Toward a Post-Kantian Constructivism

1.1. The Received View of Constructivism in Metaethics

In the last decade the conventional wisdom has consolidated around the idea that what speaks in favor of metaethical constructivism, if anything, is its ability to balance a handful of theoretical desiderata. Facing a stalemate between realism and antirealism, constructivism supposedly aims to recover the objectivity of moral facts from the prevailing noncognitivism of the mid-20th century, and to do so without running afoul of the naturalistic worries associated with critics of traditional (‘intuitionist’) moral realism (e.g., Mackie 1977). Constructivism splits the difference, rendering moral facts objective, truth-apt, and subject to genuine disagreement, while remaining naturalistically respectable.³ Enoch (2011: 324) summarizes this motivation for the Kantian constructivist nicely:

Many people are suspicious about more robust, non-procedural forms of metanormative realism. They think that there are serious metaphysical and epistemological worries (and perhaps others as well) that make such realism highly implausible. Nevertheless, going shamelessly antirealist also has problems. We seem to be rather strongly committed, for instance, to there being correct and incorrect ways of answering moral (and more generally normative) questions, and moreover our moral (and more generally normative) discourse purports to be rather strongly objective. Constructivism may be thought of as a way of securing goods realism (purportedly) delivers, for a more attractive price.

Constructivism’s claim to strike this balance is that it understands normative facts as determined by the outcome of a specified procedure, for example systematic scrutiny of one’s values or practical deliberation from the first-person point of view. More generally, it holds that normative facts are determined by what follows from occupying a certain (practical) standpoint.⁴ Thus, insofar as there is a fact of the matter about what follows, we can be wrong about our reasons, and they are in that sense objective. However, the standard for determining what they are—what does and what does not follow—involves no appeal to normativity beyond that which is implicitly involved in occupying the practical standpoint, something that we all recognize that we do.

The trouble with this way of motivating constructivism is that its naturalistic credentials appear to rely on characterizing the practical standpoint in such minimal terms that our occupying it is uncontroversial, but it is difficult to see how such a minimal characterization of the practical standpoint can provide the resources to meaningfully constrain the results. Though some Kantians remain optimistic, something like a consensus has emerged that from a theory of agency thin enough to be uncontroversial, nothing substantive can follow via a procedure of rational scrutiny.⁵

This bleak assessment of the prospects for metaethical constructivism, however, relies on a misunderstanding of the question for which it’s an answer. Returning to The Sources of Normativity with this picture in mind, one is likely to be struck by Korsgaard’s lack of evident concern for “queerness,” or anything else metaphysical. Her concern is that reasons for action are reasons for agents, and cannot be understood as entities like any other.⁶

1.2. Constructivism and Normative Alienation

In the canonical statement of Kantian constructivism, Korsgaard argues that more traditional forms of realism leave an explanatory gap between the normative facts and the agents for whom they are reasons. Realists allegedly hold that “we have normative concepts because we’ve spotted some normative entities, as it were wafting by” (1996: 44). The task for metaethical constructivism is to understand reasons for action and our capacity to act for reasons in such a way that their harmony is non-accidental: constructivism proposes to explain normative facts in a way that accounts for how they address us as agents and feature centrally in the exercise of practical reason.⁷

The demand to understand reasons in the first place arises out of the fact that insofar as we occupy the practical standpoint, we rely on them:

Normative concepts exist because human beings have normative problems. And we have normative problems because we are self-conscious rational animals, capable of reflection about what we ought to believe and to do. (Korsgaard 1996: 46)

Simply ascribing to certain facts a special kind of property leaves mysterious why it should appear in our deliberation:

There are certain things that we ought to do and to want simply because they have the normative property that we ought to do or to want them (or perhaps I should say that they ought to be done or to be wanted). The synthesis between the oughtness and the action, or the agent and the oughtness—however that is supposed to go—cannot be explained. It is like a brute fact, except that it is at the same time an a priori and necessary fact. (Korsgaard 2003a: 2)⁸

