2.1. Epistemic alienation

As we see it, there are two dimensions to this problem of alienation that are related, but distinct. The first dimension is epistemic: in order for us to be non-alienated from facts about normative properties, we must be capable of learning what these facts are. As Thomas Nagel, himself a non-naturalist, has put it, “I do not believe that the truth about how we should live could extend radically beyond any capacity we might have to discover it” (Nagel, Reference Nagel1986, 139). The worry is that non-naturalism renders it mysterious how we could non-accidentally arrive at true normative beliefs. If normative properties are sui generis , attitude-independent, and irreducible to natural properties, none of the familiar, uncontroversial avenues for epistemic connection seem to be available. We can discover normative truths neither through empirical investigation, nor through reflection on our own attitudes or commitments.

Absent an explanation of how we could have a capacity to form true beliefs about non-natural normative properties, we seem to be epistemically alienated from these properties. One guise under which this problem of epistemic alienation appears is that of the reliability challenge to non-naturalism.Footnote ⁵ The non-naturalist asserts not just that there are normative truths, but that our normative judgments often correlate with these truths. The reliability challenge is that, for the non-naturalist, this is “an unexplained striking correlation, and this may just be too much to believe,” as David Enoch puts it (Reference Enoch2011, 159).Footnote ⁶

Though the reliability challenge focuses on how we could reliably arrive at true normative beliefs, reliability is just one way of cashing out how our normative beliefs could be non-accidentally true. As such, we will understand the problem of epistemic alienation more broadly in terms of accounting for non-accidentally true normative beliefs. But in our view, the problem is not simply an instance of the more general point that, for any domain of facts we posit, we need a plausible story of how we can arrive at non-accidentally true beliefs. There are some facts—for example, regarding the number of stars in the universe—such that it would be unsurprising if we could never manage to non-accidentally access them because they ultimately have little to do with us. Normative facts are not like this. They are concerned with properties of human actions that are supposed to guide us. Normative properties, by their nature and ours, are of a kind that must be able to get an epistemic grip on us.

This is why we view existing epistemological objections to non-naturalism as part of a larger problem of alienation. In the absence of a satisfying explanation of how non-natural normative facts could get an epistemic grip on us, non-naturalism leads to epistemic alienation. As Samuel puts it, “To accept a traditional realist account of the explanation of normative facts while remaining skeptical of the realist’s non-explanation of their knowability would leave one in a state of epistemic alienation, convinced that there were normative facts but with no way of discovering what they were” (Reference Samuel2023, 170). If we are epistemically alienated from the normative, then it is difficult to see how normative properties could play the role they are supposed to play in our lives.

2.2. Deliberative alienation

The second dimension of the problem of alienation is deliberative. In order for us to be non-alienated from normative properties, these properties must play an essential role in our deliberation. Importantly, that we have non-accidentally true beliefs about normative properties is insufficient for these properties’ having an essential role to play in deliberation. Assume for the sake of argument that there are sui generis , attitude-independent, irreducibly normative properties about which we can arrive at non-accidentally true beliefs. This leaves open the possibility that facts about these properties could be “recognizably true, but irrelevant or obscure” to us—“mere facts,” as Samuel puts it (Reference Samuel2023, 162). The worry is that non-naturalism renders it mysterious how normative properties could be such as to essentially get a grip on us in deliberation.

The problem of deliberative alienation has been most forcefully raised by Christine Korsgaard. In The Sources of Normativity, she writes that metaethical views on which normative properties are attitude-independent, which she terms substantive realism, make ethics out to be a theoretical subject rather than a practical one. On her depiction of how realists see things, “we have normative concepts because we have spotted some normative entities, as it were wafting by” (Reference Korsgaard and O’Neill1996, 44). According to Korsgaard, this leaves mysterious why these normative concepts would regulate our conduct in the way they do. If our normative concepts are simply reflections of facts that exist out there, independently of us, then why would they have this essentially practical role? Even if we could establish that non-natural normative properties existed, and that we could arrive at non-accidentally true beliefs about them, we might wonder what these non-natural properties ultimately have to do with us and our practical deliberation.

