I want to discuss a problem for ethical intuitionism and an argument that seems to show that ethical intuitionists either have to embrace skepticism or naturalism. It's an interesting argument and I'm not entirely convinced that the response I set out below adequately addresses the worries that motivate it, but I thought I'd give it a shot. The argument from cosmic coincidence is taken from Matthew Bedke's Pacific Phil Quarterly paper (here or here if you can't get library access). Before we get to the argument, I should say that the view I want to defend is the view that it's possible to have non-inferential moral knowledge based on intuition alone even if we have no independent grounds for thinking that our intuitions are reliable (provided, of course, that there aren't reasons to think intuition is unreliable that we ought to take account of). The argument seems to show that if ethical properties are non-natural properties, intuitionists have to say that we cannot have moral knowledge. Once we recognize this, we cannot have justified moral belief. (Maybe you can have justified belief without knowledge, but I don't think you can justifiably believe that which you have good reason to think you aren't in a position to know.) So, given some assumptions about the metaphysics of moral properties, the argument can lend some support to the skeptical view that it's not possible to have moral knowledge (ST1) and that it's not possible to have justified moral judgment (ST2).
It’s not clear (to me) what the commitments of naturalism are, but it’s often thought there’s more to moral naturalism than just the thesis that the moral properties supervene upon the natural properties. Let’s assume that the non-naturalist agrees that this supervenience relation holds. In this post, I’ll defend a version of intuitionism on which moral properties supervene upon natural properties from the argument from cosmic coincidence:
(1) Your intuitions are physical events or states.
(2) The physical world is causally closed.
(3) Thus, your intuitions are fully causally closed.
(4) Ethical facts or properties are non-physical facts or properties.
(5) Thus, ethical facts or properties do not cause anything in the physical world.
(6) Your physically caused intuitions accurately represent non-causal ethical facts or properties only if there exists a “cosmic coincidence” between the causal order and the non-causal facts or properties.
(7) The need for cosmic coincidence, once realized, constitutes a defeater.
(8) Thus, whatever intuitive justification for beliefs in ethical facts or properties is defeated once you realize the need for cosmic coincidence.
For the sake of this discussion, I’ll grant (1)-(4). The real work is done by (6) and (7).
In support of (6), Bedke remarks:
If one’s ethical commitments are psychological, physical, and so part of the casual order, and if ethical facts or properties are not part of the causal order, how, exactly, do the causal forces of the world conspire to ensure that one’s ethical commitments, including intuitions and beliefs, accurately represent the ethical facts or properties? After all, the latter are not part of the causal order and so they cannot causally influence one’s commitments.
So as to clarify the guiding intuition that supports (7), he adds:
One could argue that there is a kind of coincidence foreclosed by metaphysical necessitation. Consider the possibility that some natural fact in the causal order N causes me to have an ethical intuition and a subsequent ethical belief that p. Suppose that N also metaphysically necessitates the ethical fact that p. In such a case, it would not be metaphysically coincidental that my intuition and belief reflect the ethical fact, for both hold in virtue of N, where the in virtue of relation is causal in the case of the intuition and belief, and the in virtue of relation is metaphysical in the case of the ethical fact. This is true enough, but it does not eliminate the kind of coincidence central to this paper. For notice how lucky I am that the metaphysical necessitation was tailored to necessitate the very fact my ethical belief represents. After all, N could have necessitated some non-p fact, q. To be clear, it couldn’t do so metaphysically speaking, by hypothesis, but it certainly is conceptually possible that it necessitate q, and more importantly, it is evidentially possible that it necessitate q or any other non-p fact given that my intuitive evidence of the ethical fact depends only on the causal order.
I’m not convinced that intuition favors the case against the intuitionist. In fact, I think we can use this to run a thought experiment that supports the intuitionist’s cause.