Realists like Scanlon and Parfit resist the idea that there is anything to be explained. It simply is the nature of the property of rightness, goodness, oughtness, or being a reason that insofar as we have the capacity for practical reason any bearer of the property is a fitting object for its exercise.⁹ One way to motivate the worry a bit more is to place it in the context of a more general concern one might have about theories of normativity: that part of the task for such a theory is to explain not only how it could be that any normative facts were true, but what they have to do with us. Korsgaard’s ‘normative problem’ is an instance of a more general anxiety felt by many philosophers, which I call a fear of normative alienation. The need to satisfy some version of this desideratum underlies many of the familiar challenges to traditional forms of normative realism.¹⁰

The threat of normative alienation appears in different guises: that normative facts could fail to be motivating, that they could fail to be rationally authoritative, and that they could fail to be identifiable. The first of these is probably the most familiar, and associated with Williams’s (1979) ‘Humean’ challenge to motivational externalism about reasons. The thought is that it is a part of the very idea of something’s being a reason for one that it has the ability to motivate her.

The second and third guises of normative alienation are associated with metaethical constructivism. Constructivism explains the reasons we have in terms of the very capacity to recognize ourselves as having reasons, and to act on that basis. Further, constructivism explains why normative facts are knowable for agents in virtue of being agents: because it is the exercise of practical reason that determines what we have reason to do.¹¹

In each case we see a similar structure: an anxiety about the possibility that we could have reasons to which we were motivationally indifferent, reasons whose relevance to our activity of reflective self-determination was at best coincidental, or reasons of which we could in principle be systematically unaware. If it were possible for reasons to be like that, they would be totally estranged from us. These more local challenges to traditional moral realism are expressions of a concern for normative alienation: that morality cannot be alien to us.¹²

This is the real problem for which Kantian constructivism provides an answer: morality is not so alien because it is ultimately explained by our capacity for practical self-determination. Moral reasons are reasons for us, reasons that rationally bind us, because they are grounded in our nature as practically rational creatures.

But what of the standard by which what I’ve called the ‘received view’ proposed to judge constructivism? In the decades since Korsgaard’s Tanner Lectures the metaethical terrain has shifted. Non-naturalist forms of realism have become increasingly quietist, or ‘non-ontological’, and their proponents today tend not to think that their views come with the kind of metaphysical consequences that threaten the modern scientific worldview. Meanwhile, increasingly sophisticated versions of expressivism claim to have recovered enough objectivity to say everything we want to say about normative facts. To the extent that expressivism and realism remain competitors, the stakes are less likely to be defined in terms of balancing naturalism and objectivity, as compared to theoretically sophisticated puzzles about natural language embedding and inference. This is not to suggest that there exists a consensus that these issues are no longer live, but I am happy to ally myself with those who claim to have moved past them, as should be all constructivists.¹³

1.3. Social Alienation

There is another sort of alienation that should worry the metaethicist, which I call social alienation. Much of morality involves getting a grip on each other, agent to agent. But a theory such as the Kantian constructivist’s, which pulls the requirements of morality out of the practical standpoint of the self, runs the risk of erasing this distinctive grip, giving us at best an indirect way of recognizing other people. At least a part of morality is irreducibly social in that it is fundamentally constituted by relations that concrete persons stand in with one another, and in conceiving of agency in individualistic terms constructivism risks being unable to explain this sociality.

The threat of social alienation in metaethics has gone largely unrecognized, and I do not have the space here to develop the objection in any detail.¹⁴ However, we can see some of its intuitive force by beginning with a more familiar version of the same basic worry: the fear that virtue ethics turns morality into an exercise in self-improvement, that even in the realm of what we owe each other (justice, or dikaiosune), all of the normative ‘oomph’ derives ultimately from the need to perfect oneself—to be excellent. As Wallace (2019: 46) puts it, “at the level at which normative requirements are explained, the interests of other people enter as occasions for the realization of virtue, rather than direct sources of requirements on the virtuous agent.” If this perfectionist gloss on virtue ethics is right it amounts to denying that we really do owe anything to one another in the first place.