Moreover, this sort of worry might be thought of as particularly pressing for non-naturalism, as opposed to normative realism in general. Naturalist realists tend to identify normative properties with natural properties that seem clearly practically relevant, such as well-being or counterfactual desires.Footnote ⁷ But for the non-naturalist, who thinks that normative properties aren’t irreducible to anything natural, the worry is that these normative properties could, in principle, have any extension whatsoever. Without a story that anchors these normative properties in something about the human agents they are supposed to govern, it would be unclear what prevents these properties from being totally alien to us as practical deliberators, even if we can discover them.Footnote ⁸

Like its epistemic counterpart, the problem of deliberative alienation can be seen as a problem of unexplained correlation. In its guise as the reliability challenge, the problem of epistemic alienation is about an unexplained correlation between our normative judgments and the normative facts. The problem of deliberative alienation is about a related but distinct unexplained correlation: between the instantiation of non-natural normative properties, on one hand, and the application of normative concepts by which we are moved in practical deliberation, on the other. As Samuel puts the point, for a metaethical theory to avoid alienation, it must account for “the non-accidental harmony of reasons for action and our capacity to act for reasons” (Reference Samuel2023, 164). If non-naturalism lacks such a story, then it is simply unclear why these non-natural properties should figure essentially in our deliberation. Moreover, it is unclear why it would be a failure of deliberation for someone to note the existence of non-natural normative properties and simply shrug, as one might upon registering the existence of some obscure non-normative property.Footnote ⁹ Thus, in the absence of a satisfying explanation of the deliberative grip of non-natural normative properties, non-naturalism leads to deliberative alienation.

As we see it, both the epistemic and deliberative components of the problem of alienation must be taken seriously by non-naturalists. A satisfactory defense of normative non-naturalism should explain the non-accidental connection between normative properties and (a) our normative beliefs and (b) the application of normative concepts by which we are moved in deliberation. In other words, non-naturalists must explain how sui generis , attitude-independent, irreducibly normative properties can nevertheless be essentially epistemically available and figure essentially in deliberation.

4.1. Individual choiceworthiness

Begin with individual choiceworthiness. According to Deliberative Intenders, we form intentions through deliberation, that is, through the consideration of various features of our options. Our deliberative method of intention formation raises a familiar practical problem: how does one get from consideration of the features of options to a commitment to one course of action rather than another? In other words, what psychological mechanism enables us to transition from deliberation to intention formation?

We think there need not be only one solution to this problem. But the following seems clear: this transition generates an indispensable role for a concept to play, namely, the role of closing deliberation by categorizing the considerations in favor of one course of action as sufficient to commit to that course of action. Such a concept would thereby help solve the practical problem of closing deliberation.Footnote ¹⁹ Moreover, this particular form of solution would be indispensable in complex choice situations where various features of options pull in different directions. Such episodes of deliberation, where deliberation is difficult to close, are highly distinctive of the kinds of creatures we are. It is thus not unreasonable to infer that we in fact possess a concept with the conceptual role of closing deliberation in this manner. Which concept possesses this role? A promising contender is the familiar normative concept expressed by the English phrase “individual choiceworthiness”: the concept of the thing to (individually) do all things considered. We therefore propose the following principle:

It’s worth emphasizing two features of ICP. First, it is often thought that the non-naturalist must accept that normative concepts have (only) representational functions—they function to track non-natural normative properties. But we are describing a version of non-naturalism on which individual choiceworthiness has a distinctively practical function grounded in our constitution as agents.Footnote ²⁰ Still, as we’ll see in a moment, it has a non-natural normative property as its referent, assuming a plausible metasemantic theory. This is a clear sense in which our brand of non-naturalism blends constitutivist and non-naturalist elements. Second, ICP is highly ecumenical. It is consistent with many different views on the nature of intention, including the planning theory of intention (Bratman, Reference Bratman1987). It is also consistent with the possibility of cases where one forms individual intentions without seeing what one intends as individually choiceworthy. And it is consistent with, but does not entail, the guise of the good theses.

We’ve connected our possession of individual choiceworthiness to constitutive features of our agency. In particular, we have explained what individual choiceworthiness is doing for us and, through the explanation, shown that it has the distinctive practical role of closing deliberation. Our next question is about reference: what explains why individual choiceworthiness refers to the non-natural normative property of individual choiceworthiness, as opposed to some nearby (natural) property? To answer this question, we sketch a metasemantics, inspired by Wedgwood’s (Reference Wedgwood2007) conceptual role semantics for normative concepts, that fits well with our marriage of constitutivism and non-naturalism.Footnote ²¹ There will be unaddressed subtleties; we point the reader to Wedgwood (Reference Wedgwood2007) for details.

