There is a powerful argument that is
frequently used in metaethics (and elsewhere). It’s based on the
idea that disagreement requires shared concepts. Thus, by looking at
who we can intuitively disagree with we can determine who we share
our concepts with. And, in when it comes to morality, it seems we can
disagree with people who classify very different things under moral
terms. As a result, it is argued, ethical terms cannot have
‘descriptive content’. I used to think that this great argument
works. Today I’m feeling less certain about this and I’d like to
hear about your thoughts. Here is why.
One of the original formulations of the
argument is from Hare 52 (as quoted in Smith 94) – the classic
cannibal case:
Let us suppose that a missionary, armed
with a grammar book, lands on a cannibal island. The vocabulary of
his grammar book gives him the equivalent, in the cannibals’
language, of the English word ‘good’. Let us suppose that, by a
strange coincidence, the word is ‘good’. And, let us suppose,
also, that it really is equivalent – that it is, as the Oxford
English Dictionary puts it, ‘the most general adjective of
commendation’… If the missionary has mastered his vocabulary, he
can, so long as he uses the word evaluatively and not descriptively,
communicate with them quite happily. They know that when he uses the
word he is commending the person or object that he applies it to. The
only thing they find odd is that he applies it to such unexpected
people, people who are meek and gentle and do not collect large
quantities of scalps; whereas they themselves are accustomed to
commend people who are bold and burly and collect more scalps than
the average.
We thus have a situation which would
appear paradoxical to someone who thought ‘good’ … was a
quality word like ‘red’. Even if the qualities in people which
the missionary commended had nothing in common with the qualities
which the cannibals commended, yet they would both mean what the word
‘good’ meant. If good were like ‘red’, this would be
impossible; for the cannibals’ word and the English word would not
be synonymous… It is because in its primary evaluative meaning
‘good’ means neither of these things, but is in both languages
the most general adjective of commendation, that the missionary can
use it to teach the cannibals Christian morals.
I just love this passage. In any case,
much after Hare’s argument, it seemed like some philosophers wanted
to apply Putnam/Kripke natural-kind semantics to moral terms. Smith,
Horgan and Timmons, and others then applied Hare’s argument against
such attempts. On the new naturalist views, people’s referential
intentions or their use of the term ‘good’ would fix the
reference of ‘good’ to some natural property that would then
constitute goodness is. Yet, the speakers can be quite unaware of
what that property is, and this would explains the intuitions behind
the open question arguments.
Unfortunately, it seems like we can
imagine cases in which the reference gets fixed to slightly different
natural properties in different communities. As a result, the content
of the goodness claims in these communities will be different. We’ll
be talking about what’s good (imagine, H2O) whereas
they’ll be talking about what’s tgood (XYZ). This would mean that
there would be no real disagreements but rather talking past one
another. Yet, it seems intuitive, almost platitudinous, that when A
says ‘x is right’ and B ‘x is not right’ in these cases they
disagree (as Smith says) just as long as their terms play
corresponding practical roles.
There is a naive answer to this
argument. It says that people in the disagreement scenarios only
think that they disagree substantially whereas in fact they really
don’t as they have different concepts. And, it is easy to motivate
why we would act as if we disagreed. This would be a way of getting
the other party to adopt our concept. Given the practical role of the
concepts, this would affect their behaviour. Some conceptual
disagreements really are worth fighting for.
The defenders of the argument have a
response to the naive answer. They say that to claim that there is no
real disagreement would be chauvinistic conceptual relativism.
On this view, part of the argument rests on our conceptual competence
with the moral terms. The awareness of when one has a real
substantial disagreement with others is part of that competence. This
is why it is platitudinous that we would be disagreeing with the
cannibals (as Smith puts it). And, it does seem offensive to be told
that, despite what you honestly think, you aren’t really
disagreeing about real moral issues. This would be to challenge your
competence.
I used to think that this must be
right. But, then I heard Stephen Yablo mention how our intuitions
about whether our disagreements are substantial or conceptual are
more than shaky. If this were right, then knowing when one is in a
substantial disagreement would not necessarily be part of conceptual
competence and the naive answer might be defended. Consider following
kind of cases. Imagine that you go see a film with your teenage son.