Korsgaard herself glosses a similar objection to Gauthier’s ‘Hobbesian’ attempt to construct moral reasons, which are essentially public, out of reasons of self-interest, essentially private. She notes that:

If reasons were essentially private, consistency would not force me to take your reasons into account. And even if it did, it would do it in the wrong way. It should show that I have an obligation to myself to treat you in ways that respect the value which I place on you. It would show that I have duties with respect to you, about you, but not that there are things I owe to you. (1996: 134)

In my view the real force of the objection comes from the implicit move from the normativity of first-person practical reason to the normativity of social relations. Korsgaard’s emphasis on the private/public distinction can obscure this, but the last sentence quoted above captures what is really at stake: there is a difference between having a duty that features another as part of its content, and having a duty owed to another.

If I promise myself that I’ll smile at strangers more, my promise concerns a stranger walking by (whose presence triggers a smile), but is not owed to them: if I am feeling grouchy, I do not wrong the stranger by failing to smile (except perhaps by rudeness). I act wrongly vis-à-vis my promise to myself in my conduct concerning the stranger. An obligation genuinely owed to another is different: it is an opportunity to do right by another or to wrong them, not just to do right or wrong.

The technical term for this distinction is that some duties are directed. If one agent is under a duty directed toward another, the other enjoys a special standing as the holder of a right to the first’s performance. When an agent violates a directed duty she does not merely do something wrong but wrongs someone in particular: the one to whom the duty is owed. As Thompson (2004: 340) puts it, the one who is wronged is the victim and not merely the occasion of wrongdoing.

Directed duties are, if not coextensive with morality, at least an important part of it, and a key instance of the kind of sociality that I’m interested in. They reflect what Wallace (2019) calls the ‘moral nexus’ that joins concrete persons, equally real: a fundamentally social relation that cannot be explained in terms of arrangements of individual normativity. A particularly dramatic way for a theory of metaethics to fail, then, would be for it to render our moral requirements in terms that ultimately collapse back in on the self.

Along these lines Tarasenko-Struc argues that Korsgaard “embraces an egocentric conception of authority, on which we originally have the authority to obligate ourselves, whereas others only have the authority to obligate us because we grant it to them” (Tarasenko-Struc 2020: 77). Just as with the simplistic virtue theorist, all of the normative ‘oomph’ of directed obligations derives ultimately, on Korsgaard’s view, from the individual requirements of self-constitution. The core of Tarasenko-Struc’s argument is that Korsgaard needs a way to explain how an obligation that derives from an obligation to oneself will not turn out, on careful inspection, to be an obligation to oneself that merely concerns another, but her conception of the original authority of the self does not have the resources to do so.¹⁵

More generally, Kantians like Korsgaard hold that morality involves relations of mutual accountability with other people, so one might think that Kantian has therefore dodged the threat of social alienation. But these relationships of accountability can supposedly be derived from the idea of the agent as self-legislator, rather than treating them as sui generis. There is widespread skepticism that this derivation can be carried off, but even if it can, in conceiving of relations of mutual accountability as mediated by more basic self-relations, the Kantian risks social alienation. Since Kantian constructivism was in the first instance an answer to how agents recognize moral reasons, it leaves us with only indirect ways of recognizing one another.

What’s important here is not whether this objection is successful exactly as I’ve sketched it, but that we can see it pointing toward a distinctive way of going wrong in metaethics: the possibility of what I have called social alienation. It is worth noting that Korsgaard explicitly has explanatory aims aligned with what I am suggesting, and indeed she holds that there is a role for sociality in, if not agency as such, at least its characteristic exercise: reflecting on essentially public reasons, responding to the call of another (here echoing the constitutive role of the hail—Aufforderung—in Fichte’s understanding of self-consciousness).

