Here is an issue I would love to hear your thoughts about. My interest in this question was prompted by this post today by Kevin Vallier over at BHL.
Some people say that it is inappropriate to hold moral principles in a way that is immune to empirical falsification. G.A. Cohen and T.M. Scanlon disagree. I have not thought about these issues but I don’t get the alternative to the picture Cohen and Scanlon argue for. A first stab might be that what would count as an empirical falsification would be that results the theory predicts don’t materialize. For example, perhaps some will claim that lowering the minimum wage is good because it benefits the worst off. This is surely an empirical claim and surely empirical evidence could possibly show it to be false. And if there are people who refuse to accept that then they are silly. But what this seems to me to show is that we need to distinguish between the intrinsically morally recommended goal and the means that are thought to be useful for bringing about that goal. Of course empirical findings are relevant to understanding what causes what. No pressure is placed upon the moral view that we should especially worry about the worst off by showing such empirical findings.
But pressure from thought experiments can also surely get us to change our minds about what the intrinsic goal of morality is. We might think that maximizing welfare is the intrinsic goal of morality but considering possibilities might get us to realize that this goal can be satisfied in ways that are not morally ideal if, for example, the way to maximize the sum total of welfare would involve vast resources going to a utility monster. This just involves ordinary non-empirical considerations in favor or against this or that as being the intrinsic moral goal. What I do not yet see is how, after we have done all the non-empirical work of settling for ourselves what the intrinsic goal of morality is, how empirical considerations provide evidence for or against such an understanding. Essentially, it seems to me that the relevance of empirical claims to determining the intrinsic moral goal is best understood to be subsumed by non-empirical moral conditionals. I think Scanlon says something very much like this.
Take a person who has a fully worked out view for every possibility of the form "if the empirical world works this way, then that is the morally correct action." That so far seems to me a completely non-empirical theory as it does not at all have to peek at which possible world is actual. I take some people out there to be saying that which possible world is actual, that is, genuinely empirical matters, are directly relevant to assessing the intrinsic moral goal. I don't understand how that picture is supposed to go.
Interesting question. Before agreeing I would want you to say something about the idea of experiments in living and the role that experience can play in testing principles. I will take a wild stab at developing this point.
Say I believe that it would be bad for me to live as a hermit because human companionship is on my list of essential prudential goods. I do all kinds of thought experiments and still believe this. Yet then I hike out to a cabin, live alone for a year, and discover that it was a mistake to put that on my list of essential human goods. On reflection I might say my error was due to the facts (i) that I had no prior access (in thought experiments etc) to what it would really be like for me to live as a hermit, and (ii) that my broadly emotional reactions to living as a hermit were themselves a source of evidence about its prudential value - a source that was not available beforehand.
Perhaps we can imagine gods who could overcome those and related limits when entertaining thought experiments, but we are not such gods. So human moral epistemology should be answerable to experience.
I would think that similar arguments could be made for moral principles and moral epistemology. In fact, writing this reminds me of a recent talk of Elizabeth Anderson's I heard -- I think she was arguing for something like this view by discussing the history of moral progress (focusing on the case of racism, and playing up the role in learning about others experience and reactions, not just our own).
Of course even if there are some such ways in which normative principles should be held subject to the tribunal of experience, this might not really be relevant to the debate on the other blog or the stuff by Scanlon and Cohen you have in mind!
Posted by: Brad Cokelet | October 08, 2013 at 01:28 PM
Hi David,
I agree that the truly fundamental moral principles are not empirically falsifiable in the direct way you are imagining. These principles make no empirical predictions. But I'm open to the idea that there could be *indirect* empirical arguments against them. The kind of argument I have in mind would be an argument for the conclusion that our epistemic justification for believing in the principle is no good. From this we could conclude that we have no reason to think that the principle is true. That on it's own is a pretty interesting conclusion. But we might even be able to go further. If some principle of parsimony is true (or perhaps some other epistemic principle), we might be able to infer that we have reason to think that the principle is not true. And that may amount to the claim that the principle is probably not true. And that would seem to be a moral conclusion.