On Wedgwood’s version of conceptual role semantics, a concept’s reference is fixed by the concept’s basic roles within our inferential and practical reasoning. These roles are captured by basic rules of rationality governing its use. Such rules specify which “output” mental states one must commit to upon (rational) acceptance of some “input” mental state, on pain of irrationality. For instance, the concept OR is a binary operator on thoughts captured by the following two basic rules of rationality: acceptance of p rationally commits one to accepting ORpq; acceptance of q rationally commits one to accepting ORpq.

Conceptual roles fix reference in the following way: a concept refers to that property which makes the concept’s set of basic rules of rationality each valid and jointly complete. A rule is valid when: necessarily the output of the rule is correct if the input to the rule is correct. A set of rules is complete when: necessarily, if the acceptance (or rejection) of an application of the concept is incorrect, then one of the rules is violated. So for instance, in order for the two basic rules of rationality governing OR to be valid and complete, OR must refer to the following binary relation: necessarily, Rpq if and only if p is true or q is true. This is precisely the binary relation of classical disjunction.

Motivated by our discussion of its conceptual role in closing deliberation, the basic rule of rationality governing the use of individual choiceworthiness is: a judgment that $ \varphi $ is individually choiceworthy rationally commits you to closing deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing.Footnote ²² Thus, given that the referent of individual choiceworthiness must make this rule both valid and complete, the referent must be such that the following biconditional is true: necessarily, a judgment that $ \varphi $ is individually choiceworthy is correct if and only if the closing of deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing is correct. Recall now Constitutive Practicality IC: in its nature, the non-natural normative property of individual choiceworthiness is the correctness condition on the closure of deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing. So, if individual choiceworthiness had as its referent a property that was not coextensive with individual choiceworthiness, then its referent would fail to make the biconditional true. Individual choiceworthiness (or some property necessarily coextensive with individual choiceworthiness) must therefore be the referent of individual choiceworthiness.

How do we rule out that some natural property distinct from but necessarily coextensive with individual choiceworthiness is the referent of individual choiceworthiness? This is an instance of a more general metasemantic problem, namely to explain how reference gets fixed where necessarily coextensive properties exist. In particular, this question does not only face the non-naturalist. Suikkanen (Reference Suikkanen2017) shows that there is an analogous question for the naturalist who relies on a causal theory of reference. As Suikkanen argues, the most promising solution for both—albeit not an entirely satisfying one—is an appeal to reference magnetism (13–15). This solution is particularly appealing in our case, given that the non-natural normative properties at issue are especially joint-carving and thus especially good candidates for reference magnets. Still, given that establishing reference when there are necessarily co-existensive properties is a general metasemantic problem, we do not take it to be one we need to settle here.

Finally, turn to the question of content. Along with a worldly referent, a concept has a cognitive significance to the possessor of the concept. In grasping a concept, one grasps a characterization of a kind of thing. This characterization that one grasps in grasping the concept is what we call the concept’s content. For our purposes, the most important feature of a concept’s content is that it guides the possessor of the concept in application: one applies the concept to that which one believes matches the characterization contained in the concept’s content. So, the more accurate this characterization is to the concept’s referent, the better possessors of the concept will be at applying the concept correctly. For example, even if the referent of our concept of water ( $ {\mathrm{H}}_2\mathrm{O} $ ) has remained the same over time, its content has changed as we have learned more about modern chemistry. Accordingly, we have gotten better at applying the concept of water.Footnote ²³

Now, even with an explanation of our possession of a concept that refers to individual choiceworthiness, we must still say something about its content in order to explain our non-alienation from normativity. Indeed, as far as we have said, we might possess a concept which refers to the property of individual choiceworthiness but whose content radically mischaracterizes its referent. That our relationship to individual choiceworthiness is not so can be explained by the second constitutive feature of individual choiceworthiness—Constitutive Supervenience IC—and the practical role of our concept of individual choiceworthiness, a practical role itself grounded in Deliberative Intenders. The explanation has two parts.

First, in order for individual choiceworthiness to be indispensable to us as a concept that closes deliberation, the application of the concept must be guided by the instantiation of properties that we have access to and which support the attainment of our practical ends. Both of these constraints would be met if the application of individual choiceworthiness were guided by the instantiation of natural properties of individual practical relevance, such as being desire-satisfying, being preference-satisfying, promoting pleasure, reducing pain, and promoting survival. These properties are both identifiable and generally among our practical ends. Put another way, if a concept whose role is to close deliberation were applied entirely independently of desire, preference, survival, and pleasure, then we would be severely hindered as agents in ways that we generally are not. There would be no systematic relationship between what we intended to do and what was in our practical interest. Second, recall Constitutive Supervenience IC: it is constitutive of individual choiceworthiness that it is closely connected to natural properties of individual practical relevance, such as desire satisfaction, preference satisfaction, promotion of pleasure, and so on. In light of this constitution principle, another way to describe what we established in the first part of the explanation is: in light of our nature, the nature of individual choiceworthiness, as captured in Constitutive Supervenience IC, is mirrored in the content of individual choiceworthiness.