Let’s imagine that a being with vast power and little better to do creates a series of planets populated by creatures that are in many ways physically and psychologically similar to us. On each planet, these creatures are wired up in such a way that they’re disposed to attribute moral properties when they take it that certain natural properties are present. So, for example, there’s a planet on which the creatures are disposed to think that if some action would cause someone pain, that counts against it. There’s another where the creatures are wired up so as to think that there’s something wrong with sodomy. There’s another where the creatures are wired up to think that non-human animal pain counts for little compared to human pain. There’s yet another where the creatures are very concerned with the welfare of plants. Internally, some of these creatures are similar to you. No matter how strange your views might be you have a counterpart out there somewhere. Your counterpart has similar beliefs, experiences, wants, desires, etc… None of these creatures know how they came to be. They have the same sorts of creation stories we do.
The argument from cosmic coincidence doesn’t rest on the thought that we don’t have moral beliefs, that there aren’t moral facts, or that our beliefs don’t fit the moral facts. Since the argument doesn’t assume these things, let’s stipulate that there are some creatures that have beliefs that fit the facts. We’ve crossed off two of the conditions necessary for knowledge. There are lots of creatures wired for failure, but let’s focus on the creatures wired in such a way that they are the lucky ones that get things right. Just so we’re clear, these creatures aren’t wired this way because someone chose them to get things right. They happen to be the ones who get things right. Given the sheer number of different wirings, the odds were that someone would get things right. We can imagine that the creature doing the wiring doesn’t know which natural properties are the natural properties on which the moral properties supervene. She wanted to cover her bases and make sure that “wherever” the moral properties are found, there’s some group out there that judges that they are where they take them to be. To give this group a name, let’s call them the “Rossians”.
Do the Rossians have moral knowledge? If they don’t, it’s not because they lack true beliefs about moral matters. And, because they’re wired in such a way that they reliably attribute the correct moral properties when they track the relevant non-moral properties, they would seem to satisfy any reliability condition on knowledge. So, if they don’t have knowledge, it’s not for a lack of a reliable basis for their moral judgments. It’s perfectly consistent with everything that we’ve said that their beliefs are sensitive and safe. Given the way that they are wired, in the nearest possible worlds where their moral beliefs wouldn’t be true, they wouldn’t have those beliefs. Given the way that they are wired, if they have a moral belief, that same belief is true in the nearby possible worlds. So, why shouldn’t we say that they have moral knowledge?
One reason might be that nothing we’ve said thus far shows that their beliefs are reasonable. For all we’ve said, they might be in the same epistemic position as the chicken sexers. Of course, you might think that chicken sexers know the sex of the chicks that they sort and so might think that this just provides further support for the claim that they have moral knowledge. If you think chicken sexers don’t have knowledge. This is easily remedied. We can add that the Rossians have good wiring and that when they take the relevant natural properties to be present, it seems intuitive to them that the moral properties are present as well. Surely if they can be wired up to track the right properties, they can be wired up to have the “right” intuitions. Now we can exert some additional pressure on the skeptic. The intuition that underlies the new evil demon objection to reliabilism suggests that if someone is the same on the inside as someone who has knowledge, no matter how bad things are external to her perspective, we can still say that she’s reasonable in her judgments and justified in making them. And now, I think, we’ve effectively silenced the moral skeptic. The moral skeptic wants to say that our moral beliefs aren’t justified and don’t amount to knowledge. If ST1 is off the table, there’s little the moral skeptic can say.
The moral skeptic cannot say that we don’t have moral beliefs. She doesn’t derive her view from non-cognitivism. She cannot say that there are no moral facts for these beliefs to fit. She doesn’t derive her view from moral nihilism. She cannot say that there are moral beliefs for the facts to fit and that none of our beliefs fit the facts. If I judge that giving to charity is either permissible or obligatory and she judges that that’s false, she’s committed to saying both that giving to charity is impermissible and that she doesn’t know that it is. This combination of attitudes constitutes a Moorean absurdity. Such thoughts are deeply irrational. Similarly, she cannot say that our moral beliefs fail to constitute knowledge and fail to be justified for purely Gettierish reasons. For one, she’d have to concede that our beliefs are true and would be committed to the Moorean absurd thought just mentioned. For another, beliefs are justifiably held in Gettier cases, so this wouldn’t matter to assessing the justificatory standing of our moral beliefs. It looks like she’d have to say that we’re not the same on the inside as someone who has moral knowledge and argue for ST1 by arguing for ST2. The trouble she faces is that the thought experiment above suggests that ST1 is false, assuming that our standard accounts of knowledge are approximately correct. In this case, she can only argue that we’re not the same on the inside as someone who has moral knowledge by arguing that our moral beliefs couldn’t be true. Again, the skeptic cannot argue that our moral beliefs are mistaken without committing herself to the Moorean absurd thought that something is impermissible and nobody knows that it is.
