Welcome to what we expect will be an interesting and productive discussion on Preston’s Werner‘s “Moral Perception without (Prior) Moral Knowledge” (which the Journal of Moral Philosophy has generously provided free access to until the end of November). David Faraci has kindly agreed to contribute a critical précis, and it appears immediately below. Please join in the discussion!
Critical Précis by David Faraci:
Thanks to PEA Soup and to the Journal of Moral Philosophy for offering me this opportunity to discuss Preston Werner’s excellent “Moral Perception without (Prior) Moral Knowledge.”
Preston’s paper is, in large part, a response to my “A Hard Look at Moral Perception” (Philosophical Studies 2015), and I want to thank him for taking the time to trudge through that paper. I’ve written a longer reply to Preston, which is also forthcoming in JMP. What I’ll say here is a mix of things I say there and some new stuff.
In my 2015, I argued that if there is moral perception (which I grant from here out), it relies on background moral knowledge that is non-perceptual, and thus there can be no “purely perceptual” moral epistemology. Moral experiences, I argued, are always grounded in non-moral experiences (at least in epistemically good cases). For instance, in Harman’s famous cat-torture case, our experience of the wrongness of the cat-torture depends on our experience of its non-moral features (the cat’s yowling, etc.). Preston largely accepts this, but in case you’re dubious, just notice that if I substituted some convincing, fake cat-torture for the real cat-torture, you’d still have an experience of wrongness, even though there wouldn’t be any. The best explanation for this, I claim, is that your moral experience is grounded in your non-moral one.
So, you experience the wrongness of the cat-torture because you experience its non-moral properties. Arguably, in order to know Q by inference from P, you need to know that P implies Q. In this case, that means you need to know that the yowling (etc.) implies wrongness. Could you know that by perception? Perhaps, but since that would be moral perception, and all moral perception is grounded in non-moral perception, you’d get a regress. At some level, your perceptual moral knowledge depends on non-perceptual moral knowledge.
Ok, that’s nice and quick and there are lots of caveats and objections. But we should probably focus on Preston’s worry. Preston argues that this picture over-intellectualizes perception. In many cases, he thinks, one experience is grounded in another, but we can’t be required to know the relevant implication, because we don’t even believe it. Here’s a motivating example:
I can, typically, effortlessly distinguish the sound of a piano from that of an acoustic guitar. But I couldn’t even begin to explain this difference or point to the low level qualities of tone and timbre that ground their differences. I have no beliefs, much less knowledge, of how I go from low level auditory information to the auditory experience of a piano. (9)
Preston thinks he doesn’t need to know that <low level sounds> imply <guitar sound> because he doesn’t even believe it. All that is required, he thinks, is that there be some “subdoxastic information states which ground reliable transitions from” one experience to the other (7). Analogously, he can grant that moral experiences are grounded in non-moral experiences, but maintain that the epistemological merit of the former depends only on the existence of such reliable subdoxastic states.
When I first read Preston’s paper, my initial reaction was to think that I hadn’t over-intellectualized perception, but rather that Preston has over-intellectualized belief and/or inference. I was tempted by the view that what Preston is calling “subdoxastic information states” just are beliefs, albeit implicit ones that we are perhaps typically unable to put into words, and that in order for Preston to know that he’s hearing a guitar, his implicit belief that <low level sounds> imply <guitar sound> must still constitute knowledge. I realized, however, that this would involve sticking my neck out more than I need to, given my dialectical interests. Writing this précis has further convinced me that I need to think more about whether I should maintain that position even silently in my heart, more on which later.
To get at what I’m more than just tempted to say, consider a passage at the close of the section in which Preston offers his objection:
The opponent of a purely perceptualist view shouldn’t be satisfied with C* [the claim that moral perception can be reliably mediated by subdoxastic states]. The alleged problem for the pure perceptualist is that moral perceptions, when they are epistemically successful, can only ground moral knowledge because they depend on justified moral beliefs. . . . If the relevant mediating states are subdoxastic states, they won’t be the sorts of things that require . . . any propositional knowledge at all . . . to generate successful moral experiences. C* is the strongest conclusion we can draw from Faraci’s argument . . . The purely perceptualist view is left standing. (10)
I think I can accept all of this, except the claim that I shouldn’t be satisfied with C*. Here it will be useful to say something about my dialectical focus. It is notoriously difficult for realists to account for even the possibility of moral knowledge. In large part, this is because realists have tended to be intuitionists, and it is notoriously difficult to account for the reliability of intuition, to explain how intuition can grant us “access” to moral truth. I suspect this is a large part of the motivation for the current trend towards perceptualism in moral epistemology. The hope is that realists can bypass these worries by appealing to something whose epistemic credentials are less dubious.
