Thursday, May 05, 2011

The Moral Lottery

There are any number of logically consistent moral belief sets one might have, e.g. whereby the putative badness of pain waxes and wanes with the moon, or even where pain is seen as intrinsically good rather than intrinsically bad. Suppose that moral realists are right to think there's a mind-independent fact of the matter as to which of these possible moral views is correct. We don't think that the abstract moral facts played any causal role in explaining why we ended up with the particular moral beliefs that we have (there will instead be some psychological, sociocultural, or evolutionary story of that). So shouldn't we think it extraordinarily unlikely that our moral beliefs are (anywhere near) true, if realism is correct? So argues Sharon Street in 'Objectivity and Truth: You'd Better Rethink It' [pdf].

First consider how the moral realist could respond if the targeted "believers" (whose reliability was being put in question) were not themselves, but some alien species. In that case, we may respond to Street by acknowledging that, in the absence of further info, we may have no antecedent reason for expecting this alien species to be morally reliable. But rather than trying to assess their reliability more or less a priori, from only very abstract info, we would do better to look at their concrete moral judgments and practices. If we look and find that the aliens seem generally sympathetic and altruistic, concerned to promote the wellbeing and non-harmful life goals of other sentient beings, then we may be reasonably confident that they're on the (morally) right track, whatever the causal process that brought this about. (If, on the other hand, they seem to enjoy gratuitous torture, then we'll judge them to be morally abhorrent.)

Of course, using our first-order moral judgments in this way is no longer going to be dialectically effective when it is our own reliability that is in question. But that shouldn't necessarily concern us (see also my Railtonian defense of 'default trust'). As Parfit writes in On What Matters:
[Street holds that] we must show that the evolutionary forces have led us to form true normative beliefs, and we must defend this claim without making any assumptions about which normative beliefs are true. What Street here requires us to do is impossible. Some whimsical despot might require us to show that some clock is telling the correct time, without making any assumptions about the correct time. Though we couldn’t meet this requirement, that wouldn’t show that this clock is not telling the correct time. In the same way, we couldn’t possibly show that natural selection had led us to form some true normative beliefs without making any assumptions about which normative beliefs are true. This fact does not count against the view that these normative beliefs are true.

I think this is exactly right. That's not to say that our first-order normative beliefs necessarily suffice as a decisive refutation of Street's skeptical argument against realism (there's more to Street's argument than there is to the whimsical despot's demand), but nor can they be dismissed as entirely irrelevant or 'illegitimate' to appeal to. If we want to work out what to believe, we need to consider all our beliefs -- everything we find most plausible, and judge to be (likely) true -- in wide reflective equilibrium, and work out how best to fit the various claims together: what to keep, and what to discard, to yield the overall most plausible conclusions.

At the heart of Street's argument is: (1) the observation that realists must consider themselves lucky (isn't it a wonderful coincidence that natural causes just so happened to influence us to hold moral beliefs that roughly align with the mind-independent moral truth!?), and (2) an epistemic principle to the effect that there's something epistemically irresponsible about retaining one's belief in the face of such luck.

I agree with the first premise: as realists, we must consider ourselves epistemically lucky. And the second premise has a great deal of prima facie plausibility to it. But it's not nearly as incontrovertible as Street seems to assume. For one thing, I certainly don't think that plausible-seeming abstract principles necessarily trump plausible-seeming concrete judgments, e.g. that gratuitous torture is wrong regardless of whether the torturer herself is committed to this conclusion. Secondly, there are more direct grounds for doubting the principle, as it often seems completely appropriate to consider ourselves epistemically lucky, e.g. if we narrowly avoided being brainwashed, or raised by a cult of counterinductivists, or some such.

To bolster her case, Street compares the moral realist to someone who stubbornly believes herself to have won the lottery, even before hearing the results of the draw:
At this point, however, the normative realist is in no better position than the person who question-beggingly insists that she won the New York Lottery, even though she has no reason to think so apart from the fact that she entered it. If we are normative realists, we think there is a “winning” coherent system (or systems) of normative thought; we also think there are countlessly many false coherent systems of normative thought, which, but for sheer good fortune on our part, causal forces might have shaped us to endorse; we think that as it so happens, ours is among (or approximating) one of the true ideally coherent systems; but when asked to give our reason for thinking so, all we can say is to repeat, in so many words, that it is among the true ones—to insist that we, and not the countless number of mistaken possible others, “see” or “sense” what is normatively true. But this is no better than insisting, without any non-trivially-question-begging reason to think so, that one has won the New York Lottery. Given the odds we can reasonably suppose to be in play in this “normative lottery” case, we should conclude that in all probability we didn’t win—that, if there is indeed such a thing as the robustly independent normative truth we are positing as a substantive normative premise, then we are probably among the unlucky ones who (just like the ideally coherent Caligula, grass-counter, hand-clasper, and so on) are hopeless at recognizing it.