Two points of clarification. First, it is part of the content of individual choiceworthiness that there is some close connection between individual choiceworthiness and natural properties of individual practical relevance; the exact contours of this close connection are not, however, part of its content. Rather, we form beliefs about these contours through first-order normative theorizing using individual choiceworthiness. Second, the thought is not that there is a causal or even explanatory connection between this constitutive feature of individual choiceworthiness and the content of individual choiceworthiness being what it is. Rather, there are reasons independent of the constitution of individual choiceworthiness for individual choiceworthiness to have a content that mirrors its constitution: the concept’s practical role in closing deliberation.

To sum up, we have sketched a harmony between us and the non-natural normative property of individual choiceworthiness. In being deliberative intenders, we are the kinds of beings who need a concept that plays a distinctive functional role in closing practical deliberation. This concept is individual choiceworthiness. Given a plausible metasemantic theory and Constitutive Practicality IC, this functional role entails that individual choiceworthiness refers to the property of individual choiceworthiness. Moreover, we should expect individual choiceworthiness to fulfill its functional role in a manner that supports the attainment of our practical ends. So, we should expect it to have a content that reflects Constitutive Supervenience IC.Footnote ²⁴ We can now tell a similar story for joint choiceworthiness and moral requirement, thus articulating a multi-level harmony between us and normativity.

4.2. Joint choiceworthiness

Recall Joint Deliberative Intenders: constitutively, we non-arbitrarily settle on courses of joint action on the basis of joint deliberation. We play duets together, go on walks together, and build houses together. Moreover, we deliberate about what songs to sing together, where to walk together, and what kinds of houses to build together. In doing so, we participate in joint intentions formed on the basis of joint deliberation.

Given our nature as joint deliberative intenders, then, we face a practical problem similar to the one discussed in the previous section: how do we go from joint deliberation to the formation of a joint intention? This practical problem gives rise to an indispensable conceptual role for a concept to play: the role of closing joint deliberation through individuals categorizing the considerations in favor of one course of joint action as sufficient to commit to the course of joint action. This concept would function to close joint deliberation through its deployment by the individuals engaged in the joint deliberation. (So, we are not assuming that groups as such possess and deploy concepts.)

Which concept possesses this conceptual role? Note that it cannot be individual choiceworthiness. Part of what is distinctive of joint deliberation is that the participants are each sensitive to properties of joint practical relevance, not just individual practical relevance. In deliberating with you about whether we will dance a tango or a waltz, we are each closely attentive to, e.g., what is mutually desirable and preferable. Yet, as already discussed, individual choiceworthiness is, as a matter of conceptual necessity, applied with close attention instead to properties of individual practical relevance.

Thus, a more promising contender for a concept that functions to close joint deliberation is a distinct but related normative concept expressed by the English phrase “joint choiceworthiness”: the concept of the thing to jointly do all things considered. We therefore propose the following principle, analogous to ICP:

So, we have connected our possession of joint choiceworthiness to constitutive features of our agency much in the way we did with individual choiceworthiness. The explanation of why joint choiceworthiness refers to the non-natural normative property of joint choiceworthiness and reflects Constitutive Supervenience JC in its content mirror almost exactly the corresponding explanations for individual choiceworthiness. Thus, we present them briefly.

Take first the question of reference. The basic rule of rationality governing the use of joint choiceworthiness is: for a joint act-type $ \varphi $ , a judgment that $ \varphi $ is joint choiceworthy rationally commits one to do one’s part to close joint deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing. Thus, given that the referent of joint choiceworthiness must make this rule both valid and complete, the referent must be such that the following biconditional is true: necessarily, for a joint act-type $ \varphi $ , a judgment that $ \varphi $ is joint choiceworthy is correct if and only if doing one’s part to close joint deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing is correct. Recall Constitutive Practicality JC: by its nature, the non-natural normative property of joint choiceworthiness is the correctness condition on (doing one’s part in) the closure of joint deliberation about whether to $ \varphi $ in favor of $ \varphi $ ing. Thus, applying Wedgwood’s conceptual role semantics as before, it follows that joint choiceworthiness refers to the non-natural normative property of joint choiceworthiness.