It’s at this point, however, that the skeptic would remind us that we haven’t addressed the argument from cosmic coincidence. Yes, the Rossians have good wiring and it’s true that if they take their intuitions at face value, they get things right as a rule. But, doesn’t this miss the point? The creatures I’ve described get things wrong as a rule if they take their intuitions at face value. It’s only in the very rare case that someone gets things right as it’s only in the rarest case that someone is wired up in the way that the Rossians are. So, as noted above, it looks like the Rossians are lucky to get things right and this sort of luck precludes knowledge. If it does, then even if the intuition that underlies the new evil demon objection to reliabilism is correct, we have no reason to say that those who are the same on the inside as the Rossians have justified moral judgments. We have no reason to think that they’re the same on the inside as someone whose moral judgments constitute knowledge.
The skeptic’s argument is too crude if it rests on the thought that epistemic luck precludes knowledge. Some luck i
s malignant, but some is benign. Let’s contrast two kinds of epistemic luck. First, there is veritic epistemic luck. In cases of veritic luck, the subject is lucky, so her belief is true, but she’s lucky, so it easily could have been false. To sharpen this up just a bit, it’s lucky that the subject gets it right given what her evidence is. Second, there is evidential epistemic luck. In cases of evidential luck, the subject is lucky, so her evidence is good, but she’s lucky, so it easily could have been that she had worse evidence. Evidential luck is thought to be benign because it’s not lucky that the subject has a true belief given her good evidence. Veritic luck is thought to be malignant because it is lucky that the subject gets it right given her poor evidence.
If the case of the Rossians is a case of veritic luck, the anti-skeptical strategy outlined here is a failure. If, however, the case of the Rossians is a case of evidential luck, the anti-skeptical strategy looks pretty good. While I don’t know if I can make a compelling case for the claim that the case we’re dealing with is a case of evidential luck, I also don’t think the skeptic can make a compelling case for the claim that it’s a case of veritic luck. The intuition that underlies the argument from cosmic coincidence is a kind of anti-luck intuition. If the skeptic cannot show that the luck at issue is malignant, we’re at an impasse.
The problem is this. Suppose we take the Rossians and all the other creatures and say that their evidence for their moral beliefs differs in content, not kind. By this I mean the evidence that they (i.e., the Rossians and the rest) have for their moral beliefs is basically this: it seems that some feature of the situation calls for a certain sort of response or counts against a certain sort of response. This evidence would consist of propositions about how things seem to them and given only that evidence. Given only this evidence, the Rossians are lucky to get things right. Suppose instead that we say that the evidence they have for their moral beliefs differs both in content and in kind. By that I mean the evidence that the Rossians have includes the propositions about how things seem and the facts that certain features call for an action or speak against an action. The Rossians wouldn’t be lucky to get things right given that their beliefs were based on this sort of evidence. The rest of the creatures wouldn’t have these facts at their disposal because their moral beliefs and intuition don’t fit the moral facts. But, if the Rossians enjoy an epistemic advantage over the others, there’s no reason to think the Rossians’ beliefs are veritically lucky.