He buys a large bucket of popcorn. In this situation, it is easy to
imagine that you could have a disagreement about whether that is a
lot of popcorn (‘that’s a lot of popcorn’, ‘no it isn’t’
sounds natural). In this case, when you think of the situation
philosophically, it is just as easy to think that the disagreement is
merely a conceptual one. You and your son would clearly have
different standards for what is a lot of popcorn and you are both
saying something about the amount of popcorn relative to your
standards. Yet, you can go on disagreeing. So, it does seem that the
real disagreement here is about what is the standard of assessment to
use in buying popcorn.
Thus, in a sense, you are disagreeing
because you do have different concepts and this disagreement seems
conceptual. It would also be hard to think that Hare’s argument
shows in this case that ‘a lot’ does not have descriptive
meaning. On the other hand, it doesn’t seem like when you put
yourself in the situation you would describe yourself as having a
disagreement about what ‘a lot’ means or which concept is the one
use. The idea of making a semantic ascent within that situation
itself seems odd even if that would be what really seems to be going
on from the outside.
So, I’m starting wonder why the
naturalist could not say something similar about the cannibal and
twin-earth cases. The point would just be that it’s not part of our
semantic competence as competent speakers to know whether we would be
having substantial or conceptual disagreements. We don’t really
care or need to care in real situations. If this were right, then the
intuition that we would be having a disagreement
would not be enough to show that we share the same concepts.
Hi Jussi, I think you’re completely right to say that we can’t always know when we’re having a substantive disagreement. But there are a few reasons why relying on the Yablo-style observation isn’t going to help out Cornell realism all that much. First there’s the dialectical point that Cornell realism is meant to avoid the open question argument, and is thus an improvement over the analytical descriptivism that (it was agreed) Hare et al had refuted. If we reject Hare’s argument for the reasons you suggest, this supposed advantage disappears. Of course, that might be fine, and we might just defend the original analytical descriptivist view. Or there might be other reasons for preferring the Cornell realist view (maybe it’s better to minimise the realm of merely conceptual disagreement, and Cornell realism might restrict this to twin earth cases).
The more substantive point is that whether or not our intuitions of substantive disagreement are dispositive, the intuitions are present, and even a successful account of how those intuitions could be confused still leaves us with an error theory. In general error theories are bound to be undesirable. And this error theory looks particularly problematic. The intuition that ethical disagreements are substantive is fairly close to bedrock in metaethics. Before being asked to give it up we would need to know that that intuition could not be accommodated except in an otherwise unattractive theory. So it seems that the main dialectical purpose of the cannibal and twin earth arguments is served even if the substantiveness intuition is defeasible. Those arguments serve to suggest that expressivism (because it respects the intuition) is a more attractive theory than descriptivism. So in order to defend descriptivism one must highlight equal deficiencies in expressivism. A certain burden of proof has been shifted, and it doesn’t get unshifted just by pointing out that the intuition is defeasible.
Another was of putting this is that what the cannibal and twin earth arguments show is that a descriptivist theory will be an unhappy solution to metaethical problems: it will leave us disatisfied in having to admit that our strongly held intuitions were in error. Of course it may be that no happy solution is out there, and an unhappy solution is still acceptable if it is all that’s available; but if we want to find a happy solution we will look away from descriptivism.
Thanks Daniel. I think I agree with you on much of what you say. Still, with my naturalist hat on, I think this won’t yet convince him.
I think here is something the naturalist might say. They might say that what the Yablo observation shows is that what’s really built into our intuition is that we can have disagreements. But, as the popcorn case shows, it’s not part of this intuition whether the disagreement is substantial or conceptual in nature. So, it’s not clear how much of an error theory this is. In addition, they might come to give general explanations that tell us why we have the tendency of making also the conceptual disagreements seem to us like substantial ones (something that must be going on in the popcorn case). For one, the semantic ascent is awfully complicated to make in the heat of the debate and the practical consequences are the same anyway.
Burdens of proof are tricky. I’ve always thought that expressivists have a heavy one. We do have some pre-theoretical intuitions about using moral language and about what kind of moral beliefs we have. I thought the expressivist always had the burden of explaining why these intuitions are mistaken.