The mistake in my 2015—one I thank Preston for illuminating—was running together questions about reliability with questions about justification. What I should have argued is only that moral perception doesn’t help realists account for the reliability of our moral beliefs. This is because saying we know something “by perception” isn’t itself an answer to questions about reliability. Most knowledge by perception seems immune to such worries, but that’s just because the story about perceptual reliability is comparatively simple in many cases (which is not to say it is simple full stop), appealing to things like causal connections between our experiences and the bits of the world they represent. By contrast, there is no obvious explanation for how our perceptual systems learn to get the right moral experiences out of the right non-moral ones. Indeed, finding that explanation seems daunting in ways that directly parallel standard challenges to explain the reliability of moral intuition.
To help illustrate this, consider a suggestion Preston makes near the end of the paper, when he anticipates the worry that we still need to account for the reliability of the relevant experiential transitions. He suggests that “perhaps the perceptual system alone has evolved to represent things in this way, in order to facilitate quick action” (14). Notice two things. First, this sort of explanation is frequently used to challenge moral knowledge. Indeed, the claim that our moral judgments have evolutionary origins stands at arguably the current most popular epistemological objection to realism. Second, this explanation does not seem distinctively perceptualist-friendly. If evolution does allow us to track the moral truth, that seems just as good an answer for intuitionists as for perceptualists!
In conversation, Preston has suggested that he agrees with most if not all of what I’ve said so far. My sense is that that’s because his concern is not with explaining reliability, but with the idea that we can appeal to moral experience as a basic source of justification for our moral beliefs.
This comes out nicely in his discussion of access internalism. Preston notes that someone might worry that all his talk of “reliable processes” means he is wedded to externalism about knowledge. In response, he points out that his argument is perfectly consistent with the claim “that in order to know P, an agent’s grounds for believing P must be accessible to her via reflection” (11). In perceiving the cat-torture as wrong, your grounds for belief—what you appeal to as a basic source of justification—are perfectly accessible to you: they are just your experiences! To demand that you further have access to the full explanation for the reliability of those experiences would be overkill. I agree with this (though as Preston notes, some things I said in my 2015 might suggest otherwise), with the single caveat that I think evidence that your belief could not be reliable is a defeater for that justification.
This brings me back to the question of how important it is, given Preston’s interests, that the connection between moral and non-moral experiences be a matter of non-inferential transitions rather than inferences, and subdoxastic information states rather than beliefs. We can see why this might seem important by noticing that the following set of claims is inconsistent:
- Moral experiences are a basic source of justification.
- Moral experiences are grounded in inferences from non-moral experiences, which are themselves grounded in beliefs about implications.
- Inferences and beliefs are subject to justification.
- Something can only be a basic source of justification if it is not grounded in anything else that is subject to justification.
I take it Preston wants to endorse (1). In argued for (2) in my 2015. (3) is pretty standard stuff. And it might be hard to see what it means for a source of justification to be basic unless (4) is true. One way out, echoing Preston, is to argue for:
(2′) Moral experiences are grounded in inferences or non-inferential transitions from non-moral experiences, which are themselves grounded in beliefs in implications or reliable subdoxastic states.
Before closing, let me gesture at a half-baked alternative. For a source of justification to be basic might just mean that it is a source that offers a kind of default justification, such that we can appeal to that source to close questioning about the grounds of our beliefs. But if that’s what we’re interested in, is it really necessary that basic sources of justification also be fundamental ones—ones whose justificatory force is independent of anything else that needs justification? I think perhaps not. Perhaps experience is a basic source of justification because it is generally reliable, and that’s why I can appeal to moral experience to justify my moral beliefs—at least until I gain too much evidence that my moral experiences in particular are not reliable. If that’s the case, then (4) is too strong, and that provides another way out of the inconsistency.
