Experimental Metaethics

Steve Campbell from Michigan kindly directed my attention to a new paper that uses the methods of experimental philosophy to investigate the objectivity-question in metaethics. This paper by Geoffrey Goodwin and John Darley is entitled ‘The Psychology of Metaethics: Exploring Objectivism’ (downloadable from HERE). I have a couple of questions below but I think I need to explain the paper a bit first.

The method of the paper was to give a set of claims to the ‘subjects’. These were different factual claims, moral claims, claims about conventions and of matters of taste. The subjects were asked first whether they agreed with the claims and then whether they were true, false, or ‘mere opinions’. They were finally told that many other people hold the opposite views about the given claim. The final follow-up question was whether these people were mistaken, whether the subject himself could be mistaken, whether both could be right, or something ‘other’ was the case (whatever that could be).

The idea is supposed to be that agreeing to the possibility of mistakenness reveals that subject thinks that there is a disagreement. Thinking that there is a disagreement is then taken to be evidence that the subject thinks that the truth in question is objective and judgment-independent.

The answers were surprisingly predictable and the conclusions the authors drew were pretty modest. Perhaps the only slightly curious thing is that people think that some moral claims are true/false whereas others are mere opinions. But, by and large, on the criteria of the authors vast majority of the subjects thought that morality is almost as objective as factual matters of science.

I’m quite sceptical about this sort of methodology. First, there are questions about the sample of Princeton undergraduates who may not be representative in these matters (intercultural and class variation would be interesting). Second (as Antti Kauppinen has argued) there are questions about whether polling gives us anything more than gut-reactions whereas what would be more interesting would be some sort of reflective self-understanding of the subjects. Finally, I’m not sure how good of a meter the possibility of being mistaken is for the acceptance of the sort of judgment-independent objectivity the authors are after. Many response-dependent domains seem to contain the idea that one party is mistaken.

Anyway, I am (and Steve is) interested in whether there could be any arguments from what folk-metaethics is like to what the right metaethical view is. It’s true that we have some (new?) data here. Many metaethicists refer to the realist phenomenology and to the presumption of objectivity it entails. The study seems to support this line of thought. But I take it that any non-cognitivist worth their salt has an explanation for why it seems to us that morality is objective. Could the realist/objectivist get more out of this sort of studies to support their views? Is there any (even ideal) experimental data that could have more weight?

29 Replies to “Experimental Metaethics

  1. Jussi,
    Interesting stuff. I’m not sure if I follow you exactly, so let me just draw a distinction. You could look at non-cognitivists as having two agendas (among others, of course): to offer an explanation that debunks the claim that we ordinarily take morality to be objective and to offer a story debunking the objectivity we (purportedly) attribute to morality. I could very well be mistaken about this, but as a sociological observation, it seems to me that only in a few places is the agenda to specifically argue against the claim that folk morality purports to be objective (esp. with regards to the phenomenology — I’m thinking especially of Blackburn and Horgan and Timmons, assuming that you’d classify these folks as non-cognitivists despite their own labels). In any case, given that some have taken up the task of showing that morality does not even purport to be objective, then showing that it does purport to be objective would be an interesting result. It would also be interesting if non-cognitivists could debunk the apparent phenomenological aspects of the objectivity, but not its non-phenomenological aspects, such as those we might find in experimental results.

  2. Josh,
    sorry about not being very clear. I was thinking that non-cognitivists would only be taking part in the second of your projects (as far as I understand them). I think you are right about Blackburn and others who I really did have in mind. Come to think of it there are some people (Loeb perhaps) whose views migth be in conflict with the apparent aim of objective truth of the folk thougths. But I wonder if they would question the method used. Give few minutes of Socratic questioning and they might be able to also reveal less objectivist folk intuitions.

