Preston Werner (Hebrew University of Jerusalem), "Toward a Perceptual Solution to Epistemological Objections to Nonnaturalism"

Journal of Ethics and Social Philosophy, 2023

May 15, 2023

Nonnaturalist moral realism is the view that moral truths are objective, mind-independent, and not reducible to natural facts. One important set of objections to nonnaturalism is epistemological. Since nonnaturalists are almost exclusively non-skeptical about moral knowledge, showing that moral knowledge is incompatible with a nonnaturalist metaphysics of moral properties would present a serious problem for nonnaturalism.

A great amount of journal space in the last few decades has been dedicated to epistemological objections to nonnaturalism and responses to those objections. But none of it, as far as I was able to tell, considers whether some particular moral epistemology may be in a better position to answer these objections than others.

In my paper, “Toward a Perceptual Solution to Epistemological Objections to Nonnaturalism”, I begin to ﬁll this gap in the literature. In some of my other work I have defended the theory of Moral Perceptualism, which says that for many human beings, moral properties are part of the contents of perceptual experience, in the same way that colors, shapes, and objects are.

The general thesis of the paper is that Moral Perceptualists are in a unique position for responding to epistemological objections to nonnaturalism. I think this means two things for the philosopher interested in nonnaturalism: First, that conditional on Moral Perceptualism, epistemological objections to nonnaturalism lose much of their force. Second, and as a corollary, insofar as Moral Perceptualism is uniquely placed to respond to these objections, this gives nonnaturalists a reason to endorse the existence of moral perception.

2. Two Epistemic Headaches for Nonnaturalists

Entire papers could (and have) been written on how best to understand the epistemological objections to nonnaturalism. As I read the history, I think there are basically two interrelated traditions to these objections - The ﬁrst tradition goes back to Benacerraf (1973). Benacerraf’s objection was to mathematical platonism, and relied on the now widely rejected causal theory of knowledge. But descendents of the view and the objection persist, and have been extended to nonnaturalist metaethics. The second tradition is largely seen as beginning with Street (2006) and Joyce (2006), who argue that a non-skeptical nonnaturalism is incompatible with the evolutionary etiology of our moral belief-forming mechanisms. (Evolutionary objections to moral realism have been around since Darwin, but Street is credited with providing the most sustained and detailed recent elaboration of this objection.)

I won’t retrace the complicated histories of each of these traditions here. For brevity, let me just state the two epistemic principles that I ultimately think are at the heart of these objections in their most charitable and forceful interpretations:

Epistemic Access: In order for our beliefs about some domain D to constitute knowledge, we must have epistemic access to the D-facts.
Explanatory Connection: In order for a belief B to constitute knowledge that P, P must play an ineliminable role in an explanation about why B exists.

In the rest of the paper (and the rest of this post), I describe how Nonnaturalist Moral Perceptualism can meet both of these principles.

3. Epistemic Access and Moral Perception

Before showing how moral perception (MP) can meet Epistemic Access, a word on what exactly Epistemic Access comes to. Epistemic access, as I understand it, involves establishing that some (metaphysical) relation holds between the D-beliefs and the D-facts that can ground positive epistemic status. Epistemic access is both weaker and stronger than reliability. It is weaker because it does not require accuracy—an epistemic access relation can hold without a majority of beliefs being true. But it is stronger because it requires some such relation to hold; even beliefs that are reliably true (because for example their contents are necessary) may not meet an epistemic access condition. Finally, note that the sort of relation that underwrites epistemic access need not be causal. It could be, for example, introspective, conceptual (as with self-evident truths, if they exist), or constitutive (as with a belief that I have at least one belief).

How can MP meet epistemic access? The simple answer is that, according to moral perceptualism, we can perceive moral properties, and surely perception is a kind of epistemic access if anything is. I think this simple answer is ultimately right, but initially it may seem to just move the bump in the rug. After all, according to nonnaturalism, moral properties are not of the same kind as colors, shapes, tables, and chairs. Nonnatural properties, in particular, would be causally ineﬃcacious. However, perception is widely thought to be an essentially causal process. So we have what looks like an inconsistent triad:

Causally Ineﬃcacious (CI): Nonnatural moral properties are causally ineﬃcacious.
Perceptual Access (PA): We have epistemic access to nonnatural moral properties through perception.
Causal Condition on Perception (CCP): Perception is an essentially causal relation.

I think this seeming inconsistent triad is what best explains our strong sense that something has gone awry in a theory of epistemic access to moral properties via perception. But if that’s right, that’s bad news for the objector. A closer read shows that the triad is not, contrary to initial appearances, inconsistent. CCP does require that any instance of perception is causal, but that does not entail that all properties represented in experience must themselves be causal. When Norma witnesses some teenagers harming a kitten, and, by hypothesis, perceives the badness of the action, her perceptual state is essentially causal, even if the badness itself is not.