Jussi,
What are your thoughts then on the Jackson-Smith type of functional analysis of moral terms? My initial reaction to Yablo’s counterexample is that the conceputal/substantive disagreement distinction is still pretty clear in the case of moral terms, even if it can be doubted with respect to other terms or predicates. The popcorn example suggests that there can be disagreements where it’s not obvious whether there’s a disagreement in concepts or in the referents of these concepts, but the example works only because there’s no accepted functional characterization for terms like ‘a lot of popcorn.’ ‘Good’ is tricky, but perhaps ‘right,’ for instance, can be given a Jackson-Smith type platitudinous analysis. If so, then this analysis constrains discourse on moral concepts in the way the naturalist supposes and it would therefore be possible to classify disagreements as conceptual or substantive.
If you’re saying that it’s not part of our intuition that the disagreement is substantive, then maybe I haven’t quite understood how the popcorn example is meant to work. You say that even if we recognise that the disagreement is merely conceptual (our concepts of “a lot” involve different standards) we can go on disagreeing. We need to know more about this. Clearly there is a practical disagreement: we disagree about how much popcorn to take typically. Can this disagreement be expressed? One might say: your standards for how much popcorn it one should typically take are too gluttonous/frugal. The point here is that the disagreement can be expressed because we have more fundamental vocabulary. But if the descriptivist is right, then our disagreements involving the more fundamental vocabulary will also be conceptual (involving relativity to different standards). E.g. we have different concepts of gluttony. The danger is that the descriptivist is reading this kind of relativity into the most fundamental normative concepts. That will mean (according to descriptivism) that the substantive disagreement is no longer expressible by subjects who are aware of their concepts and restrict themselves to indicative utterances. Of course the substantive disagreement can be expressed by imperatives (“Don’t take so much popcorn!”), which shows that the disagreement is there. It would be odd if our normative vocabulary were so impoverished that we couldn’t express such disagreements indicatively. And it really would be an error theory, in that most speakers take it that the difference between “Phi-ing is wrong” and “Don’t phi” is that the former is stronger than the latter. Yet on the descriptivist view, once we understand the semantics of “wrong”, when someone says the former to me they may in no way be telling me what to do. That’s a pretty serious threat to the normativity of normative language. So in summary I’m surprised if you deny the intuition of substantiveness.
Daniel, this claim: ‘The intuition that ethical disagreements are substantive is fairly close to bedrock in metaethics.’
Is this supposed to be an intuition we have about all cases of ethical disagreement, or just some? If it is an intuition about all cases, then you might be right about the problem it presents for the Cornell realist. But I’m not sure whether I think this claim is plausible.
It’s certainly not as plausible as the claim that people judge that some, or most, of their ethical disagreements are substantive. But this judgement seems to be compatible with the Cornell realist story – they aren’t committed to the claim that all ethical disagreements are about conceptual matters – we might share a common intention to use ‘good’ in a certain way, and just be wrong about its application.
This story would also explain why we think that moral disagreement is usually substantive – we typically discuss ethical matters with people who are using good in the same way as us. Or am I missing something?
Thanks, and thanks for the post Jussi.
Michael,
there’s a new paper in the essay’s on Jackson book where Horgan and Timmons argue that the disagreement argument is fatal for functional analysis of moral terms. That view would, as you say, come to claim that the disagreement is conceptual. And, H & T say that this is implausible.
I was thinking that here would be one way for Jackson to argue that that doesn’t succeed. In fact, given Jackson’s recent arguments for subjectivists being able to make sense of the disagreements, I would not be surprised of this.
I don’t know. It’s true that there aren’t many platitudes about ‘a lot of popcorn’ but surely there are platitudes about ‘a lot’ and ‘popcorn’ and assuming compositionality that seems enough. A lot is more than little and so on. It’s not immediately clear that there are much more platitudes about thin moral terms.
Daniel,
you are right. I should not have put it like that. It’s true that in the popcorn case that the intuition is that the disagreement is first-order – substantive. It is about the lotness of the popcorn and not about words seemingly. But, maybe one could claim that what the intuition does not include is the presupposition that the disagreement is about whether the bucket of popcorn belongs to the extension of our shared term. This just seems like too much built into the disagreement-intuition. Maybe we don’t really think/care about what disagreement actually requires. So, you are right, at some point there is a bit of error theory here.