This may not ultimately be the way to go, but I do think it has something going for it. It seems to me that it would be quite surprising if anything of deep importance turns on whether the states that explain the move from non-moral to moral experiences count as “beliefs,” or whether those moves themselves count as “inferences,” and are therefore subject to justification. But, again, just a hunch.
Hopefully I’m understanding Preston’s (and my own) thoughts correctly here. If so, to close I’d just like to invite him to say more about his motivating views. What are the advantages of perceptualism over intuitionism when it comes to identifying basic sources of justification for moral beliefs? That is, if we grant that perceptualism doesn’t help explain the reliability of our moral beliefs, why is it nevertheless important that we appeal to experience rather than intuition when we seek to justify those beliefs? I look forward to seeing what Preston has to say to these and to everyone else’s questions, both here and in his future work.
First of all, let me thank PEA Soup for hosting this discussion, as well as David Faraci for his excellent précis, as well as for a lot of helpful discussions we’ve had about these issues for the last few years. David makes a bunch of good points in his discussion of my paper. I’ll try to respond to them relatively briefly, so as to make this a reasonably sized blog comment.
Over-intellectualizing beliefs ?
Does a purely perceptual view help with deep epistemic objections (e.g. debunking) to moral realism?
Relatedly–an argument to the effect that we couldn’t be reliable would be a defeater for perception as a basic source of moral justification.
Advantages of perceptualism, given that it can’t solve debunking stuff.
Why pure perceptualism?
Let me begin by addressing David’s last point. Assuming pure perceptualism can’t make progress on the deep epistemic problems for moral realists, such as Streetian debunking arguments and Benacerraf-type worries (though see the end of this comment), why should we think it represents any sort of improvement over traditional a priori moral epistemology, like intuitionism? Rather than get bogged down in the details here, let me just programmatically list a few reasons why one might like a pure perceptualism. Obviously, everything I list is going to be contentious and nuanced in ways I won’t go into (unless people want to challenge me!):
A. If you’re generally skeptical of synthetic a priori knowledge for whatever reason, but you want to hold on to moral realism, or some other objective-y view of normative truths, pure perceptualism can help.
B. If you think perceptual experience has a default justificatory status that intuitions do not, then you’ll think intuitionism faces an epistemic burden that pure perceptualism does not.
C. Pure perceptualism seems to fit better with certain features of standpoint epistemology than intuitionism. (I think this is an important feature, though it hasn’t been noted in print anywhere as far as I know–and anyway, I don’t have the argument for this claim fully worked out either.)
D. Pure perceptualism may provide a better explanation of moral progress than intuitionist views, at least given auxiliary assumptions. (I’m thinking here of an idea somewhat like the case made in Sarah McGrath’s “Moral Knowledge of Perception”, suitably generalized.)
E. If you’re attracted to a bottom-up methodology for normative theorizing, pure perceptualism looks more attractive than intuitionism, at least prima facie. Intuitionism, at least on its face, looks like it fits better with a principle-first normative theory. (Note this is distinct from saying that pure perceptualism entails particularism. I want to hold on to my principles! I just have a different story about the structure of their justification.)
F. Doesn’t it just seem like a cool view?
2. Over-intellectualizing beliefs?
The key to my response to David’s paper is appealing to the role of what I call “subdoxastic information states”. I claim that David’s objection to a “pure perceptualist” view assumes that only beliefs could play the role of transitioning from low-level perceptual information to the representation of more complex properties in perceptual experience. Think of the difference between the sound of a piano and the sound of a guitar. Those sounds are constituted out of lower level auditory properties of tone and timbre, but–I claim–we don’t have beliefs that ‘justify’ the inference from these low-level properties to the experience as of a guitar or as of a piano.
David tentatively suggests that this may rely on an over-intellectualized notion of belief, and that we should count as having implicit beliefs that justify these sorts of transitions. As I see things–and I don’t think that David would disagree–what is of relevance here is that these information states, be they doxastic or not, are subject to (normative) epistemic assessment. We can certainly ask whether these states are reliable, but can the states be justified or unjustified? I think this is a contentious issue amongst epistemologists. My inclination, though, is to say that these information states don’t count as properly agential, and so are not subject to questions about justification in the way that explicit beliefs are. I think it would be odd to epistemically criticize someone for experiencing something as sounding like a piano that isn’t a piano (assuming normal circumstances).