However, there are two reasons to doubt whether Korsgaard can accommodate a robust understanding of the sociality of morality: first, because it’s not clear that her theory of agency is up to the task of grounding the kind of sociality she appeals to in lecture 4.2 of Sources. Second, the sense in which agency is social for Korsgaard is, so to speak, inside-out: what it is to be an agent is to be such as to potentially stand in recognitive relations with others, if any others show up, because the reflective relation that one stands in to oneself as an agent (the ‘second-person within’, as she puts it in Korsgaard 2007) is generalizable. But for Fichte or Hegel agency is outside-in: being an agent requires already standing in recognitive relations with actual, concrete others. For Korsgaard the Aufforderung, and the recognition it presupposes, is constitutive of various normative facts but not of agency itself. For Fichte, it is through the Aufforderung that we become agents.¹⁶

I do not claim to have conclusively shown here that Korsgaard faces a problem of social alienation, nor that she lacks the resources to address it, either of which would require an article-length discussion on their own. Here I hope only to have given enough of a sense of what kind of problem social alienation presents to put pressure on Korsgaard (or any other constructivist) to consider it, and to have raised substantial doubt in her ability to meet it.¹⁷

My suspicion—and for now it will have to remain merely a suspicion—is that something like this worry underlies much of the dissatisfaction with Kantian constructivism. As I noted above, the worry has not been explicitly addressed in these terms but I think it offers a plausible diagnosis. Even if the Kantian can establish that agency involves the lawlike form of the will’s exercise, some object, it simply cannot explain how anything recognizable as morality can follow from this alone, because agency, so understood, is a reflective self-relation, and morality involves relations with others. No amount of reflection can bring them into view in the right way if they are not there already.

1.4. The Dilemma

Those moved by the worry about normative alienation tend to gravitate toward more agent-centered approaches to metaethics (constructivism, subjectivism, Humeanism). If the problem is that normativity could be too distant from agency, the solution is to bring it closer. But doing so risks obscuring the social relations that constitute morality’s core. The problem of social alienation suggests taking a more subject-centered approach to metaethics, one that emphasizes, in its explanation of normativity, those with whom morality is principally concerned. The problem of social alienation is relatively easy to address for a realist: facts about relations between agents and subjects, or facts about subjects themselves, are the normative facts that there really are. But a realist, subject-centered account will have difficulty explaining how reasons are reasons for agents.

Constructivists hold that the best way to explain the tight connection between morality and the agents it binds is to think of morality itself as arising from agency. Thus, a constructivist theory will need to include both of the following components:

• [1]  An account of the practical standpoint (generically, agency, though Humeans tend not to use the term)

• [2]  An account of the determining/grounding/explanatory relation between [1] and a normative order (i.e., morality, or, as in Rawl’s canonical constructivist theory, the framework for a just political order)

Kantian constructivism conceives of the practical standpoint in terms of the lawlike form of the will, and its relation to morality as entailment: a particular moral order is determinately generated by explicating what follows from occupying the practical standpoint. But like other metaethical accounts that center the individual agent, constructivists typically limit themselves to bringing others into view indirectly (as a consequence of accepting universal prescriptions, or as the content of a valuing attitude, for example).

To navigate the demands imposed by worries about both kinds of alienation, constructivists will need to both maintain the focus on agency as the source of normativity and conceive of agency in such a way that it relates us to others in the right way. Given the schematic explanatory strategy above, one might expect that the best way to do this would be to go looking for a different theory of agency—the first component in the schema—that builds sociality in from the beginning. The hope would be that such a theory of agency, when used to explain morality (the second component in the schema above), would have a chance of making sense of how the irreducible sociality of ethical life is possible. That way the sociality the constructivist needs to explain will, in some (perhaps abstract) form, have already been accounted for as a part of morality’s foundation.