Next, take the question of content. We should expect the application of joint choiceworthiness to be guided by the instantiation of properties that we have access to and which support the attainment of joint practical ends. Otherwise, we would not be very good at acting with others, given the concept’s central role in closing joint deliberation. Both of these constraints would be met if the application of joint choiceworthiness were guided by the instantiation of natural properties of joint practical relevance, such as being mutually desire-satisfying, being mutually preference-satisfying, promoting mutual pleasure, reducing mutual pain, and promoting mutual survival. Put another way, if a concept whose role is to close joint deliberation were applied entirely independently of properties such as mutual desire, mutual preference, mutual survival, and mutual pleasure, then we would be severely hindered as agents who act with others in ways that we generally are not. In fact, we are generally quite good at acting with others. In light of Constitutive Supervenience JC, another way to put what we have just established then is that the nature of joint choiceworthiness, as captured in Constitutive Supervenience JC, is reflected in the content of joint choiceworthiness.

4.3. Moral requirement

Finally, turn to Socially Organized: constitutively, we do not just act together—we live together. We organize ourselves into nations, economies, legal systems, cultures, religious groups, and other highly organized social forms of life. While other species engage in cooperative activity, only we actively organize ourselves into and maintain large-scale institutions. Our “group-mindedness” is, as argued by Tomasello and Vaish (Reference Tomasello and Vaish2013), unique to humans among all primates. Our highly social, institutionally embedded forms of living raise a third and again familiar practical problem for us: how do we coordinate with one another at such a scale, given our diverging individual (and smaller group) practical interests?

As many before us have argued, moral concepts are particularly well-suited to solve this problem.Footnote ²⁵ As Tomasello & Vaish put it:

Living together successfully requires forgoing personal benefit for the sake of the group, accepting certain forms of unfreedom, taking an interest in the well-being of strangers, according equal worth to all, and following, maintaining, and enforcing rules. Moral concepts enable us to do these things—and at a large scale. Thus, while moral concepts are intimately connected to the concept of joint choiceworthiness, the latter is too thin to do all the practical work we need done to live together. We need moral concepts too.

So, we take on board much of what has already been said about the function of moral concepts in enabling cooperative living. Like the concepts of choiceworthiness and joint choiceworthiness, moral concepts are primarily for solving practical problems, problems that arise in our being the kinds of creatures constitutively disposed to organize ourselves into large-scale cooperatives. However, we do not think that accepting as much requires adopting a naturalistic reduction of moral properties. Rather, we suggest that the practical function of moral concepts enables us to grasp irreducibly normative moral properties. It is through living together that we have discovered—rather than invented—morality.

Indeed, we propose that the function of a concept of moral requirement (and interdefinable moral concepts like moral permissibility) is to enable us to successfully engage in long-term forms of cooperative living. This function corresponds to a meaning-constituting rule of competence for the concept: a judgment that $ \varphi $ is morally required commits you to treating $ \varphi $ as practically necessary for successfully engaging in long-term forms of cooperative living. Thus, given that the referent of moral requirement must make this rule both valid and complete, the referent must be such that the following biconditional is true: necessarily, a judgment that $ \varphi $ is morally required is correct if and only if it is correct to treat $ \varphi $ as practically necessary for successfully engaging in long-term forms of cooperative living. As captured by Constitutive Practicality MR, the non-natural property of moral requirement is the correctness condition on treating $ \varphi $ as practically necessary for successfully engaging in long-term forms of cooperative living. Thus, applying Wedgwood’s conceptual role semantics as before, it follows that moral requirement refers to the property of moral requirement.

Moreover, as has been discussed at length by others, for moral requirement to solve the distinctive sorts of coordination problems we face in trying to live together, we need a concept of morality whose prescriptions are closely related to natural properties such as being other-regarding, beneficent, impartial, and altruistic.Footnote ²⁶ According to Constitutive Supervenience MR, the property of moral requirement is closely related to precisely these natural properties of moral relevance in virtue of its constitution. Thus, the function of moral requirement sets up not only reference to the irreducibly normative property of moral requirement but also a content which reflects the constitutive nature of its referent as captured in Constitutive Supervenience MR.

The key idea then is that non-naturalist moral realists can take on board much of what naturalists have said about the practical—and, in particular, social—function of moral concepts. However, they can do so not as a way of reductively explaining the nature of moral properties but as a way of explaining our grasp of irreducibly normative moral properties.