Of course, this is all assuming there are no defeaters hanging around to make us skeptical of these information states in some particular circumstance. And we might think that we have defeaters in the moral case, unlike in mundane cases like the piano case. That takes us to Faraci’s central worry for pure perceptualism:
3. No progress on the possibility of reliability
Ok, so David’s biggest worry for the pure perceptualist view is that it makes no progress on the deepest epistemic problem(s) for moral realism–of showing how we could reliably track moral truths, on a realist understanding. This worry gets expressed in a bunch of different ways by different people. But the basic idea is that the realist needs to explain how we could reliably form moral beliefs, and pure perceptualism fares no better than intuitionism (or other a priori views) on this count.
It’s true, I think, as David says, that the pure perceptualist has no easy solution to these kinds of worries. (I once did, but that was in my dissertation, so thankfully no one has to know.) So let me just say a few ways that I think the pure perceptualist does better than traditional a priori views, even if we’re not out of the woods yet.
First, I think pure perceptualists can explain how we can in principle have access to the moral properties, through perceptual awareness. Supposing we’re reliable, the pure perceptualist can note that this reliability is not coincidental in one important sense, which is that our beliefs are responsive to our perceptual awareness of the moral properties. Intuitionists can make similar claims, but I think they rely on much more contentious assumptions about our access to some domain of abstract properties. There is some futzing with how the perceptual awareness of moral properties is supposed to be possible, but I think there are good reasons to think it is.
Second, and closely related, I think that pure perceptualists can meet an explanationist kind of requirement on knowledge that the intuitionist can’t. One might think that a condition on knowledge is that part of the best/good/normal explanation for the belief appeals to the fact in question. Suppose Emily sees the cat being lit on fire and forms the belief that that is wrong. The pure perceptualist says that her belief is formed on the basis of her seeing (in the factive sense) that it is wrong. The intuitionist says that she has an intuition on the basis of her seeing a bunch of non-normative properties. But is the intuition explained in terms of the fact that it was wrong? It’s less clear (though not impossible) to see how her intuition would be explained by the fact that it was wrong, since the intuition isn’t naturally seen as dependent on anything outside of her head. (Re-reading this paragraph, I don’t think it will convince anyone not already convinced by my view. But let me leave it at that unless others raise specific objections.)
Thanks again to David and PEA Soup, and I’m looking forward to the rest of the discussion!
To be slightly self-indulgent, let me just say real briefly how I ought to think realists should confront debunking and Benacerraf-type problems, of the sort that David raised above.
Unlike a lot of other realists, I don’t think there is going to be a nice, neat, all-encompassing strategy for addressing these kinds of concerns. Rather, I think these kinds of objections arise in a number of ways, and some combination of the following approaches have to be taken, in accordance with the specific worry:
1. Point out that the argument overgeneralizes. (Vavova 2014 manifests this strategy wonderfully.)
2. Point out that the argument actually doesn’t have the force against realism that it seems to, once we look more closely.
3. Argue that, with the right positive theory of the mechanisms of moral knowledge, we can rebut the challenge. (It’s the ‘access’ and ‘explanationist’ understandings of the challenge that I respond to above where I think this third strategy is appropriate.)
Say you are looking out on a scene with two stationary objects A and B. You turn your head rightward, then leftward, etc. As you turn your head A and B occupy different positions in your visual field: they move leftward, then rightward, etc. Say A’s motion is faster than B’s motion. Then, ignoring other cues for simplicity, A will look to be closer to you than B. This is depth perception via motion parallax. Consider, then, two propositions suggested by the foregoing:
(1) A’s motion is faster than B’s motion
(2) A is closer than B
In the sense used in the current exchange between David and Preston your perceptual knowledge of (2) is grounded in your perceptual knowledge of (1). It seems to me it might be helpful to use (1) and (2) as a probe for exploring the epistemological issues under consideration.