  3. Hi Jussi,
    Interesting results. I’m skeptical too. Obviously, no ad populum inference will do. But there might be an inference to the best explanation in the offing (or as some like to call it, affirming the consequent).
    It would be nice if we could get social psychologists even more interested in the things philosophers are interested in. It would be interesting to find out what we ordinarily think we’re doing when we talk moral talk. I fear it would be all over the map. But one point. I’ve long thought that it might be the case that knee-jerk undergraduate anti-realist or relativist rhetoric actually reflects deeper realist, absolutist convictions. Fear of finding that you can’t defend what you believe when you are very closely identified with those belief could lead to an attempt to avoid the need to do so, and so to “you believe what you want to, I’ll believe what I want to” and so on. So whatever instrument one comes up with to measure folk views would have to do something to take account of the fact that entrenched practices of dissembling or being misleading about one’s metaethics could be at work.

  4. While I definitely agree that we cannot take surface responses to questionnaires as decisive evidence, I think we should also be careful about overstating the worries on this front. This probably isn’t the place for a referendum on experimental philosophy, but generally I’m a fan. I think that data like the ones you report, Jussi, (and many, many others) are at least suggestive evidence (so long as they are gathered in a manner consistent with the norms and practices of the social sciences). That doesn’t mean that other data, such as those that might come out of more extensive, “Socratic” questioning might be evidence, as well. At the end of the day, I suspect we’ll want to look at all of these data, and then render our analyses. That, at least, would seem to be more reliable than a few of us sitting in our armchairs and letting our “data of one” stand in as representative for the folk. (Especially given our corrupted intuitions, etc. etc.) I mean, once we open the door to saying things like, “Folk morality is objective” or “Folk morality only appears objective, but it really isn’t,” it seems to me that we should look at folk moral discourse. And, the best look at folk discourse will be informed by the norms and practices of the social sciences, presumably.

  5. Whoops, I left out a negation (never a good idea). This:
    “That doesn’t mean that other data, such as those that might come out of more extensive, “Socratic” questioning might be evidence, as well.”
    …should be this:
    “That doesn’t mean that other data, such as those that might come out of more extensive, “Socratic” questioning might not be evidence, as well.”

  6. Robert,
    that’s really interesting. I did think when I first read the paper that my experience of many undergraduates is the same as yours – that they are relativists and so on. I do like your explanation for whey they say they are. This would make a nice psychological study. I think there is some peer-mechanism going on too – nihilism seems cool…
    There’s couple of questions here. Of course there are some good scientific means to find out what people think. But I’m not sure the questions philosophers ask are about that sort of empirical issues. So, I’m often sceptical about whether the experimental method can provide better responses to the same questions. I do agree about abiding to the norms and practices of social sciences. Yet it seems to me that social scienticists are often very careful in their methodology – in the kind of samples they use, in the alternative ways of collecting the data for instance from the statistics instead of asking people and so on.

  7. Jussi,
    You write, “Of course there are some good scientific means to find out what people think. But I’m not sure the questions philosophers ask are about that sort of empirical issues.” But surely what the folk think is relevant to some philosophical questions. Philosophers frequently make claims about the content of the ordinary concept of RIGHT or FREE WILL or RACE or whatever, and at least part of that content is settled by how we use the corresponding terms (very loosely, how we think about these things).
    Regarding social scientific methods, I think that (almost?) everyone in the x-phi community would agree that x-phi should comport with the best practices of the social sciences, when it takes on social scientific practices. (That doesn’t mean that we’re there yet, of course.) So I don’t think that worries about some aspects of some study’s methods mean that we should discount that study out of hand. For instance, if you worry that some study’s results might be tainted by some bias because of the sample (say, all of the respondents are from Princeton), the right response — as social scientists themselves would surely insist, and as I thought you were suggesting in your original post — is not to discount those results or this kind of endeavor out of hand, but to do some similar studies at University of Nebraska and Cal State University East Bay (or wherever the suspected bias might be revealed) to see if such a bias exists. Then we’ll correct our analyses if we find such biases.