The main problem with this analogy concerns the "odds" that we assign to the various outcomes. Given our understanding of the lottery mechanism, we know that we should assign equal odds to each possible outcome, and hence the odds we should assign to some particular ticket (#139583923, say) winning is extremely low. For Street's argument to work, it must be that we are similarly required to assign equal (or roughly equal) odds to the truth of each possible normative system. But why think that? I certainly don't assign equal odds to all possible normative systems -- I think it's overwhelmingly more likely that pain is intrinsically bad than that it's intrinsically good -- and I don't see anything in Street's argument that suggests I should change my mind. (It's not as though the normative truth was itself settled by a chance process that gave equal objective probability to each possible outcome.)

Street might try to push the analogy by asking us to imagine a lottery winner who likewise insists that the (a priori) odds of ticket #139583923 winning are much higher than for any other ticket. There are a couple of things to say about this. Firstly, I agree that such a person would be crazy. But it's far from clear that there's necessarily anything incoherent about their beliefs: they might just have wacky priors, like the epistemic counterinductivist. I think they're being irrational as a matter of mind-independent normative fact, but of course Street cannot think that. What's more, my reasons for thinking that equal odds are required in the lottery case (viz. something like Lewis' "Principal Principle" relating objective chances to rational credences) simply don't carry over to the normative case, where the truth of the matter was not settled by a chance process.

So I think Street's argument strictly fails: There's nothing incoherent about assigning higher odds (a priori) to anti-pain normative systems than to pro-pain views. So the moral realist needn't conclude that she is "probably" among the misguided when she holds to her anti-pain view. It's true that our views are formally analogous to others (like the pro-pain view) that we hold to be mistaken. But I don't think there's anything necessarily wrong with the form of those other views -- they're just wrong on the substance. So I'm not particularly troubled by sharing the form of the wrong-headed view, just with different substance. That's how the right view would have to look.

Suppose you grant that the realist needn't (on pain of incoherence) take herself to be "probably" among the misguided. Still, you may think, perhaps it's still the case that she really should conclude this. But now you're positing a mind-independent normative truth. And if the argument against normative realism depends upon the truth of realism, then that can't possibly be a sound argument!

So there's no sound argument against realism to be found here. But there is at least a challenge for the realist, to say more about how moral knowledge is possible on their view. I can't give a full answer here, but here's a rough sketch of a view: There's a fact of the matter as to which psychologies qualify as fitting or 'substantively rational'. (The fitting psychology might, among other things, endorse modus ponens as a rule of inference, accept inductive over counter-inductive norms, and take the badness of pain as a provisional moral datum. The rough idea being that the 'fitting' psychology is one that reflects or fits with the normative facts.) Inferences drawn, and conclusions reached, by agents with fitting psychologies are ipso facto justified -- and if true, eligible to qualify as knowledge.

That seems a coherent view, and one that allows for (fitting) agents to know the mind-independent moral facts. One might think of it as similar to a kind of reliabilism (the view that true beliefs formed via reliable methods thereby constitute knowledge). Reliabilism is clearly immune to Street-style skeptical arguments: the mere fact that there are other, less reliable, processes does nothing at all to undermine the processes that I actually use (even if I have no independent way of verifying that the process I use is one of the few reliable ones). And it brings out the disanalogy with the lottery case: presumably there's no way to be a reliable predictor of lottery results, the way that an anti-pain moralist might be a reliable predictor of moral truths. Similar observations may be made when using the fittingness view in place of reliabilism.

In short: it's epistemically undermining if this psychology I'm using was 'lucky' to get things right. That casts doubt on the rationality of my psychological processes. There's no such problem if I'm merely lucky to have acquired a reliable psychology in the first place. That doesn't cast doubt on the rationality of the psychology I have; it merely suggests that I could easily have ended up with some other, less rational, psychology instead. To which the appropriate response is just, "Well thank goodness I didn't!"


  1. I think you are largely right about how the Parfit-Street discussion turns out.

    I also agree that the kind of realist we're talking about has a job to do in normative epistemology. I don't think your sketch helps very much. Suppose we Ramsify out your word "fitting". It just sounds like your explanation is that there is some property F had by certain psychologies, and people with those psychologies make justified judgments about normative facts. This does very little to dispel any concerns I have about how it could be possible to know anything about these mind-independent normative facts.