I doubt that your perceptual knowledge of (2) is epistemically based on your perceptual knowledge of (1). The kind of grounding just noted need not be epistemic basing. I think Preston would also say something along these lines, and David seems to be conceding this for now. But there’s a new question David raises: how can we explain the reliability of the psychological mechanisms that generate content (2) from content (1)? In this case the answer would appeal to some geometrical facts.
Now transition to moral perception. Say we have these propositions:
(3) Those kids are making the cat suffer
(4) Those kids are doing something wrong
And say you have a perceptual representation of (4) and, as both Preston and David think, it is grounded in your perceptual representation of (3). We could also look deeper than (3) at propositions about the kid’s movements and the cat’s expressions, and maybe there are reasons to do so, but I won’t here. Two points. First, just as your perceptual knowledge of (2) need not be epistemically based on your perceptual knowledge of (1), for all that has been conceded by the talk of grounding, it might be that your perceptual knowledge of (4) is not epistemically based on your perceptual knowledge of (3). I’m skeptical about immediate perceptual knowledge of (4), but I think one needs to wheel in additional considerations to support such skepticism. Second, if we consider the new question—how can we explain the reliability of the psychological mechanisms that generate content (4) from content (3)?—then just as we had to appeal to geometrical facts before, now I’d think we we have to appeal to moral facts. One question I have is whether this is supposed to be a problem. I don’t really see that, not if what we are trying to explain is reliability. If there’s a lingering sense of worry then I think it is about something other than explaining reliability, for example explaining how perception might put you in contact with the fact that makes (4) true, or something like that.
I think this is a really great exchange between Preston and David so far, and I’m excited to see that there now is a “literature” on this pure perception view that you are both discussing.
I think there is an ambiguity in the term “grounding” here.
Metaphysical Grounding = You cannot experience moral properties on the basis of perception without experiencing non-moral properties on the basis of perception.
Epistemic Grounding = Perceptual evidence for the moral status of particular actions will always evidentially depend on perception of non-moral properties and an inference from some moral principle.
Sometimes the conversation seems to slide between these two distinct kinds of grounding.
For example the cat example supports the metaphysical grounding claim, but it’s unclear to me how it would support the Epistemic Grounding claim.
However, after giving the cat example – David says:
“So, you experience the wrongness of the cat-torture because you experience its non-moral properties. Arguably, in order to know Q by inference from P, you need to know that P implies Q.”
The first sentence is clearly true if we’re thinking Metaphysical Grounding…but the next sentence seems to assume the grounding relation is the epistemic kind.
Forgive me if this has been discussed and I missed it, but how do we get from the obviously true metaphysical claim to the epistemic claim?
Hi Eli,
Thanks for this! As I’m reading it, your comment is mostly friendly to me, right?
I agree with most everything you say about the case of (2) and (1). The grounding in question is not a matter of epistemic basing. However, I may be misunderstanding what you mean when you say “how can we explain the reliability of the psychological mechanisms that generate content (2) from content (1)? In this case the answer would appeal to some geometrical facts.”
Can you say a bit more about how you’re envisioning the geometrical facts as explaining the reliability on their own? Surely we would need to also explain how the visual system (in this case) works such that it is responsive to those facts, right?
Presumably we could give such an explanation, either at the level of some individual agent, or at the level of evolutionary history, for the (2) from (1) case.
As I see (what I’m calling) Faraci’s ‘deep’ epistemic worry, there won’t be an analogous explanation of reliability in the case of (4) and (3).
And perhaps I’m misunderstanding David here (please feel free to correct), but as I’m reading him, he thinks that the lack of an explanation of reliability of the move from (3) to (4) serves as a defeater for forming beliefs on the basis of (4).
All that said, I think I’m inclined to agree with the general spirit of your comment (But feel free to tell me if I’m missing something). The skeptic asks for an explanation of reliability. If I get to assume reliability and just provide an explanation, I think the pure perceptualist can do that. If we don’t get to assume reliability, then I think the pure perceptualist can’t meet the challenge, but the challenge will overgeneralize anyway.
Hi Andrew,
First, let me digress and say that your moral perception paper was one of the first ones I read when I was beginning work on my dissertation, and stimulated a bunch of my thought on these issues, so thank you!