The central premise is the premise asserting that our epistemic justification for believing in the principle is no good. I take it that we could have empirical evidence for such a premise. For example, you could learn that there is a device implanted by an evil genius in your brain, and that that is what is causing you to have the intuitions that support the principle. Or, less exotically, some empirical results from experimental philosophy might cast doubt on these intuitions. Or perhaps some considerations from evolution.
Posted by: Chris Heathwood | October 08, 2013 at 01:32 PM
Brad,
I agree with that. It surely is the case that more vivid experience with what an option would really be like can reasonably influence our view of how valuable that option is. Full info views have championed this thought for a while. And it could be that in fact our best access to such accurate and vivid info is attained via some actual experiences rather than merely imagining possible experiences. A well told fiction can try to bridge the gap, but surely imperfectly. So this suggests there is an important epistemic role for accurate, vivid depictions of options and that actually trying out options in real life is often the best way to get those accurate and vivid depictions. I wonder if that is all that those who think basic moral principles are falsifiable have been seeking or if they want more than that?
Posted by: David Sobel | October 08, 2013 at 01:46 PM
Well I was poking around on that blog for a minute and there does seem to be some other idea in play.
Vallier writes: "To say, for instance, that socialism is required by justice even if it is empirically infeasible seems to me the mark of an ideologue and an essentially ideological research program."
Maybe he is thinking of moral principles in a broadly Korsgaardian way (they are proposed solutions to problems not descriptions of facts out there in the world). He could then be thinking that it is a mistake to *assert* that justice requires socialism because that is an *unworkable* solution to the agreed upon problem. And then maybe he could embrace some sort of pragmatic conception of truth and argue that the claim about justice is false?
I don't think this is an attractive position, of course!
Posted by: Brad Cokelet | October 08, 2013 at 01:52 PM
Chris,
Couldn't we reason non-empirically that if such and such empirical findings come in, that would have thus and so impact on our moral view? And this without peeking at which possible world is actual. If so, then I am thinking we could become fully morally opinionated about all the possibilities and we would not have any reason to change our mind about any of them based on which world is actual.
Posted by: David Sobel | October 08, 2013 at 01:52 PM
I had a thought sort of similar to Brad's. Pragmatists, for instance, sometimes seem to suggest that various theories, including moral theories, ought to be judged against a general set of goals defined in a very broad way. Say moral theories are responses to the problem of social cooperation, or something of that nature. Then you could find out via empirical investigation whether belief in a certain moral theory or acceptance of a certain moral theory engenders social cooperation, or whatever.
The problem with this sort of thought, though, is that it seems like the moral work is being done by a conception of social cooperation and its moral importance which seems independent of the empirical data. Furthermore, at best the empirical data can measure belief in a view---and it is a further controversial moral claim that a moral view ought to be assessed given the effects of people's belief in the view.
Posted by: Dale Dorsey | October 08, 2013 at 02:00 PM
While this is something of a special case, with narrow application, I'd be inclined to say that some moral principles can be empirically refuted if they entail that we ought to do things that as a matter of empirical fact we cannot do.
Posted by: Dale Miller | October 08, 2013 at 02:19 PM
Hi David,
Hmmm. Let's try a concrete, simple case. Suppose I intuit that pain is necessarily intrinsically bad. On this basis, I judge that it is likely that pain is necessarily intrinsically bad. As you describe, I reason non-empirically that if in fact this intuition is caused by an evil genius, then it provides no epistemic justification.
So far, I've yet to peek at which world is actual. Here are two words that could be actual:
Wn: A "normal" world with no evil geniuses.
Wg: A world in which my intuition is caused by an evil genius.
If I discover that Wn is actual, I'll be pretty confident that pain is necessarily intrinsically bad. If I discover than Wg is actual, I'll have no confidence that pain is necessarily intrinsically bad.
I think these last claims can basically be rephrased as follows. If Wn is actual, it's pretty likely that pain is necessarily intrinsically bad. If Wg is actual, the likelihood that pain is necessarily intrinsically bad is much lower. Thus, discovering which world is actual will help me determine how likely it is that pain is necessarily intrinsically bad.
Posted by: Chris Heathwood | October 08, 2013 at 02:32 PM
I guess I was thinking that empirical facts are not playing a crucial moral role if we can work out a priori what the moral relevance is of all empirical facts. I guess you are thinking that empirical facts are playing a crucial role if they are needed to determine what has intrinsic moral value. Usually, I take you to be saying, empirical facts are at best needed to learn what has instrumental moral value but in odd cases they can be needed to determine what has intrinsic moral value.