  8. Josh,
    I agree with the first point in moderation – *part* of the content of ordinary concepts is settled by facts about the use of the terms. It’s also true that the actual is relevant for the philosophical questions. But, there are still big issues. The terms you mention are good examples. It seems that there have been times and places where ‘folk’ have grossly misused these terms, even the vast majority of them. Are ‘we’ incapable of similar errors? If we are not, then there is no inference from actual use to correct use that is constitutive of the content of these concepts.
    I know the methods of X-Phi is catching up quickly with other social sciences. Much more comparative studies are done and interviewing beyond polling. I’m very happy to see this. But then there seems to be some studies that aren’t really putting in the effort.

  9. I agree, Jussi: we can mistakenly use terms and we can mistakenly characterize them. (I guess I’m echoing what I said during the discussion on Doug’s last post here.) But, of course, the only reason we can know that we are mistaken is that the mistakes are incompatible with some other uses and characterizations of our terms (plus the way the world is). The only reason we were mistaken to say that atoms are indivisible is that we were also saying that some specific things were atoms, and those things turned out to be divisible. So, it seems like the best thing to say is that it’s the analyst’s job to sort out our sometimes conflicting data, but also that it’s good to get as many data as possible from the folk when folk usage is relevant.

  10. “Many metaethicists refer to the realist phenomenology and to the presumption of objectivity it entails. The study seems to support this line of thought. But I take it that any non-cognitivist worth their salt has an explanation for why it seems to us that morality is objective.”
    I suppose that’s a line of response some noncognitvists will be tempted to take. But another one runs more like this: philosophers are in the business of finding neat taxonomies and categorizing the world as they find rational (among other things). In this work they may come up with neat concepts such as “true” and “false” in the descriptive senses. But then they might get so caught up in this exciting work that they fall for the mistake of assuming that just because they have this really rational and philosophically convenient taxonomy, that’s going to be the one the unreflective human mind, or language in general uses. So the philosopher might be tempted to ascribe the use of “true” and “false” to a belief in a mind-independent, descriptive proposition. But doesn’t this beg the question against the noncognitivist? Isn’t it exactly the noncognitivist’s thesis that when we use words such as “true” and “false” in an ethical context, we are not purporting to describe reality but in fact expressing our sentiments, or something like it?
    This way we can flip the coin. Isn’t the cognitivist doing just what he’s chastising the non-cognitivist for doing: injecting his own armchair analyses into real-world data? While I suppose having the data is an improvement over not having it, I fail to see how it can enlighten the debate about the ontological and metaphysical commitments of folk morality, since to make sense of the data, you need to assume either of the positions you are trying to decide between.

  11. Well, I’m not sure philosophers have come with concepts of ‘true’ and ‘false’ – they seem to be a part of the conceptual framework that we find ourselves immersed in. Also, issues about begging the question are notoriously difficult. You are right about the expressivist’s understanding of truth though.
    I’m not sure the data is completely neutral. One question in the study was whether the subjects think that one party is mistaken if they disagree. Indeed many people thought that this is the case (less so in the convention and taste cases). A realist has an easier explanation of why people say this – they are trying to fit their views to the moral reality.
    An expressivist can interpret the claim about mistake in a deflationist way as a claim that someone who had the attitudes the subject applauds would disapprove the attitudes of one side on the issue (or something like that). I think the data has done something if there is a difference in how good these explanations are. I’m not sure that either one of them is much better than the other but in any case this is a question we can assess independently of assuming either one of the views.

  12. No, I didn’t say (or didn’t mean to say) that philosophers invented the concepts of true or false. They do, however, provide various analyses of these concepts, which may or may not correspond to the way we actually use them.

  13. As for assessing which explanation is the best: the results are exactly what you’d expect on both accounts. So you’d need an *independent* reason to suggest that either is a better explanation than the other. Which, as far as I can see, makes the data neutral.