5.1. Solving the problem of epistemic alienation

First, a brief reminder of the problem of epistemic alienation. The domain of the normative seems to be essentially about us—normative facts are facts about human agents qua human agents, and normative properties are properties of human actions qua human actions. In light of this, normative properties must be a kind of property that is essentially discoverable by human agents. That is, it is not only that we must have some prospect of arriving at non-accidentally true beliefs about normative properties. Our capacity to discover normative properties must be a robust capacity, in the sense that it guarantees that many normative facts are not obscure or inaccessible to us, as opposed to, say, facts about the number of stars in the universe. The problem of epistemic alienation is that, if non-naturalism lacks an explanation of all of the above, it renders us epistemically alienated from the normative.

Here is how the resources developed in the previous sections provide such an explanation. Consider the properties of individual choiceworthiness, joint choiceworthiness, and moral requirement. Ex hypothesi, these are non-natural, irreducibly normative properties. How can it be that we have a robust capacity to arrive at non-accidentally true beliefs about these properties? For simplicity, we will focus on explaining our answer to this question as it pertains to individual choiceworthiness (for the rest of this section, just choiceworthiness), given that our answer to the problem has the same structure when it comes to both joint choiceworthiness and moral requirement.

Recall that on our view, facts about what is constitutive of us and normativity explain (i) our possession of the concept individual choiceworthiness, and (ii) the reference of that concept to the non-natural normative property of individual choiceworthiness. These in turn help us explain (iii) our robust capacity to arrive at non-accidentally true beliefs about which actions are individually choiceworthy.

These first two steps explain how we can have beliefs about which actions are choiceworthy at all. This is, of course, only a precondition for having a robust capacity to arrive at non-accidentally true beliefs about which actions are choiceworthy. As we have argued, there is another constitutive harmony that explains this final step. Although choiceworthiness is an irreducibly normative property, it nevertheless supervenes on certain natural properties, the core of which we have labeled properties of individual practical relevance. This is constitutive of choiceworthiness; if it did not supervene on such properties, it would not be the property of choiceworthiness. Furthermore, we have argued that individual choiceworthiness plausibly includes in its conceptual content at least the vague shape of these connections. This makes it difficult for us to have systematically false beliefs about which kinds of actions are choiceworthy.

Crucially, however, our claims about the conceptual content of individual choiceworthiness do not entail that we are always in a position to arrive at true beliefs about what is choiceworthy, simply in virtue of our competence with the concept. The connections identified by the content of the concept cannot tell us exactly which actions are choiceworthy, nor will they determine any particular first-order theory of choiceworthiness. They can, however, constrain the conceptual possibilities, ruling out certain kinds of radical mistakes, such as thinking that choiceworthiness has to do with promoting suffering or frustrating our own interests. The rest, on our view, must be figured out through reasoning and theorizing. A first-order theory of choiceworthiness would follow only from a particular conception thereof, not from the concept itself.Footnote ²⁷ This is an advantage of our view, as it seems rather implausible that which actions are choiceworthy would be fully and directly settled by apprehension of conceptual truths.

So far, we have explained how the construction of individual choiceworthiness makes it difficult for us to have systematically false beliefs about which actions are choiceworthy. But what we need to show is that we have a robust capacity to arrive at non-accidentally true beliefs. Here is how that works on our view. We start with conceptual truths about what is of core relevance to choiceworthiness. When we form beliefs about conceptual truths by exercising our competence with the concept, those beliefs are uncontroversially non-accidentally true. These non-accidentally true beliefs are vague, but they give us a foothold in developing a larger corpus of non-accidentally true beliefs about what is choiceworthy. From these conceptual truths, we build out through reasoning and theorizing to arrive at further beliefs about what is choiceworthy.Footnote ²⁸ If all goes well, we arrive at further true beliefs. Moreover, because this reasoning and theorizing will be constrained by the conceptual role of individual choiceworthiness in solving our practical problems, these further beliefs will retain a connection to our foothold in the nature of choiceworthiness. Of course, we can still go wrong and end up with false beliefs. But the aim of our view is not to deliver normative infallibility. In fact, a good explanation of how we can non-acidentally arrive at true normative beliefs should not render falsehood and mistake impossible.Footnote ²⁹

We contend that our capacity to work outward from what is entailed by the conceptual content of individual choiceworthiness, and thereby to form further true beliefs, is a robust capacity to arrive at non-accidentally true beliefs about choiceworthiness. Of course, there are different conceptions of non-accidentality. On a modal conception, for our beliefs to be non-accidentally true, they simply need to be modally robust enough. Any beliefs that are true as a conceptual matter are clearly modally robust enough (as is our possession of the normative concepts discussed, given our nature). For further true beliefs arrived at through reasoning and theorizing to be modally robust enough, we need only suppose that reasoning and theorizing are modally robust enough methods for forming further true beliefs. To suppose this is only to reject pessimism about first-order normative reasoning and theorizing (cf. Enoch, Reference Enoch2010, 428).