In other words, if the moral nexus between persons, as a formal matter, is a part of what it is to be the kind of creature whose capacity for reflective self-determination explains the reasons by which it is bound, then it will be no mystery when those reasons turn out to be explained partly by the concrete social relations that creature stands in.¹⁸

Inspired by Fichte and Hegel, who are animated by a similar concern with Kant’s moral philosophy, it’s this suggestion I now want to develop. Rather than appealing to law as the constitutive form of agency, my proposal appeals to mutual recognition of agents with one another, insofar as they are agents. I will call this the post-Kantian theory of agency. In the next section I elaborate this theory of agency, which is better suited than Korsgaard’s to address the social problem of alienation, and in §3 I return to the broader question of how to explain morality based on a theory of agency alone, to see what a constructivist can do with post-Kantian agency.

2.1. What Is It to Recognize Another?

Recognition is a normative attitude. It is an attitude in the sense of a position, posture, or stance: an orientation with respect to something. In other words, recognition is not just a special kind of belief or desire, but a more general way of being-toward its object.²¹ As Wittgenstein (1953: iv.22) famously says, “My attitude towards [a friend] is an attitude towards a soul. I am not of the opinion that he has a soul.”

It is a normative attitude in that it is a distinctively practical kind of orientation, one that manifests in acting toward its object as having a certain kind of practical significance. In characterizing normative attitudes this way, I do not mean to deny that they are cognitive. Though they are in the first instance practical, insofar as normative attitudes at least implicitly classify their objects as significant in a particular way they have an implicit conceptual structure.²²

The key features of recognition are that it:

1. features in or structures practical deliberation;

2. manifests in patterns of behavior and other forms of embodiment;

3. attributes a status to its object; and

4. can be made explicit in judgment, but is not necessarily so.

In general it is natural to use “recognition” in this sense when we speak of recognizing the authority of a person or institution, or recognizing a right or claim as binding or legitimate. What we recognize is not a fact, a property, or an object, though of course, we do in some sense recognize an object under an aspect, or as-of-a-kind.²³ But even insofar as what we recognize is an object under an aspect, the aspect under which we recognize it is not a descriptive property, but a status, a normative significance. Thus, accordingly, to recognize something as authoritative is centrally be disposed to treat it in a certain way, to practically orient oneself with respect to it in light of this normative significance, to deliberate accordingly, and not necessarily to explicitly judge that it is authoritative.

Consider the example of recognizing something as holy. To recognize an object as holy might be to use it in certain rituals, to handle it with greater care, or to stay away entirely. Recognition of holiness is also embodied in ways beyond treating something as holy in deliberating and acting with respect to it: recognition of holiness is embodied in experiencing anxiety when the holy object is mishandled or at risk of damage, in feeling calm or awe in its presence (as appropriate), in having the bodily skill required to treat it as holy, in holding oneself in a certain posture in its presence, and in being habituated to a set of ritual practices.

Such recognition is typically accompanied by the belief that the object is holy, but the recognition of its holiness is neither equivalent to nor exhausted by having the corresponding belief: such recognition is in fact consistent with never making explicit to oneself the belief that the object is holy—one could even lack competence with the concept of holiness, or, in peculiar cases, explicitly deny that the object is holy while nonetheless consistently treating it as holy.²⁴ Recognizing something as holy is, thus, an overall practical orientation toward it as something holy, and not identical to the mere belief in the object’s holiness.

2.2. What Do We Recognize the Other As?

Insofar as recognizing another classifies them a certain way, granting them at least an implicit, practical significance, it has a particular content. In a case where recognition never rises to the level of explicit judgment there is nonetheless some concept in the offing that gives unity to the behavioral manifestations of recognition, a concept that would feature in a judgment making the recognition explicit. When we recognize another, in the sense relevant to post-Kantian agency, we recognize them as something.

Mutual recognition, however, does not classify its object as having a merely normative significance, like the significance of holiness, which demands a certain discrete pattern of treatment. Rather, mutual recognition classifies its object, the other mutual recognizer, as a recognizer, thereby granting it a metanormative significance. It classifies the other as something that can attribute normative significance. In other words, to recognize another as a recognizer is to recognize that its relationship with its environment is significance-granting.