  14. Simen,
    I think the data does provide some support for a cognitivist understanding of what (e.g.) (M) “‘murder is wrong’ is true” means, insofar as it rules out a certain (arguably influential) rationale for noncognitivism.
    According to Simon Blackburn “‘murder is wrong’ is true” means the same thing as “murder is wrong,” and, as a general rule,
    (Tmod)* “‘s’ is true” expresses “s.”
    Where “s” (and this, btw, is where Blackburn deflationism account goes awry) stands for any statement (not merely any proposition).
    From (Tmod) Blackburn is able to argue that the fact that people sometimes say “it is true that murder is wrong” does not evidence that they are realists / think “murder is wrong” is objective; (Tmod) entails that, say, someone who says “it is true that boo to Monet paintings” expresses “boo for Monet paintings” — plainly not an objective remark. Mutatis mutandis, “it is true that s” merely expresses ‘s,’ someone who endorses it is not committed to taking ‘s’ as (e.g.) a possible object of belief.
    The study casts doubt on Blackburn’s deflationist argument for noncognitivism by indicating that people don’t take “‘s’ is true” to express ‘s.’ (For, if they did, they would be equally likely to predicate “is true” (and, mutatis mutandis, “is false”) of aesthetic statements as of moral statements or factual statements.)
    No doubt there are other noncognitivist arguments that are compatible with the data — maybe “‘s’ is true” expresses an especially an especially passionate pro-attitude (e.g. person a: “Risk is such an awesome board game,” person b: “that’s so true!”), but inasmuch as it debunks Blackburn-style deflationism, the data lends some support to cognitivism.
    *Note: (Tmod) is not exactly the deflationist principle Blackburn endorses in Ruling Passions, but I think it is a fair characterization of what he’s going for.

  15. Jamie,
    I think the data show that ordinary English speakers (assuming Princeton undergraduates are a representative sample of those) do not always take “S’ and “‘S’ is true” to be materially equivalent. This is because I’d expect, if ordinary language speakers did affirm this equivalence, that they’d have no problem saying, of any S with which they agreed, that “‘S’ is true” and not a “mere opinion.” In point of fact, they are willing to say of factual claims with which they agree that “‘F’ is true,” but are not (nearly as) willing to say that of aesthetic claims with which they agree. If “‘S’ is true” means “S,” why would they be unwilling to endorse “‘S’ is true” when S is instantiated for a certain class of statements with which they agree, but not when it is instantiated for another?
    Just to be clear: I do think that “S” and “‘S‘ is true” are materially equivalent if the universe of discourse is restricted to possible objects of belief, just not when any sentence can be instantiated for S.

  16. Whoops, I failed to answer your question. Yes, I think the data do suggest that (for certain classes of statements) “S” and “‘S’ is true” are not materially equivalent; if most speakers disagree that those two statements are equivalent, then it seems to me that those who say they are equivalent are misunderstanding one of the statements.

  17. I see.
    In that case, Blackburn is in good company. The data refute Soames, Tarski (insofar as Tarski was talking about natural languages, which I admit isn’t very far), Horwich, and Quine, to name a few. I bet that takes some of the sting out of the refutation.

  18. Jamie,
    If two English statements meant the same thing we’d expect competent English speakers to agree with one of those statements if and only if they agreed with the other. So if (e.g.) “I agree that pasta is tasty” meant the same thing as “I agree that ‘pasta is tasty’ is true” then we’d expect that Princetonians surveyed would assert the former iff they also asserted the latter. But that didn’t happen. So, we have evidence that “pasta is tasty” and “‘pasta is tasty’ is true” aren’t equivalent.
    As for your elliptical appeal to authority, Frank Jackson, Graham Oppy and Michael Smith (see “Minimalism and Truth-Aptness”) all seem to agree that “pasta is tasty” and “‘pasta is tasty’ is true” are equivalent if and only if “pasta is tasty” is truth-apt (which it is just in case it is a possible object of belief). So they disagree with the Tarskis and the Soames and the Quines of the world (if, indeed, those folks thought that all instances of “S” and “‘S’ is true,” even those where what is instantiated for ‘S’ isn’t a possible object of belief, are equivalent). I guess we need a tiebreaker.