On an explanatory conception of non-accidentality, for our beliefs to be non-accidentally true, there must be an explanatory connection between the truth of those beliefs and our holding them.Footnote ³⁰ This too, can be delivered on our view. Beliefs that are true as a conceptual matter easily satisfy the condition: if p is a conceptual truth that I come to believe by exercising my competence with the concept, then my belief that p is explained by the truth of p. For further true beliefs arrived at through reasoning and theorizing, it must be that the explanatory connection to the truth is preserved in exercises of those capacities. Though this is not as straightforward, we think this too is plausible. If the reasoning and theorizing through which we derive further normative truths reflect the metaphysical relations between the initial conceptual truths and these further normative truths, then it is plausible that the explanatory connection between truth and belief is preserved.

Thus, on our view, it is no mystery how we can arrive at non-accidentally true beliefs about which actions are choiceworthy. Indeed, in virtue of what is constitutive of us as human agents, and what is constitutive of choiceworthiness, we have a robust capacity to discover what is choiceworthy. Thus, facts about what is choiceworthy—though they are non-natural, irreducibly normative facts—could not be obscure or inaccessible to us, like facts about the number of stars in the universe. We may often get things wrong, but it is always possible in principle for us to get them right. As such, we have shown how, even though choiceworthiness is a non-natural and irreducibly normative property, we are not epistemically alienated from it.

We take the above explanation of our non-alienation from individual choiceworthiness to apply mutatis mutandis to joint choiceworthiness. When it comes to moral requirement, the story is also largely the same. However, it focuses not on individual or joint deliberation in general, but specifically on the kind of deeply social cooperative living that is distinctive of human agents. As a reminder, in Section 4 we explained how, in order to successfully pursue this kind of living in our practical deliberation, we need concepts that are less thin, and have constraints on their correct application that relate closely to large-scale forms of cooperation. This is why, in addition to thinner practical concepts, we have distinctively moral concepts that refer to distinctively moral non-natural properties. In virtue of this match, we are capable of forming some non-accidentally true moral beliefs simply by being competent with moral concepts. From there, we move from concept to conception through moral reasoning and theorizing, building out a larger corpus of (hopefully) non-accidentally true moral beliefs.

5.2. Solving the problem of deliberative alienation

Now to the problem of deliberative alienation. Once again, we may begin from the observation that the normative domain seems to be essentially about us. Normative properties are properties of human actions, and it is essential to them that they are supposed to guide the conduct of beings like us. As such, normative properties must be a kind of property that figures essentially in the activities that guide our conduct – i.e., in practical deliberation. This is not just to say that these are the kind of properties that could in principle figure in our deliberation. There must be a robust connection between the nature of our deliberation and the nature of these normative properties that rules out a scenario in which normative properties are totally practically obscure or irrelevant from our deliberative standpoint. The problem of deliberative alienation is that, if non-naturalism lacks an explanation of all of the above, it renders us deliberatively alienated from the normative.

As with the problem of epistemic alienation, we think the resources we have developed thus far can provide such an explanation. The question we must answer to respond to the problem of deliberative alienation is, why would non-natural, irreducibly normative properties be such as to get a grip on us in deliberation? To illustrate our answer to this question, it will be helpful to consider further why Korsgaard thinks realism is subject to the problem of deliberative alienation and constructivism is not.

As quoted before, Korsgaard objects that on a realist view, “we have normative concepts because we have spotted some normative entities, as it were wafting by” (Reference Korsgaard and O’Neill1996, 44). A large part of her worry in making this remark is that realism cannot explain why normative properties are not entities of merely theoretical interest. As she understands realism, the role of our normative concepts on a realist view would be, in the first instance, to correspond to normative properties existing out there in the world. The worry is that, even if normative properties so construed can perfectly well get a grip on us epistemically, that would be the end of the story. We would have registered a normative truth, but we would have no explanation about why that should have any upshot for our deliberation. In other words, it would be mysterious why it would be a failure of deliberation to simply register these normative truths and move on, treating them as practically irrelevant curiosities about the universe, as we might register the existence of some distant celestial entity.