As for your question about grounding–I don’t think you’ve missed anything, though I have been thinking about this quite a bit because in an earlier draft of a different paper I’m working on, I conflated these issues until I was straightened out by my colleague David Enoch.
Certainly what is relevant for the debate at hand is what you call epistemic grounding:
“Epistemic Grounding = Perceptual evidence for the moral status of particular actions will always evidentially depend on perception of non-moral properties and an inference from some moral principle.”
I definitely reject this. It’s less clear to me what David thinks now, but he used to accept it (see, for example 2.1 of his 2014 paper). But perhaps you’re worried even about a weakened version of the principle:
Weak Epistemic Grounding = Perceptual evidence for the moral status of particular actions will always evidentially depend on perception of non-moral properties along with some other information states.
I am less sure what to say about this principle. (I don’t think anything I say depends on accepting or rejecting it, right?) One reason it strikes me as plausible is this: If we learn that we were mistaken about the presence of the non-moral properties that metaphysically ground the moral properties, we would think we had lost evidence for the presence of the moral properties as well. For example, if Emily finds out that she was looking at a cat facade, she would not just give up her beliefs about the presence of a burning cat, but also her beliefs about the presence of badness.
I don’t think that’s conclusive proof of Weak Epistemic Grounding, but it at least seems plausible enough to assume unless we have counterexamples in hand.
With regards to Eli’s comment, I’m happy to just nod along with the view Preston attributes to me.
Andrew, I’ll echo that you’re not missing anything. In my 2015, I was pretty careful (I think!) to separate those out; I was playing a bit faster and looser here to keep the length down. Just to be clear about what my view is: In epistemically good cases, I think that moral experiences are psychologically grounded in non-moral experiences. (I write “psychologically” instead of “metaphysically” because I think we should leave it open that the relation is causal, but lots of people take metaphysical grounding to exclude causes.) I say “in epistemically good cases” because I don’t see any reason to think it’s psychologically impossible to (say) just have a random experience of wrongness. But I do think that all plausible cases of successful moral perception involve such psychological grounding. My evidence for this is the convincing fake test stuff; also, denying this would arguably require direct, independent contact with moral properties.
Turning to the quotation you mentioned:
You’re absolutely right that the first claim is about psychological grounding. Then the idea was supposed to be (following my 2015) that this entails inference in a very broad sense that includes subpersonal processing; you or your perceptual system “infers” the moral experience from the non-moral one and that “inference” has to meet certain epistemic standards. The trouble, which I think Preston nicely illuminates, is that while this may be true about standards of reliability, I didn’t pay enough attention to standards of justification. I still think that if one experience is psychologically grounded in another, then whether the former experience generates knowledge depends on the reliability of the experiential transition (one kind of epistemic grounding), but I no longer think (or at least am willing to question) that forming beliefs on the basis of the former experience is justified only if one is justified in believing the relevant experiential implication (another kind of epistemic grounding).
As long as I’m here, let me say something brief about Preston’s comments, as well. First, thanks, Preston, for offering your A-F about the appeals of perceptualism. Those are all really interesting, and I need to think about them a lot more. One thing they definitely suggests to me is that I’ve been thinking about intuitionism as a broader class of views than you, because I never meant to assume intuitions would be general or principle-first. I think this raises some interesting questions about how to draw the line between intuition and perception, again something I need to think about more.
Finally, let me say a little about why I think we should be more worried about reliability than Preston is. I don’t think the 1-3 Preston appeals to in his second comment are very promising; I don’t agree that the problem overgeneralizes or that it has less force against realism than it seems (is this Clarke-Doaney stuff, Preston?). If anyone is interested, I discuss these issues here (generally) and here (about non-naturalism).
So, that brings us to the idea that perceptualism has an advantage where explanationism is concerned. I just want to register some doubts here. I take it perceptualism could offer an advantage in two ways. First, if explanationism is true, but the a priori is incompatible with explanationism, then obviously perceptualism wins out. But unless they’re skeptical of the (synthetic) a priori generally, explanationists are just going to give an account of their compatibility (see, e.g., Bengson’s “Grasping the Third Realm”).