Posted by: David Sobel | October 08, 2013 at 02:42 PM
I think that one way to think of the point I'm making is that our ability to work out a priori the moral relevance of all the empirical facts can itself be affected by empirical facts. There can be empirical defeaters for a priori justification.
To be sure, I don't think that empirical facts can themselves determine what has intrinsic moral value. But they can affect what it is reasonable for us to believe concerning what has intrinsic moral value. That's why I described the argument as "indirect."
Is this only in odd cases? The example I used was of an odd case (an evil genius case), but the more interesting cases may not be odd at all. One such possibility is that certain classes of our intuitions might be unreliable because, say, they are connected in some way to the "emotional part" as opposed to the "rational part" of our brain (that's kind of a caricature of part of Joshua Greene's argument). Another possibility is that evolutionary considerations cast doubt on some or all of our moral intuitions.
Posted by: Chris Heathwood | October 08, 2013 at 03:17 PM
I now think what I reported myself as thinking at 2:42 was confused.
Posted by: David Sobel | October 08, 2013 at 03:26 PM
Good, thanks Chris, I see better what you are saying now.
Posted by: David Sobel | October 08, 2013 at 03:42 PM
Hi David. Not sure I'm following the discussion properly. It sounds like you doubt that moral claims are empirical claims or have testable consequences. Is that right? I think it's true that moral claims don't have testable claims in isolation. But then it's arguable that most scientific claims don't have testable claims in isolation. But various kinds of claims, including moral claims, do have testable claims when conjoined with suitable auxiliary hypotheses. So, for instance, my belief that good people tend to keep their promises to others even when that requires personal sacrifice and my belief that you are good person lead me to believe that you will keep your promise to Bilbo even at some personal cost to yourself. If you keep your promise to Bilbo, this seems to provide some confirmation of my moral assumptions (taken together). If not, then I have to reconsider my belief about the promise-keeping behavior of good people or my belief that you are good (or both). If there is nothing you and other people could do that would affect my two moral beliefs, then we might think that I am treating them as unfalsifiable. (I wrote about this in my moral realism book, pp. 136-38.) Is this responsive to you query?
Posted by: David Brink | October 08, 2013 at 04:24 PM
I'll say a bit more about what I have in mind as the alternative. We can see it if we focus a bit on Cohen's criticisms of Rawls's constructivism. According to Cohen (and let's assume he's correct for the moment), Rawls formulates political principles based on empirical facts, like moderate scarcity or certain features of human moral psychology. But the *real* normative principles can't be fact-sensitive in this way because they must *derive* their normativity from some logically prior principle that is itself fact-insensitive (basically a regress argument; one any constructivist would happily block by accepting a fact-sensitive moral principle as brute).
In the Rawlsian case, some moral principles depend on contingent empirical facts for their truth, even if they are not the *ultimate* normative principles. Rawls's two principles, for example, are supposed to be fact-sensitive in this way, dependent on a whole host of contingent empirical facts. So that's the alternative, I think: stuff in the orbit of constructivist approaches in normative ethics and political philosophy.
For what it is worth, one reason I don't think Cohen's criticism of constructivism is sound partly because I think you can be a constructivist in political philosophy but hold that there are fact-insensitive moral principles that specify the relevance of facts about human nature (a reply like the one you offer). I think any Rawlsian, especially a political liberal, should respond in this way.
Posted by: Kevin Vallier | October 08, 2013 at 04:29 PM
David,
Tell me if I am getting the picture. Suppose I start out thinking that morality requires maximizing happiness and I start out thinking that Joe is a very moral person. Then the empirical fact that Joe fails to maximize happiness puts consistency pressure on my overall ethical thinking. That seems right and interesting. But it would not, if I am thinking about it correctly, show us a way for empirical matters to put pressure on completely internally consistent ethical perspectives.
Posted by: David Sobel | October 08, 2013 at 06:04 PM
Kevin,
I don't really feel like I can see from that you wrote exactly what Cohen was trying to say or what is at issue. Perhaps I'll try to find what he says and read it for myself to see if that helps.