I was assuming that the descriptivist can give some kind of pragmatist story about the prescriptive force of the utterances. This would explain why, even if they are disagreeing about which one of the terms ‘a lot’ with different extensions we should use, there is a point to disagreeing about this (even if we do this on the first-level of the discourse). This would also explain why we can have the practical disagreement by using indicative sentences (and even when the disagreement is really conceptual).
Kirk, to answer your question, I don’t think that every case of ethical disagreement is substantive. I agree with Hare that sometimes people use normative vocabulary in an inverted commas way, when they are really disagreeing merely about e.g. what the standards of their society are (this disagreement can be called ethical, though it isn’t ethically substantive, and isn’t conceptual either). What I take issue with in Cornell realism is that ethical disagreements get to be non-substantive just because those I am disagreeing with are e.g. from twin earth. Suppose that the twin-earthers are merrily torturing people, and yet given the way that the semantics of “wrong” work (according to Cornell realism), they can say truly that this behaviour is not wrong. I disapprove of the torture on twin earth in the same way I disapprove of any torture. I disagree with the twin earthers about it too: I disapprove of their behaviour, they approve of it. And yet according to Cornell realism this disagreement is not expressed when I say “Torture is wrong” and they say “Torture is not wrong”. My intuition is that the substantivity of expressed disagreement goes along with the substantivity of actual disagreement as seen in the underlying attitudes.
Jussi,
This is basically the response to the argument from disagreement that Gunnar Bjornsson and I give in our paper on metaethical contextualism. We argue that ordinary intuitions of disagreement aren’t fine-grained enough to reliably track the difference between disagreement over a proposition and disagreement in attitude. (To Daniel’s point, I’ll observe that an expressivist view of the disagreement is compatible with a descriptivist view of the contents. So if a particular descriptivist view supports an expressivist disagreement, then this is a descriptivist theory that answers your objection).
Sorry, I see that Jussi made effectively the same response to Daniel before I posted.
Thanks Steve. I’ve probably heard you explain this but didn’t think about the implications enough in the context. Sorry about this – should have remembered.
No worries, Jussi–I hope my post didn’t sound pissy. We only make the point in passing, and not with the care that you do here, but I wanted to draw people’s attention to the fact that at least some metaethical work is already appealing to this idea. I’ll be interested to see what the community has to say.
I have a few questions for both Steve and Jussi. What do you conclude from your observations? I hear Steve’s skepticism about determining whether our intuitions are tracking substantive or conceptual/semantic disagreement as general. Is that right? If so, how do you see that skepticism as compatible with Putnam’s Twin Earth thought experiment? It seems to rest on an intuition that our disagreement is semantic. Or are they not compatible? If so, does this mean you reject Putman’s thought experiment as well? If they are compatible, does that mean that you think it is only our intuitions about whether our disagreements are substantive or conceptual/semantic in the moral case are suspect? If so, do you think this is a problem with the particular thought experiments that have been devised (e.g. Hare’s) or can there be no carefully crafted thought experiment that could target a disagreement of one kind over the other, as many assume Putman’s Twin Earth experiment does? If so, can you explain why this is so? What is it about apparent moral disagreement that warrants such strong skepticism, when a parallel skepticism in other areas isn’t?
Janice,
I was wondering about whether this would cast doubt on the idea that the possibility of disagreement rules out descriptivism. It’s true that the sceptisism about our intuitions is general but it’s not like the conclusion can be that all our disagreements are conceptual/semantic. I take it that the point is that our intuitions about this aren’t decisive.
I think this should be compatible with Putnam’s Twin Earth. I thought the original point of H & T was that the natural kind semantics cannot work for moral terms because we have very different intuitions about the twin earth and moral twin earth cases. In the water twin earth case, if we knew the constitution of the stuff the twin earthlings call water we would not disagree with them about whether there is water in the class (of XYZ). We would know that they are talking about different substance. We might talk about water with them but in this case we would adopt a referential intention to talk about whatever they are talking about that would change our speaker’s meaning. So, in those cases, intuitively there is no disagreement about which we could ask whether it is a substantial or conceptual. And, this is supposed to support the Putnam hypothesis.
The moral twin earth cases is different. Here we do have an intuition that we would disagree with the moral twin earthlings even if they systematically called torture good and suffering causally guided their use of the term (just as long as the term has the same practical role). So, this is why the question about disagreement and shared concept rises here and not in the Putnam case.