In later work, Korsgaard contrasts what she sees as the realist picture with a constructivist understanding of normative concepts and properties. As she puts it:

Thus, according to Korsgaard, realists and constructivists understand both normative properties and normative concepts fundamentally differently. For the realist, normative properties are fundamentally theoretical entities, to be grasped through concepts whose primary role is to “describe some piece of external reality” (Reference Korsgaard2003, 117). This renders it mysterious how normative properties could play any essentially practical role in our deliberation. By contrast, for the constructivist, normative properties are solutions to practical problems, and so they are fundamentally practical rather than theoretical entities. Correspondingly, they are grasped through fundamentally practical concepts, whose role is to enable us to solve the practical problems we face.

Strikingly, on the view we have been developing, both normative properties and concepts resemble Korsgaard’s understanding of them much more than the understanding she attributes to the realist. First, consider how we understand the role of normative concepts such as individual choiceworthiness. On our view, the conceptual role of individual choiceworthiness is to enable us to close deliberation by coming to an all-out judgment about what is ultimately favored by the practically relevant features of our options. As such, the conceptual role of individual choiceworthiness is in the first instance to solve a practical problem – the problem of how to make non-arbitrary choices in complex situations. The conceptual role of individual choiceworthiness is not in the first instance to accurately represent some piece of external reality.

Korsgaard seems to think that normative concepts’ having an essentially practical role precludes them from referring to non-natural, irreducibly normative properties. This is where we disagree. In our view, it is precisely because normative concepts have an essentially practical role that they are able to refer to essentially practical normative properties. As the version of conceptual role semantics we have laid out explains, if individual choiceworthiness did not have the essentially practical conceptual role it does, then it would not be able to refer to the property of individual choiceworthiness, given its essential practicality as captured by Constitutive Practicality IC.

When it comes to normative properties themselves, we similarly see their status as non-natural and irreducibly normative as compatible with their essential practicality. On our view, it is constitutive of normative properties that they are poised to help us solve practical problems. For example, in being the correctness condition on closing deliberation, it is in the nature of the property of individual choiceworthiness to help us solve the problem of how to make non-arbitrary choices in complex situations. Relatedly, it is constitutive of normative properties that they are closely connected to subvening natural properties – the very same properties that are of central relevance in deliberation. For example, the property of individual choiceworthiness, by its nature, marks the relevant subvening properties as having normative significance, guiding us to respond to them in deliberation.Footnote ³¹

We think something similar is true of other normative properties, such as joint choiceworthiness and moral requirement. Each of these normative properties marks subvening natural properties as having a particular kind of normative significance, relative to the kind of practical problems it is in the nature of the property to help us solve. The concepts that refer to these normative properties do so in virtue of having different conceptual roles, all of which are essentially practical. For example, the conceptual role of moral requirement is to point us towards those actions that are necessary for successfully engaging in long-term forms of cooperative living. This concept fulfills its conceptual role only in the context of a particular kind of practical deliberation. Thus, on our view, it cannot be correctly deployed in a merely theoretical manner any more than it can be on a constructivist or constitutivist view. It is in the nature of practical thought that, in putting us in touch with normative properties, it also puts us in a position to be guided by them in deliberation.

So, there is a sense in which, like Korsgaard, we understand normative properties as solutions to practical problems. However, we disagree with Korsgaard’s contention that this precludes these properties from being non-natural and irreducibly normative. Unlike constructivists, we deny that the existence of normative properties depends on their being the outputs of a procedure of construction. However, we do think that, given our nature as deliberators and the nature of normative properties, it is no mystery how they get a grip on us in deliberation. Of course, as we have already explained, there is no guarantee that we will always get things right. We can always fail to deliberate well, and in doing so, fail to be guided by what is genuinely choiceworthy, or morally required. However, when we do deliberate well, we correctly identify the solutions to our practical problems and as such are guided by normative properties. Because of the essentially practical nature of these properties, as well as the concepts that reflect them, the very process by which we are guided by them rules out the possibility that we could register them as obscure or irrelevant from our standpoint as deliberators. Insofar as we grasp normative facts in deliberation, we cannot grasp them as “mere facts,” as Samuel worries (Reference Samuel2023, 162). As such, we have shown how, even though properties like choiceworthiness are non-natural and irreducibly normative, we are not deliberatively alienated from these properties.