What evidence do the Rossians have for their moral beliefs? The intuitionist should say that their evidence includes both psychological facts (e.g., that it seems that such and such a feature counts for or counts against) and normative facts (e.g., that such and such a feature counts for or counts against). The intuitionist should also say that this second kind of evidence isn’t available to the others. Elsewhere I've argued that your evidence includes the proposition that p if you know p non-inferentially. Non-inferential knowledge is all you need to have p as a reason for belief. I didn’t say what it takes for p to be a reason for belief. There’s currently some controversy as to whether false propositions can constitute evidence or reasons to believe. (There’s no question that they can be treated as if they are evidence or reasons to believe, but that’s not the same thing.) If I’m right and your evidence includes anything you know non-inferentially, the intuitionist should say that the Rossians have normative propositions as part of their evidence. If I’m right and only true propositions can constitute evidence, the Rossians’ evidence differs in kind from the evidence the other creatures have for their intuitions misrepresent the moral domain and their moral beliefs are mistaken.
If evidence consists of facts or true propositions and your evidence will include any propositions you know non-inferentially, it’s a mistake to say that the Rossians are lucky to get things right given what their evidence is if that’s based on the claim that the Rossians and the rest of the creatures are in roughly the same evidential position (i.e., that their evidence differs in content, not kind). It seems to me that there are two promising lines of argument for the claim that false propositions do not constitute evidence. First, there’s the linguistic evidence that suggests that evidence ascriptions are factive. As Unger noted long ago, the following remarks are clearly defective:
(9) What was his reason for thinking that he was out of milk? It was that his fridge was empty. Of course, he didn’t know that the fridge was empty.
Those who don’t think evidence ascriptions are factive have to offer some explanation as to why (9) seems defective. If something can be your reason for believing even if it’s not true, something can be your reason for believing even if you don’t know it’s true. Why then does (9) seem contradictory? We can strengthen the case for the factivity of evidence and reasons ascriptions. Those who deny that such ascriptions are factive will have to offer some explanation as to why (9) seems defective even if it’s in perfectly good order. The explanation will have to say that there’s something weaker than entailment that holds between the reason-ascription and the further claim that the proposition ascribed by the that-clause is true. One way to test to see if a connection is weaker than entailment is by considering the reinforcement data. You can properly reinforce information that is merely pragmatically implied, but not information that is entailed. If you try to reinforce an obvious entailment, the result is a statement that seems defective, a redundant conjunction:
(10) I have a dog. I have just one dog.
(11) He knows he has a dog. Indeed, he believes he has a dog.
If (9) weren’t a contradiction and it didn’t follow from the fact that his reason was that he was out of milk that he was out of milk, then this should seem felicitous:
(12) His reason for believing that he was out of milk was that his fridge was empty. Not only that, his fridge was empty.
Intuitively, it seems (12) is a redundant conjunction along the lines of (11).
The second line of argument focuses on the relation between evidence and explanation. Our evidence or our reasons for belief can figure in explanations in two ways. When we know that p is part of our evidence, we know that so long as p is not a brute fact, there’s some explanation as to why p. We also know that if p is part of our evidence, p explains certain support facts. It explains, for example, why it’s likely that q if, say, the probability of q on p is high. Whether a piece of evidence figures in an explanation as the explanans or the explanandum, since we know that only facts figure in (correct) explanations, only facts constitute evidence. Views that deny that evidence consists of true propositions cannot account for these connections between evidence and explanation.
With this in place, we can now see why the intuitionists ought to say that the Rossians’ have evidence that differs both in content and in kind from the evidence that the other creatures might have for their moral beliefs. It differs in content because they attribute moral properties in different situations than the other creatures do. It
differs in kind because the Rossians have moral propositions as part of their evidence and the others do not. So, the intuitionists should say that the case described is a case of evidential luck rather than veritic luck, in which case the skeptical argument isn’t all that threatening. I can anticipate two objections to the view developed here. The first is that the intuitionists don’t have anything good to say about cases of error. It’s a consequence of this view that the Rossians are the only subjects that have moral propositions as part of their evidence, so how can the intuitionist say that the rest of these subjects are justified in their beliefs? The second is that the account I’ve described is only available to the naturalists.