    Maybe I would be more persuaded if I heard more of a story about fittingess. Could we do more than giving me a list of beliefs that people with fitting psychologies would find plausible, or a list of attitudes that it would be fitting to have?

  2. Can you imagine a possible (evolutionary) genealogy that (a) could hold true of each one of our "normative belief generating systems," and (b) would undermine your belief in their reliability (and, a fortiori, the reliability of your normative belief generating system)? I think I'd be clearer on what a fitting psychology is if I knew how to confute the claim that I have one.

  3. Street's argument seems another example of something Lewis commented on; many philosophers don't think general skeptical problems have been solved and aren't totally comfortable with the situation, but have learned to ignore the standard skeptical worries most of the time. However, when a new worry comes up that superficially looks different enough from the standard skeptical worries, it doesn't trigger their habit of ignoring standard skepticism, and their discomfort with the situation comes out. I can certainly think of some general skeptical arguments that seem to me to be very similar to Street's argument, and the fact that reliabilism seems relevant, of course, further evidence that the problem really is just standard skepticism, not anything special about morality.

  4. I haven't read Street's paper but may get to it later so I'll just comment on your take of her argument.

    I actually doubt these parts of her premises:

    "We don't think that the abstract moral facts played any causal role in explaining why we ended up with the particular moral beliefs that we have (there will instead be some psychological, sociocultural, or evolutionary story of that)."


    " (1) the observation that realists must consider themselves lucky (isn't it a wonderful coincidence that natural causes just so happened to influence us to hold moral beliefs that roughly align with the mind-independent moral truth!?)"

    First of all, the first passage. I don't know why it is assumed to be so obvious that moral facts play no causal role in explaining why we ended up right or close to it. Even if these facts are abstract, it would still require some serious argumentation IMO to arrive at the claim that they do not causally influence our normative accuracy. And why must we assume that moral facts are abstract?

    Also, can we say that reason is reliable and that part of getting things right morally is reasoning correctly? Can't we make a partial analogy with mathematical truth? Perhaps there is an element of pure reason in moral deliberation as well. That seems reasonable to me.

    Also, Street seems to assume that truth is like a vanishing point on the horizon, so demanding that it's either a perfect fit with the moral bull's eye or outright falsity. I think like truth in science, we should see things in degrees so here's a parity argument.

    Isn't it reasonable at least in science to have degrees of truth or else avoid very implausible conclusions? Consider Newton's physical theories. They have been supplanted by Einstein's and quantum mechanics because the empirical evidence suggest they do not predict phenomenon accurately under certain conditions (such as speeds approaching luminal speeds). But if knowledge entails truth and Newton's theories were, strictly speaking, not true, then tollensing, that would seem to imply that Newton's ideas did not supply us with any knowledge. But that seems preposterous. Likewise, with Einstein's relativity which seems to go against the most accurate set of theories so far developed, namely quantum mechanics. Can we say that Einstein did not supply us any knowledge whatsoever? And what if the physics 50 years from now say that our theories are inaccurate? Is it reasonable to say that physics has not supplied us with any knowledge so far? Again, it seems very implausible to conclude this. The only way to remedy it seems to have a notion of truth by degree. Newton was on the right track as Einstein and so forth. If this permissiveness can be granted to science, can't moral truth be as forgiving? We may be on the right track and partly right; some of us, closer than others, etc though many of us may be strictly speaking wrong but not wholly so.

    Street's possible reply of course will likely be that unlike physical reality, moral truth cannot constrain our theories. They are just out there, floating around somewhere. But it does seems to be the case that moral truths do constrain in some ways, our moral judgments. It seems a priori that no moral theory should justify torture for fun. And why can't psychological, sociocultural and evolutionary stories nudge us closer towards the truth as well? If even there is a slight net nudge (as it seems to me, far more than a slight net nudge!) towards moral truth, then it seems we have grounds for a degree-of-truth approach.

  5. Nick - I survey a few possibilities here. I tend to oscillate between understanding #3 (an orientation towards the true and the good) and #5 (the psychology with highest a priori expected value).

    Miles - I don't think there's any "neutral" test for moral reliability, so I can't see genealogies telling us much one way or another. (As explained in the post, we need to assess the output beliefs on their substantive merits -- but of course there's no way to do that without taking some moral assumptions for granted.) There are some trivial exceptions, e.g. if we were to learn that our beliefs were crafted by an omniscient evil demon who aimed to make us believe moral falsehoods.