  19. I for one do not think that the folk *have* the sorts of concepts (“objective”, “realist”) that are required to frame metaethical debates. (And this is not a flaw on their part.)
    I will grant that there are some questions you can ask that will gesture in that direction, e.g. is your claim true, do those who make the opposite claim make a mistake, etc. But we all know that this stuff is just the first step in any serious metaethical debate; it’s what the non-cognitivist has to account for, not what he has to deny.

  20. Angus,
    I think I’m with Jamie on this one. One way to explain to asymmetry with the taste judgments is to say that the ‘folk’ think that there are no standards at all for making taste-judgments. This is why they are mere opinions and not true. But no expressivist thinks that there are not standards for making moral judgments. For the very least I could have better attitudes – more coherent, informed and so on. This is why it may be the case that my judgment is false or true.
    Another way to put this is to think that whether we attribute truth to p depends on whether there are any standards of assertibility for p that go beyond whether I merely feel like p. No realist consequences whatsoever follow from this.

  21. I worry that the options given on the survey, ‘true’, ‘false’, ‘mere opinion’, are not exclusive. I would have thought opinions could be true or false, even mere ones. Perhaps the third option should have instead been ‘neither’.

  22. Angus,
    It wasn’t an appeal to authority. I’m not defending any view about truth (here). I was pointing out that a lot more than quasi-realism is riding on the results of the survey, if the results are interpreted as you interpret them.
    I’m pretty sure every philosopher believes that only truth-apt things can be true, so I don’t follow your point about Jackson, Smith, and Oppy. Do they really disagree in that article with anything that Soames or Tarski say about truth? In any case, I think it would be surprising if there were a dispute between those three Melbournians and Soames that could be settled by a survey.

  23. Jussi,
    If what you say is an accurate exposition of Jamie’s views, I think we may all be in agreement on a point that — probably, given that both you and Jamie have “misread” me in the same way, due to inarticulateness on this end — you think is in dispute. Specifically, I agree that “one way to explain to asymmetry with the taste judgments is to say that the ‘folk’ think that there are no standards at all for making taste-judgments.” And there are probably plenty of other ways to explain the study’s data without assuming the falsity of expressivism.
    All I claim is that the study evidences the falsity of a particular argument for expressivism — one that holds that statements of a certain grammatical structure are, ipso facto, truth-apt. Expressivists who endorse this form of minimalism say that, even though moral statements do not express propositions (read: possible objects of belief) / have some other special, non-formal property (some relationship to “standards,” perhaps), they are nonetheless capable of being true and false. In short, the fact that people talk of moral “truths” and “facts” is not in tension with expressivism. (Blackburn appears to affirm something along these lines; he writes “either…’is true’ is not a property of utterances, or…it is not a robust or substantive property. It is not one that invites difficult philosophical questions” (Ruling Passions: 75).)
    The study’s findings belie this argument; aesthetic judgments have the same grammatical structure as (and, it seems, all of the same formal properties of) moral judgments, but survey says they are regarded as incapable of truth and falsity by (arguably) a representative sampling of ordinary speakers of English. If, as seems plausible, one way to determine what sorts of statements are truth-apt is to look at what sorts of claims ordinary speakers of English regard as capable of being true and false, then the study undermines a certain argument to noncognitivism.