One could fairly suggest at this point that the problem here is with how CCP is articulated, rather than with the spirit of the objection. Instead, we should reframe CCP as:

Strict CC: Necessarily, if a property F is part of the contents of S’s perceptual experience e, then F (or the fact that F is instantiated) is at least partially causally responsible for e.
The Strict Causal Condition, combined with CI and PA, does give us a genuinely inconsistent triad. The problem is that Strict CC is almost certainly false. Many properties which are commonly taken to be perceptible fail to meet Strict CC. In the paper, I explain in more detail, but to whet your appetite, here is a non-exhaustive list: Absence properties, Aﬀordance properties, biological kinds, intentions, colors (on some metaphysical views of colors). Once Strict CC is weakened to accommodate even some of these examples, nonnatural properties will also be (in principle) perceptible, because they have causally eﬃcacious properties as their partial grounds.

4. Explanatory Connections and Moral Perception

Let’s move on to the notion of Explanatory Connection (EC). Return to Norma: Norma sees the teenagers harming the kitten, perceives it as bad, and on that basis forms the belief that action is bad. Her belief about badness is, for the moral perceptualist, explained by her perception. But for nonnaturalist moral perceptualism to meet the Explanatory Connection condition, her perceptual experience of badness must be explained by badness.

Probably the most forceful way to argue that (EC) can’t be met is, channeling Sharon Street (2006), by appealing to the ultimate evolutionary source of our moral faculties. Our moral belief-forming mechanisms were not selected for tracking nonnatural facts. So whether they line up with them or not, there is no explanatory connection between the nonnatural facts and our moral beliefs. Skepticism (or antirealism) looms.

In response to this worry, as well as to the problem of EC in general, I say two things. First, the evolutionary worry, at least as it relates to EC, is not as powerful as it ﬁrst seems - as Andreas Mogensen has pointed out, this is to run together two types of explanation:

Imagine that insects in one species, S1, have a certain pattern of colouration that serves as camouﬂage: it resembles the surrounding foliage. Natural selection has favoured this pattern of colouration because it allows the insects to avoid predators.
Suppose the pattern of colouration arises because juveniles eat a certain kind of moss during a critical development period. However, the fact that the juveniles have this diet is irrelevant in explaining why having this kind of colouration confers greater relative ﬁtness: the colouration would be equally advantageous if it came about as a result of a diﬀerent set of developmental factors. (Mogensen 2015, 198)

To take a similar but more relatable case, consider my action of taking a break from writing this post to eat some chocolate. While it’s true that there is some deep evolutionary explanation about why I seek out sugary treats, my particular action in this particular circumstance can be wholly and adequately explained in a much more proximal way: I like chocolate, I was hungry, and I knew I had some chocolate left in the fridge.

What this shows us is that meeting EC doesn’t require a complete nonnaturalist vindicating evolutionary theory of our moral judgments. Rather, meeting EC requires a more local explanatory connection between agents perceiving moral properties and those properties in the world. How can we get this? Here - simplifying the paper a bit - I say two things by way of addressing the worry:

Any plausible metasemantic theory for perceptualist nonnatural moral reference will ensure an explanatory connection of the sort needed by the nonnaturalist moral epistemologist.
There are independently plausible and defended metasemantic theories for perceptualist nonnatural moral reference.

I defend 1 on the grounds that, on any plausible metasemantic theory, the feature(s) that play the reference ﬁxing role (causation, teleology, reference magnets, etc.) will themselves constitute an explanatory connection. I defend 2 by pointing to the recently growing literature on nonnaturalist metasemantics (Dunaway 2020, Suikkanen 2017, Werner 2020). Given moral perceptualism, the badness of the teenagers harming the kitten can (proximally) explain Norma’s perception of badness. The Explanatory Connection can be met.

5. Conclusion

To my dismay, moral perceptualism remains a relatively unpopular view in moral epistemology. To nonnaturalists’ dismay, epistemological objections to nonnaturalism remain inﬂuential and compelling. Maybe, just maybe, the nonnaturalists can adopt moral perceptualism, and we can all rest easier together.

Works Cited and Some Recommended Reading:

Benacerraf, Paul. “Mathematical Truth.” Journal of Philosophy 70, no. 19 (November 1973): 661–79.

Dunaway, Billy. Reality and Morality. Oxford: Oxford University Press, 2020.

Farennikova, Anna. “Perception of Absence and Penetration from Expectation.” Review of Philosophy and Psychology 6, no. 4 (December 2015): 621–40.

Joyce, Richard. The Evolution of Morality. Cambridge, MA: MIT Press, 2006.

Mogensen, Andreas. “Evolutionary Debunking Arguments and the Proximate/Ultimate Distinction.” Analysis 75, no. 2 (April 2015): 196–203.

Street, Sharon. “A Darwinian Dilemma for Realist Theories of Value.” Philosophical Studies 127, no. 1 ( January 2006): 109–66.

Suikkanen, Jussi. “Non-Naturalism and Reference.” Journal of Ethics and Social Philosophy 11, no. 2 (February 2017): 1–24.

Vavova, Katia. “Debunking Evolutionary Debunking.” In Oxford Studies in Metaethics, vol. 9, edited by Russ Shafer-Laundau, 76–101. Oxford: Oxford University Press, 2014.

———. “Evolutionary Debunking of Moral Realism.” Philosophy Compass 10, no. 2 (February 2015): 104–16.