Posted by: David Sobel | October 08, 2013 at 06:07 PM
That's right. Not any set of moral propositions together have empirical implications. The same holds for non-moral propositions.
Posted by: David Brink | October 08, 2013 at 06:10 PM
Fair enough.
Posted by: David Sobel | October 08, 2013 at 06:14 PM
Hi Kevin (if I may),
I was hoping you might say more about how "you can be a constructivist in political philosophy but hold that there are fact-insensitive moral principles". It seems to me that part of what it is to be a constructivist is to deny that there are any such principles, even if the principles in question merely specify the relevance of facts about human nature, and thus serve to ground some set of fact-sensitive principles. In other words, I was under the impression that, in order to qualify as a constructivist, you've got to think that all moral (political) principles can be explained (ultimately) by appeal to facts about us. If this isn't right, I guess I'm left wondering what's supposed to explain the difference between the constructivist's view and Cohen's.
Posted by: Chris Howard | October 08, 2013 at 06:30 PM
I think Scanlon and Cohen both have in mind something like this. Ordinary moral (and other normative) claims plainly do have an empirically vulnerable component. I don't think I paid close enough to the Cohen stuff to remember give a good example, but for Scanlon there are loads of obvious examples, because claims about what reasons people have are factive, in that the reason itself has to be a fact (and the reason is often an empirical fact). But, we can just 'factor out' the empirical bits and be left with pure normative claims, which have no empirical content. (I know this is what Scanlon thinks.)
I'm pretty sure neither of them considered the sort of thing Chris Heathwood mentions.
Posted by: Jamie Dreier | October 08, 2013 at 06:34 PM
Great discussion! I don't want to cut it short but I just want to thank everyone so far for the very helpful thoughts.
Posted by: David Sobel | October 08, 2013 at 07:44 PM
To add to David's reply to David Brink. Surely my belief that Joe is a good person is, at least in part, an empirical claim that can (easily) be falsified. My belief that good people do not rape, murder, etc. seems not to be empirically falsifiable. So, if Joe does bad stuff the 'choice' as to which belief to jettison is not really a choice.
Posted by: David McNaughton | October 08, 2013 at 08:02 PM
Hm. Well, okay, but that's it, after McNaughton no more Davids.
Posted by: James Dreier | October 08, 2013 at 08:33 PM
Empirical confirmation works best when we can test hypotheses by relying on auxiliary assumptions about which we are very confident and for which we have good independent support. So in some cases it is easy to know which assumption should be given up when the empirical claim they jointly imply does not obtain. The fact that confirmation is sometimes this easy does not mean that it always is. So I would take David McNaughton's example to support my claim, rather than undermine it.
Posted by: David Brink | October 08, 2013 at 09:10 PM
Addendum: According to the picture I was sketching, moral claims confront experience as a corporate body, rather than individually, with the result that confirmation of moral claims is holistic. This means that when the world does not behave as our moral claims jointly imply that it should, we have to determine where to lay the blame and, sometimes, which moral claim to give up (e.g. our principle about the promising keeping behavior or the belief that David is a good person). Sometimes this is easy when we are much more confident and have greater independent support for some than for others. In some cases, some claims may seem especially fixed or unrevisable. Consider the claims that causing unnecessary suffering is wrong, that genocide is wrong, or that rape is wrong. (1) One might go Quinean about these, claiming that these allegedly fixed points are not unrevisable in principle or come what may, but simply harder to imagine revising. If we tried hard, we could imagine other evidence that might lead us to question those maxims. (2) Whether or not that Quinean move is attractive, the holistic point still stands. Even if some moral commitments are fixed points, unrevisable in principle, not all are, as the holistic examples illustrate. But then moral claims are not per se immune to observational evidence. (3) Interestingly, the most plausible candidates for unrevisable status are comparatively discrete moral maxims such as causing unnecessary suffering is wrong, genocide is wrong, rape is wrong, rather philosopher's principles, such as contractualism, the difference principle, or self-ownership (to pick a few at random).
Posted by: David Brink | October 09, 2013 at 07:53 AM
I was going to say something incredibly insightful that would lead to total consensus, Jamie, but I'll respect your wishes.
Posted by: David Faraci | October 10, 2013 at 11:38 AM