – highlighted by your examples – is that it can take normative complements. So any semantics for it had better predict that. I don’t think it is possible to make extra-theoretical judgments about what an expressivist semantics for ‘because’ would have to be committed to, about these cases – I think the only way to evaluate it is to look and see what resources the expressivist has, for giving an account of ‘because’, and see if they are up to the job.
Biforcated Attitude Semantics – the expressivist semantic framework outlined in Being For – gives a recipe, to start out with, for what the semantics might look like, for predicates which take sentential complements which can be either normative or descriptive – this includes attitude reports, for example, as I discuss in chapters 10 and 11, but also deontic modals, as I discuss in the appendix to chapter 12. You would treat them as relations between the things I call ‘semantic values’. If I were going to try to develop an expressivist semantics for ‘because’, that is where I would start. I am not certain I would get anywhere, but if I would not, then the lesson that I would draw is that the problem is deeper and more general than, and upstream from, the specific problem that you are trying to raise.
Hi Mark,
thanks – that’s very helpful.
I agree that this issue is not limited to the issue of blame and wrongness. I thought that on your view, to get enough structure for the attitudes, the expressivist needs some attitude for which the speaker is for when she uses a given predicate. I was wondering whether this problem generalises to all accounts of that type indepedently of what predicate and attitude the expressivist has.
Maybe I had an additional desideratum in mind for expressivist semantics. I agree that the expressivist has to explain how to construct the meaning of the complex phrases from the constituents. But I also thought that the complex attitudes attributed to speakers should be somewhat psychologically plausible. And, I was wondering whether with certain complex claims there is a threat of a problem of an odd kind of self-referentiality of attitudes given the expressivist semantics.
One expressivist account of reason and because claims I’ve heard of (I think Gibbard has an account of this type) is that these claims express being for or planning to use the relevant considerations as salient in one’s planning or in one’s being for. On that model (i) would come roughly close to something like:
For (using being for blaming for promise- breaking as salient in deliberation for blaming Ann for breaking a promise)
or
Planning to use one’s planning to blame promise-breaking as salient in deliberation in forming one’s plan to blame Ann.
Maybe someone could have such plans or being for attitudes but they do seem curious. If one already plans to blame in each case of a type of an act, why plan to use this as salient in forming a plan in one instance? After all, one already plans to blame in that instance.
Hi, Jussi.
Two things: first, just to be clear again, the structural idea in Being For – the solution to the negation problem – does not require attitudes at all. It just requires some relation. For example, ‘stealing is wrong’ could express being for avoiding stealing, where avoiding stealing is not an attitude at all, but a success term for not stealing. That’s just an example (I believe it shows up in chapter 4).
Second point: let’s distinguish intentions or decisions from the states expressed by normative sentences. Gibbard has a model where they are closely related, but they might not be. Now ‘I’m going to do A because P’ might express (or report) a conditional intention whose condition is realized. (I take it that this is something like the model you are suggesting Gibbard endorses.) Even if ‘P’ expresses a positive attitude toward doing A, it might not be an intention at all, and ipso facto not a conditional intention of the required sort. So I take it that’s an answer to your last rhetorical question…
Right. That’s again helpful. I see the first point. So, ‘is wrong’ would be under such a proposal a predicate that is related to the act of *avoiding* such and such instead of *the attitude of blaming* for such and such. I had thought that the attitudes played more of a role but I was mistaken there it seems.
I wonder if the problem still applies so that it doesn’t have much to do with attitudes either. If the predicate ‘wrong’ matches up with the act of avoiding, you’ll get sentences like:
(v) I avoid breaking promises because it is wrong to break a promise.
Perhaps something like this can be said about the second point too. One could think of sentences of the form:
I have attitude Z towards doing A because P.
Here the attitude Z is picked so that it matches what is according to the expressivist the positive attitude towards A expressed by P. And, that seems to get us back to where we started.
Jussi, great post, and an important problem.
It seems like the realist has similar sentences to deal with: “I blame Jane for doing X because X is blameworthy.” This is a little trivial but I think it makes sense. To my mind, this is a way of focusing attention on the fact that you blame Jane for the right kind of reasons and not because someone is paying you to blame Jane, etc.
Now the problem for the expressivist (or buck-passer) is to draw a distinction between right and wrong kinds of reasons that does not appeal to anything normative (on pain of circularity). I have an article forth-coming in which I argue this cannot be done. But that would be the hurdle.
Heath,
thanks. Could you say a bit more why you think that the realist has a similar problem?
I thought she could think that blameworthiness is a distinct property which provides reasons to blame and which in essence is distinct from the blaming attitudes. Of course a buck-passer might think that there is nothing more to blameworthiness than being such as to deserve blame. In that case it would be more difficult to account for the seeming sensibility of your blame but that would be a substantial consequence of buck-passing and not that of any metaethical view.
Jussi,
I take it that the worry is a kind of triviality or circularity in the expressivist’s specification of moral reasons, or reasons for action more generally. So now here’s a similar difficulty for the realist. Consider the claim
(i) There is decisive reason to blame Jane because there is decisive reason to blame Jane.
Obviously this is trivial or circular in the problematic sense. Now define or analyze “is blameworthy” as
(ii) “s is blameworthy” =df “there is decisive reason to blame s”
Now if (ii) is true, then
(iii) There is decisive reason to blame Jane because Jane is blameworthy
is trivial or circular in the problematic sense, and this is so no matter how realist you are about reasons to blame and, hence, blameworthiness.
One might object that (ii) is not a good analysis of blameworthiness. I think it’s pretty good myself—you might want to throw in something about the right kind of reasons, but that won’t fundamentally change the issue—but in any case one will be able to construct realist predicates like this. Or put the other way around, so long as some realist predicates “sum up” the balance of reasons, we will be able to find verbs for which those predicates appear to provide reasons but for which the sentences in which they occur are problematic or semi-problematic. E.g. if “wrong” is understood realistically as “ought to be avoided” then the realist will be able to say things like
(iv) Stealing ought to be avoided because it is wrong
which is true but somewhat unhelpful. Sentences like (iii) and (iv) will be problematic or semi-problematic, even if they do not appear problematic on the surface.
Now, having said all that, it may be that I do not fully understand what problem you have in mind for the expressivist.
Thanks. I see now – this is very helpful. I think this is the crucial step in the deduction you give:
(ii) “s is blameworthy” =df “there is decisive reason to blame s”
This looks to me like a buck-passing account of blameworthiness. And, I might think that it is this that creates the problem and not realism per se. I did agree in the original post that a buck-passer about a problem has to deny the sensibility of the kinds of claims we are talking about.
But, I’m starting to think that, unlike the expressivist, the realist can deny the buck-passing accounts of the relevant properties. Take Scanlon. He claims that wrongness does give reasons not to do the act but that’s because he doesn’t think that wrongness is the balance of reasons but rather that it is the the substantial property of being forbidden by the non-rejectable principles.