Alternatively, perceptualism could enjoy some sort of narrow advantage, where “perceptual awareness” can account for explanatory connections better than whatever explanationist-friendly account we give of the a priori. If that’s what Preston has in mind, I’d like to hear more about why he thinks this. Personally, I’m inclined to think we’re not yet in the position to say anything about this one way or the other. The question of how moral facts/properties can explain anything psychological is a foundational problem in metaethics, one that seems prior to the debate between intuitionists and perceptualists. Until we’ve answered that question, why think the answer will support perceptualism? This was the point of some of what I said in précis about the suggestions Preston makes about (e.g.) evolutionary explanations. It seems unclear both (a) how evolution could grant us to access to the moral truth and (b) how, if it does, this fits better with perceptualism than intuitionism.
Hi David,
Thanks again for your comments, and for making me think more carefully about this stuff. Just a couple of brief things in response.
First, I should clarify what I said about intuitionism. I think I mistakenly implied that intuitionism must be a ‘principle-first’ sort of view. I don’t think intuitionists per se are committed to that sort of claim. I’m thinking of the intuitionist as being committed to two claims:
Foundationalism about moral justification
Some a priori source of justification for the foundationally justified moral beliefs.
The reason I think the pure perceptualist more naturally fits better with a bottom-up story about moral epistemology than intuitionism is simple to state, but probably harder to flesh out in a convincing way. But the basic idea is that, intuitionist stories about the mechanisms of epistemic access–self-evidence, conceptual competence, ‘intellectual perception’–look to me like the sorts of things that would be responsive, if at all, to general truths, rather than particular truths. Why would intuitions track particular moral facts rather than general ones, given these mechanisms? And if they can track both, as I think some theories would imply, why not just directly intuit the general principles, so we can finish normative ethics in a few short steps?
Obviously, I’m not really giving an argument here. I’m just sort of gesturing at the sort of argument I think could be given. But that’s the general idea.
On reliability stuff. Rather than rehashing what has been said more convincingly by others, let me just point in their direction, before saying a tiny bit to follow up.
On the overgeneralizing stuff, as I said, I think Vavova (2014) does a great job at pointing out that if the skeptic is asking for a positive reason to trust our basic source of justification, it will entail skepticism about virtually everything, including all perceptual belief. (Our explanation of the reliability of perception is going to essentially appeal to evidence garnered by perception.)
As for having less force–I think that there are some ways that the reliability worry gets raised that make it a trivial matter to be met. I suppose Clarke-Doane’s stuff falls into this camp, as does a nice recent paper by Dan Baras (“Our Reliability is in Principle Explainable”). Clarke-Doane & Baras differ in subtleties, but as you know the basic strategy is to show that safety and sensitivity can be relatively easily met for many moral beliefs, at least if we read safety and sensitivity as modal conditions. It basically relies on (a) arguing that sensitivity is trivially met since the moral facts are necessary, and (b) arguing that safety is met, since we couldn’t have easily evolved to have different moral beliefs (or ‘proto-beliefs’).
Insofar as skeptical worries are raised in terms of safety and sensitivity conditions, I think the Clarke-Doane style approach might work. (I’m not sure about (b), to be honest. But I’m not convinced that it won’t work.) But, if I’m being honest, I think–contra Clarke-Doane–that this reveals that this is not the correct way to understand the spirit of the epistemic challenge. (Baras, for his part, sort of acknowledges that this style of move feels “fishy”.) Instead, I think we should understand safety and sensitivity as hyperintensional, in something along the lines of Matt Bedke’s “No Coincidence” paper (in Oxford Studies in Metaethics).
Anyway, all this is just to say that, on some readings of the skeptical challenge, such as the way Clarke-Doane understands it, it can be dismissed as ineffective. (In defense of Clarke-Doane, he argues that there is no better way of understanding the skeptical challenge in his 2016 paper, “Debunking and Dispensability”. I disagree, but let’s set that aside. At least he can be credited with not just intentionally understanding the skeptic’s argument in an implausible way.)