Janice,
Good question. Answering for myself, I’d want to distinguish between mere semantic disagreement and different kinds of (let’s call it) attitudinal disagreement. In Putnam’s cases, there’s only ‘disagreement’ over what word to use. I believe we can reliably distinguish those cases from cases of attitudinal disagreement, so Twin Earth arguments are fine.
What Gunnar and I are skeptical about is the ability of ordinary intuitions to reliably differentiate between cases where two assertions express conflicting beliefs, and when they express conflicting pro-attitudes. That’s the question in metaethical cases, and I take it that it’s also what’s at stake in ‘that’s a lot of popcorn’: there’s a conflict in attitudes towards (something like) how much popcorn is an appropriate amount.
But Jussi did talk about ‘semantic’ disagreement, so maybe he intended a stronger claim?
Well, I’m not sure how much stronger I meant (did I even use semantic – I know I did call it conceptual?). However, it seems to me that if the attitudes in question are the same for both speakers and we are assuming that they are related to terms that have descriptive content, then the disagreement in attitude requires arguing with terms that have different extensions. You might think that the disagreement is merely expressive in nature but I would want to say that you are also trying to get the other party to have the given attitudes towards the same things. In effect, in the descriptivist framework this would require that you get them to adopt the term with the same extension as yours. So, in effect, it seems to me that there is also a disagreement about which term to use together or of trying to get the other party to see that they’ve misunderstood to which things their term applies. This seems conceptual to me in a way. Or, so the naturalist might argue.
Sorry to come to this so late, but I’m a bit confused about the popcorn example, specifically how it serves as a counterexample to the argument in question. When my father and I disagree over what is ‘a lot’ of popcorn, it seems (as Steve mentions) that we are disagreeing about something like what is an appropriate amount of popcorn. But this is a substantive, not a “merely conceptual” disagreement about what is an appropriate amount, a disagreement that then informs the disagreement concerning what is ‘a lot’. ‘A lot’ certainly has descriptive content, something like ‘(significantly) more than appropriate’. But as with ‘good’, it seems to me that ‘appropriate’ has no descriptive content. It may have limiting content—neither one kernel nor an amount it is impossible to consume can turn out to be ‘appropriate’. But it seems that father and son are having a substantive normative disagreement concerning what is an appropriate amount of popcorn, just as I may disagree with the natives about what is good. So why isn’t the popcorn example just more fodder for the disagreement argument, rather than a counterexample to it?
No worries – thanks for taking part. Here’s what I had in mind.
First, it’s not clear that there is any normativity involved. The father and the son can be happy amount of popcorn the son is eating. They agree that it is an appropriate amount and the amount the son ought to eat. They can yet disagree whether the amount in question (that is appropriate) is a lot of popcorn or not. Nothing rules out that it is appropriate to eat a lot of popcorn.
Second, the reason why I think we should not think that they are having a mere substantial disagreement is that it would be quite an uncharitable way of understanding their behaviour. On that interpretation of their discourse, we would have to attribute an obvious error in application of their shared concept to one of them. One of them would be getting things wrong by their own lights. But it shouldn’t be that hard to measure whether the amount in question matches up to what you take to be a lot of popcorn. So, given that charity is a constraint of making sense of others, we should avoid this option.
A fairer understanding of what is going on thus seems to be that they have different concepts of a lotness, they correctly apply those concepts, and disagree about which is to be used in the context. They could express their disagreement on the higher semantic level, but why would they bother? Fighting over the words in the first-order talk seems to be just as good. This strikes me true about the observation; we are not always that bothered about whether our disagreements are substantial or over the words and there might not even be that strict of a distinction.
David and Jussi,
I suggest that Jussi’s example might not be the best for his point. Consider the following ‘disagreement’, between a teenager (A) and an adult (B)
A: ‘She’s really old’ pointing to a 40-year old woman.
B: ‘No, she’s not old!’
Presumably both realize that use of ‘old’ is standard-relative, and that each is employing a different standard. The ‘disagreement’ is over something like what standard to use. I agree with Jussi, except that I wouldn’t want to say that there are different ‘concepts’ involved.
Mark Richard has a great paper on this, ‘Contextualism and Relativism’.