    Aaron - yeah, that sounds plausible. (In fairness I should note that Street claims to be exploring a kind of 'internal' skepticism, based on norms we already accept, rather than a totally 'external' skepticism that casts all norms into doubt. And her conclusion is anti-realist constructivism, rather than skepticism per se. But even so, I do think there's something to your point...)

    NChen - If Street's argument is successful, it casts doubt not only on our claim to "perfect" truth, but even to being anywhere close. What non-question-begging reason can be given for insisting that "no moral theory should justify torture for fun"? One can imagine an internally coherent alien viewpoint according to which sadistic torture is the greatest good to which all actions should aim. (Or do you deny that such abhorrent views are internally coherent?) Sure, we think that's obviously wrong, but (Street would point out) the alien would think the same of us, and if moral realism is true then what non-question-begging reason could be given for thinking that we are more likely than the aliens to be the lucky ones who have latched onto the mind-independent truth?

    "can we say that reason is reliable and that part of getting things right morally is reasoning correctly?"

    Reasoning (i.e. reflective equilibrium) helps us to iron our inconsistencies. But if Street is right that there are multiple radically divergent ideally-coherent moral belief sets (as seems plausible), then whether reflective equilibrium leads us to a coherent anti-pain view or a coherent pro-pain view is going to be radically contingent on our starting points. So Street's challenge remains: what non-question-begging reason is there to expect that human starting points are more likely than alien ones to be right about the mind-independent moral facts?

    "I don't know why it is assumed to be so obvious that moral facts play no causal role..."

    Hard to argue for what seems obvious (perhaps you could say more about the basis for your doubts?), but here's one attempt: The physical world is causally closed. Moral properties, if there are any, aren't (identical to or reducible to) physical properties. So moral properties aren't causally efficacious.

  6. Richard said:

    >If Street's argument is successful, it casts doubt not only on our claim to "perfect" truth, but even to being anywhere close. What non-question-begging reason can be given for insisting that "no moral theory should justify torture for fun"? One can imagine an internally coherent alien viewpoint according to which sadistic torture is the greatest good to which all actions should aim. <

    The “what non-question begging...” reply would seem to prove too much. It seems to me that all explanation, scientific, moral, common sense, mathematical must come to an end somewhere. What if we were to meet an alien people that denied our view that material things exist independent of our observation or that 2+2=4 or any unprovable commonsense claim? It seems that the right response in such scenarios is not to say that our systems differ from theirs and that there is no fact of the matter but to say that we must be misunderstanding each other in some way. That seems to be the most reasonable conclusion if some such scenario as described obtains, not to doubt our most fundamental moral convictions. There's a feeling of non transparency that I get from such skeptical cases. I always feel that there has to be a misunderstanding somewhere between us and the aliens, or that the aliens are collectively suffering from kind of mental illness. As unlikely as these later may be, both (or that the aliens are just plain wrong) seem far more likely than the claim that their understanding is on equal moral ground as ours.

    >(Or do you deny that such abhorrent views are internally coherent?)<

    I am denying that they are wrong or that there's a serious verbal/conceptual confusion, not that they are necessarily inconsistent.

    >Hard to argue for what seems obvious (perhaps you could say more about the basis for your doubts?), but here's one attempt: The physical world is causally closed. Moral properties, if there are any, aren't (identical to or reducible to) physical properties. So moral properties aren't causally efficacious. <

    I am not sure I understand this line of reasoning. I suppose that Street is assuming that naturalism in moral properties is false. But even so, suppose that moral properties are not physical or natural properties. But it is not obvious to me that that would automatically mean they are causally non efficacious. If they are not physical properties but they are properties of some sort, I can only assume that they must be abstract properties or perhaps we say that the moral realm is populated only by abstract “moral entities” whatever they may be. I can think of no other objects other than natural or abstract ones. But some abstract objects are causally efficacious. What is the difference between a car that causes a crash and that car's singleton set causing it? I assume that all sets are abstract but surely impure sets, sets containing material objects as members occupy space and causally move things in the material world? Likewise, if the monists are right and that the only fundamental thing in the world is just the world and everything else is gotten through abstraction, everything other than the world is in some sense an abstract object. But that would not mean that they are then really causally non efficacious. Also, we seem to know and reason with purely abstract objects such as the objects in math and if they are not causally efficacious it is unclear how that works but it does seem to work.