As for meeting the explanationist challenge: I have been very worried about this challenge, both for any moral realist, and for the pure perceptualist in particular. Partly, this is because I think this is the right way to understand skeptical challenges. Partly, this is because this is an extremely difficult challenge to meet. And partly, because I think it’s closely related to a challenge that has been largely (though not completely) overlooked by moral realists, but especially by non-naturalists, which is the issue of metasemantics. (And in fact, I hope I’m right about each of those claims, as I have no less than three distinct paper drafts on these issues in the works…)
I do think that there are ways to meet the explanationist constraint compatible with an a priori moral epistemology, for example the one that you (David) mention, Bengson’s. I can only say here that I think those approaches–or at least the ones that I am familiar with–all have, in my view, other very deep problems.
To summarize my current thinking, which I know will leave you unsatisfied, I’ll just sketch out how I think things would have to work for the explanationist challenge to be met. First, we have to assume reliability. I don’t mean this in a question-begging way–I’m happy to admit we might not be reliable, but for the reasons stated above about overgeneralization, I don’t think a positive argument for reliability is a fair burden to place on the anti-skeptical realist. (There are some complications I’m eliding over here for brevity, and I fully grant that these complications are not trivial.) So, if we’re reliable, we still need a story that connects up our moral beliefs to the moral facts in order to meet the explanationist challenge. This is the role that perceptual awareness is supposed to play, and the role that I think the a priori moral epistemologist has much more trouble meeting without falling into other problems.
I think this gets us to the following conditional: If we’re reliable, and pure perceptualism is true, then we have moral knowledge.
That’s a big ‘if’, to be sure, but I think that’s the best that the honest realist can do.
Preston, I think we probably agree on more than we disagree on here. I’m not sure what I think about the general vs. specific truth stuff. I have some inchoate views about how to square the synthetic a priori with explanationism that I won’t go into here, though I’ll say I likewise find Bengson’s view problematic.
Regarding the assumption of reliability: I agree that if the challenge in question disallowed the assumption of reliability, it would overgeneralize. In fact, I think that’s putting things mildly. I’m not even sure it would be a coherent challenge, because I understand the challenge partly as one to explain reliability, and it doesn’t make much sense to demand that someone explain X without assuming X obtains, at least for the sake of argument.
But I think the impression that this is part of the challenge is illusory. I think it’s fine to assume reliability if one merely assumes it obtains in order to explain it. By contrast, in my view modal conditions like safety and sensitivity provide defeasible abductive evidence for but do not constitute or explain reliability. When you’re providing evidence for X, the rules of the game are different; you no longer get to assume X. That’s why someone can’t agree with me that modal conditions (merely) provide evidence of reliability but then say “oh, well, that’s fine, assuming we’re reliable we would meet those modal conditions so we’re reliable.” That would be question-begging. (This is all in the first paper I linked to in my last comment, though this system doesn’t make them look much like links!)
Anyway, I take it we agree there’s a real challenge here and that it can’t be met as easily as Clarke-Doane thinks (whether or not you buy all the details of my reply). I also like your focus on the metasemantics; I’ve just always felt more at home in epistemology.
Let me just end with what continues to give me pause: the suggestion that perceptualism has an advantage because perceptual awareness “connects up our moral beliefs to the moral facts in order to meet the explanationist challenge.” (It might be worth noting here that I don’t really think of the reliability challenge as distinct from the explanationist one. Indeed, I don’t usually talk about “reliability” at all; I do that in this context just for dialectical simplicity. I think the challenge in this vicinity is a challenge to explain how our beliefs could be non-accidentally true, and I think that the right understanding of non-accidentality involves explanatory connections.)
Anyway, the issue is that to my ear, talk of being aware of something doesn’t just mean having an experience or even a factive experience of it, but rather having such an experience because of it. If that’s right and “awareness” invokes an explanatory connection, then saying “if we have perceptual awareness then we can meet the explanationist challenge” more or less just means “if we have a story about how moral truth explains our experiences then we can meet the challenge to show how moral truth explains our experiences.” Given that, I continue to worry that w/r/t this challenge moral perception seems appealing because most perception has an explanatory connection built in. But given that precisely the question in this case is how there could be such a connection, saying that it’s through “perceptual awareness” looks to me like not much more than a label for the thing the perceptualist needs, something I’m still not convinced she has any more hope of attaining than the intuitionist.