Let’s think about some of the creatures that get things reliably wrong. These poor subjects attribute moral properties when they take certain natural properties to be present when and only when it’s incorrect for them to attribute these moral properties. To give them a name, let’s call them the “Randians”. Intuitively, the Randians are just as reasonable and just as rational as the Rossians since they both form their moral beliefs by taking their moral intuitions at face value, they are internally coherent, they reason just as carefully, etc… It’s true that the Randians get things wrong as a rule, but they are no worse off than those systematically deceived by a Cartesian demon and the demon’s dupes count as rational in their beliefs. I don’t think it’s difficult for the intuitionist to accommodate the intuition that the Randians are rational. While their beliefs aren’t based on (genuine) evidence, this isn’t due to a failure on their part. They count as rational, in part, because they respond in the way that they should have responded if the evidence had been the sort of evidence they took themselves to have. What about the intuition that their beliefs are justified? The intuitionists might go in one of two directions here. There’s a difference between saying that something is rational and saying that it’s justified. If someone’s actions or beliefs are justified, it’s not true that they should have been otherwise. If someone was rational or reasonable, it doesn’t follow from the fact that they should have done things otherwise that they are anything less than perfectly virtuous. If the intuitionist wants to explain why the Randian’s beliefs are just as justified as the Rossians, they can say that the justification of a belief doesn’t depend upon whether it’s based on evidence but whether it’s formed in such a way that the believer formed the beliefs she should if her evidence was what she took it to be. Myself, I’m not inclined to say that the Randians’ beliefs are just as justified as the Rossians. Surely the Rossians’ behavior is better justified than the Randians’ behavior. The Rossians do what they’re obliged to do and the Randians don’t. If, however, the Randians really believed what they ought to have believed, it seems that they’d be justified in acting as they judged that they ought to act. Since they’re not, I’d rather classify both their actions and their attitudes as excusable at best.
There is a second worry that arises for the intuitionist. The intuitionist view has to juggle two commitments. The first is that it’s possible for some subjects to have moral facts as part of their evidence. The second is that these facts are not natural facts. You’ll recall that the argument from cosmic coincidence assumed that moral intuitions are physical states or events. If we assume that that’s so and we say that intuitions provide us with our reasons or our evidence for our basic moral beliefs, how can we also say that our evidence for these beliefs includes moral facts?
To deal with this worry, it’s important to stress that intuitions can provide reasons even if the intuitions aren’t themselves the reasons and aren’t constituted in part by those reasons. On the account of evidence defended earlier, if you know p non-inferentially and p is a reason, p is a reason you have that you can rely on in your reasoning. To have a reason is for the thing you have to be a reason and for you to have the right to treat it as such. If you think intuitions or experiences by virtue of which it seems to you that p justify believing p to be true, these give you the right to treat p as a reason. Whether p is a reason depends upon whether p is true. There’s no obvious inconsistency here in saying that the intuition is not part of the non-natural order but gives you a right to reason from premises that are true by virtue of how things stand in the non-natural order. (I suspect the worry here is similar to worries that McDowell has about attempts to make sense of perceptual knowledge without the resources of disjunctivism. I've tried to address McDowell's epistemological argument for disjunctivism here if anyone is interested.)
Let's take stock. I've tried to show that the argument from cosmic coincidence doesn't give us a good reason to reject the combination of intuitionism and non-naturalism. What we need to bolster that argument is an additional argument that would show that the evidence we have if we're in the position of the Rossians doesn't differ in kind from the evidence we have if we're in the position of the Randians. I cannot think of any good reason to think that our evidence for our moral beliefs would have to be limited to non-moral facts. Notice that I haven't said that there aren't ways of leveling the playing field by taking evidence away from the Rossians. You might think that if the Rossians are exposed to the Randians the experience of interacting with others with radically different moral views might defeat the justification they otherwise might have had for their moral beliefs. Intuitionists don't have to deny that this is possible since their focus is on the justification intuition provides in the absence of defeaters. I've been interested in an argument that seems to show that there's an important difference between moral and non-moral beliefs by virtue of the fact that one concerns natural properties and the other doesn't. Since disagreement cases can pose the same threat to non-moral beliefs as moral beliefs, these cases don't point to any principled difference between, say, moral belief based on intuition and non-moral belief based on observation.