  7. Do you think there is something special about the moral case, or can genealogies never debunk? For example, assuming the truth of the genealogy, what about this as a "challenge" to the aesthetic realist: humans evolved their capacity for making aesthetic judgments (that mountain is beautiful; this duck is ugly) because the mere ability to converge on aesthetic judgments (quite independent of whether they were true or false) conferred a reproductive benefit. Thus, the deliverances of your aesthetic sensibility are the deliverances of something that is not "for" delivering true beliefs; there is no reason to think its deliverances are reliable.

    The aesthetic realist might reply: we need to assess the output of our aesthetic sensibility on their substantive merits - but of course there's no way to do that without taking some aesthetic assumptions for granted. But this seems mistaken. The story told about how we came to have our aesthetic sensibility obviated the need to assess its output on the merits by suggesting that we have no reason to think we can effectively do that.

  8. Miles - yes, I think the genealogical story is similarly impotent as an objection to aesthetic realism (though some kind of aesthetic relativism seems independently plausible to me, I don't think the genealogical story is doing any work here). The claim that "we have no reason to think we can effectively [assess the outputs on their merits]" presupposes a skeptical premise that the realist need not grant.

    We often have reason to think a process reliable even if was not designed "for" reliability. Supposing I'm looking at a bunch of randomly-assembled turing machines. One happens to give correct answers to all the arithmetical questions I input. I now have inductive reason to consider it reliable at arithmetic, even though it was not made "for" delivering true beliefs (it was assembled randomly).

    So the skeptic must supplement this claim with some further principle disallowing our appeal to (believed) normative facts when assessing the reliability of our normative belief-forming mechanisms. But it's not at all clear why the non-skeptic should go along with this skeptical principle.

    P.S. I don't think that genealogies can never debunk. It's going to depend on what we take the relevant philosophical data to be. If I have some religious experience, and I conclude that God exists as the "best explanation" for my having of this experience, then an alternative causal explanation is clearly going to be relevant (and, I'd expect, successful!) for undermining that inference. If we take the relevant data to not just be our experiences of it seeming that there are moral facts, but further take those external facts themselves as data to be explained, then debunking explanations of how we might form the belief even in the absence of the reality is no longer so relevant -- because my acceptance of the (putative) reality does not depend on it being the best explanation of my psychological states. (It's an interesting question when it's legitimate to take external / non-psychological facts as 'data' in this way -- feel free to add your thoughts in comments to the linked post.)

  9. NChen - Street doesn't want to deny that "all explanation... must come to an end somewhere." She notes that in the empirical case, we can give an internal story about how it is that our senses are reliable (if we evolved as we think we did). Her worry is that there is no such (even merely internal) story available for the moral realist: we must take our correctness to be, in some sense, a matter of brute luck, rather than of causal sensitivity to the facts.

    Now, I think we can give an internal story (as set out in my main post) if we're allowed to appeal to the contents of our normative beliefs in assessing reliability. I guess it's here that Street's worries about question-begging arise. In the empirical case, we could give an account of why evolved sensory apparatus can be expected to be reliable, without appealing at any point to the particular sensory experiences we've had. The normative case seems to involve a much tighter circle, so I can understand finding the sort of explanation I'm giving here to be more troubling than in the empirical case. (The empirical case involves considerations that are at least partly independent of our particular actual sensations, in a way that the moral case is not.)

    Back to the aliens: I certainly don't think that "their understanding is on equal moral ground as ours." The aliens are indeed "just plain wrong", or so I believe. But (as Street would say) that's just to repeat our belief that we've latched on the objective moral truth and they haven't. It's not to give any explanation of how we came to be the lucky ones (let alone how we came to be right without appealing to brute luck!). You can even call the wrong view a kind of "mental illness", but again it's not clear that this adds anything to the claim that they're wildly and horrendously wrong. The question is: what logical or epistemic mistake have they made? If perfect logic can just as well lead to their conclusion as ours, then isn't it a matter of luck (rather than, say, logical acumen) if we're the ones who end up in the right place?

    On causation: I find it odd to say that sets cause things whenever their members do. (That's a lot of causal overdetermination!) But regardless of how you want to talk about causation, it's clear that the moral qua moral can't be relied upon to "push" agents in the right direction any more reliably than the immoral does. We may respond to the natural components of these facts, but it would seem crazily supernatural to think that the moral properties as such can be relied upon to push us in the right direction. If you want to say that the badness of pain is (in some indirect sense) part of what causes us to believe the anti-pain view, then symmetry requires that the badness of pain (not properly recognized as such, of course) is also what causes the pro-pain people to their view. It's the same physical stuff pushing us around in either case, and there's no antecedent reason to think that physics is going to push us in the morally right direction. Or, if you think otherwise, I think the onus is really on you to spell out your account of this...


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)