  24. Color me skeptical about what experiments like this could possibly contribute to resolving metaethical disputes.
    As Jussi mentions in the opening post, there’s a huge issue with “how good of a meter the possibility of being mistaken is for the acceptance of the sort of judgment-independent objectivity the authors are after”. I almost fell out of my chair when I saw how the authors of the study defined their target (possibility of being mistaken), but cited Blackburn as an example of the view that there are no moral facts and that morality is not objective. But isn’t Blackburn supposed to have built his iconoclasm on claiming that he can “climb the horizontal ladder to objective, mind-independent moral facts”? If I had been a subject in the study, I would have happily agreed that people can be mistaken about moral truths (and quite a few judgments of taste), but it’s set up so that the experiment would never discover that I was an expressivist. So that’s just the taxonomic concern, although I realize you can’t design an experiment to capture every nuance of every metaethical view in the literature.
    Another issue is that unless you beg the question against some kind of error theory, this style of experiment can’t rule out the possibility that 100% of respondents are simply mistaken about the metaphysical status of their claims. Jussi’s followup about Socratic operations on the subjects is also highly suggestive. If entering into this or that kind of discourse can nurture contrary intuitions — say, can turn an objectivist into an error theorist, and vice versa — then how can this kind of experiment even *define* its target phenomenon? How would you know when you were reporting the “right” intuitions about objectivity?
    I have to confess to being a bit confused about the discussion in this thread about “moral phenomenology” or how something “seems objective to us”. I’m not quite sure what it even means to say that something, as a matter of observation “seems mind-independent”. What would it look like *to* *me* if my moral judgments were mind-independent? What observations should I be on the lookout for that would (or ought to) convince me that they’re not mind-independent?

  25. Andrew,
    about the last point. It’s not that moral judgments are mind-independent. Of course making those judgments depends on our minds. The idea is that what the judgments are about is independent about the judgments we make. The phenomenology is generally taken to support this. It feels to us like we try to get our judgments right, we try to recognise the values there are rather than invent them and project them. Of course it may not feel it that way to you – it could feel to you that you invent the values as you will.

  26. Jussi,
    the dealbreaker seems to be how one cashes out the dependence relation. Certainly the primitive noncognitivisms of the Positivists, or cognitivist subjectivisms that say that moral statements are true or false as a function of what the speaker approves of, are dependent on attitudes. But don’t quasirealism or norm-expressivism have truth values (or their justification) *dependent* on attitudes without the *content* of those judgments being identical to stating or expressing the attitude?
    AIUI, the new noncognitivisms can make sense of my thought that some of my own moral judgments can be wrong, because there is a system of norms of justification to which I have pledged my allegiance from which I might make errors of inference. So according to the paper’s authors, I (as an expressivist) believe that morality is “objective”, and yet they cite Blackburn as an example of someone who thinks morality is not objective.
    It certainly “feels like” I’m trying to get my moral judgments right. I just don’t equate this with the feeling that I’m trying to get them objectively right, because I still don’t know what that means, and I still don’t see how this study could even in principle address the issue.

  27. I thought I should highlight a few papers that might be relevant to this discussion.
    First, regarding ‘objective phenomenology’, Simon Kirchin’s paper “Ethical Phenomenology and Metaethics” in Ethical Theory and Moral Practice addresses this issue. He argues, persuasively in my view, that although the phenomenology of moral discourse presents moral constraints as ‘external’ to us, this is not the same as experiencing morality as ‘mind-indpendent’ in the way specified by realism. Blackburn, for example, points out that those in the grip of an emotion often see the external world as constraining their actions (see ‘Errors and the Phenomenology of Value’).
    Also, regarding minimalism and truth. I was not aware that Blackburn tried to argue for expressivism on the basis of minimalism. It seems to me that the adoption of minimalism simply blocks an objection to expressivism. Also, minimalism about truth is not best thought of the view that an appropriate grammatical structure is sufficient for the correct appllication of the truth-predicate – a robust disciplinary regime for the use of the relevant sentences is also required. (This is the distinction between ‘pure syntacticism’ and ‘disciplined syntacticism’, see for example Lenman’s paper in Philosophy and Phenomenological Research 2003.)
    Regarding the paper which began this discussion, I totally agree with Campbell Brown’s post – the fact that ‘opinion’ is contrasted with ‘truth’ seems to me to totally undermine the data.

  28. The most shocking thing about this data are the responses with respect to questions of taste.
    75% of respondents thought there was no correct answer to the question, “Is Shakespeare a better writer than Dan Brown?”
    Oh. My. God.

Comments are closed.