If I recognize you as a recognizer of the holy, and I see you handling an unfamiliar object gingerly, using it to perform an unfamiliar ritual, I will recognize what you are doing as treating it as holy, and I will think ‘ah, there appears to be a holy thing that I missed’, and then go on to treat it as holy myself. Of course, I may come to disagree—I didn’t miss the object’s holiness, it is you who are confused. But in recognizing you as a recognizer of the holy I am at least defeasibly disposed to treat as holy that which I see you treat as holy. I treat you as an authority in matters of holiness, though perhaps an imperfect one.

One might worry that this is simply not plausible. To borrow Anscombe’s famous example, if I see someone I recognize as a recognizer of holy things treating a saucer of mud as holy, I may not be the least bit inclined to wonder whether the saucer of mud is, after all, holy—not even defeasibly. This intuition is understandable but misguided. If one could truly witness such a performance and remain entirely unmoved, this would reveal that one did not, after all, recognize the worshipper as a recognizer of the holy, or that if one had at one point done so, the performance was so implausible that it jeopardized the recognition. That’s what it means to recognize someone as a recognizer, in the relevant sense.²⁵

This helps to explain why I have not opted for the more familiar suggestion that the ur-ethical relation is the (mutual, perhaps) recognition of another’s humanity. The idea of recognizing the humanity in another has a long history of being drafted into doing serious moral work, and it might have seemed like a natural one to reach for here: part of what I’m suggesting is that the kind of sociality that shows up later in the Kantian story—often as the formula of humanity—ought to be promoted to a more foundational role in our understanding of agency.²⁶ Thus one might simply propose that recognition of the humanity of others is itself constitutive of agency, rather than something that follows from and depends on being an agent, the constitutivist account of which makes no reference to the humanity of the other.

But if we understand by “humanity” something like holiness—a normative significance that demands a pattern of special treatment—it will require at once too much and too little of the ur-agents recognizing it. It requires too much because on the most natural sense of the term, and the one that tends to be presumed when it is relied on to help answer first-order moral questions, it builds too much into the constitutive constraints on agency.²⁷ It requires too little because it does not deliver the mutuality we have been chasing, or if it does, it does so at best coincidentally. My recognition of another’s humanity need not have anything to do with their recognition of mine.

Now, it’s possible that the sense of ‘humanity’ in use in Kant’s Groundwork²⁸ may be better understood as a metanormative significance like that of a recognizer—in respecting the humanity in others, Kant commands us to act in light of the recognition that others are reflective, self-determining agents like ourselves, capable of changing the normative significance of the objects of their own wills by taking them to have different normative significance. This would bring the Kantian story closer in line with the recognition I have in mind.

However, even stipulating that to recognize another as human involves something like recognizing them as a recognizer, that won’t be enough to render the mutuality of mutual recognition more than coincidental. For mutual recognition to be a relation that plausibly plays a role in getting agency up and running—in explaining our own reflective relations with ourselves—it must be a relation through which we are awakened to ourselves as individuals, a relation that is bound up with our capacity to recognize ourselves. If you appear to recognize something as holy but it doesn’t seem holy to me, that makes you a defective recognizer of holiness, nothing more (necessarily) than an instrument that delivers occasional false positives. That won’t be enough to force open a space between my immediate judgments of holiness and my reflective role as an arbiter of such judgments. What generates our capacity to achieve reflective distance from ourselves, according to the post-Kantian view, is the distinctive kind of pushback we get from another we recognize as a recognizer of recognizers, and as recognizing ourselves.

This gets to the fundamental disagreement between the Kantian and post-Kantian conceptions of reflective self-consciousness: the Kantian thinks that one can first achieve reflective distance from one’s own inclinations, and then take oneself to be a recognizer of recognizers, thereby attributing that same power to others, without first coming to recognize oneself as recognized by a recognizer like oneself.²⁹ The post-Kantian insists that reflective distance in the first place presupposes the kind of distinctively normative pushback one gets from others with whom one stands in mutually recognitive relations, not just the kind one gets from a malfunctioning recognition machine. On the post-Kantian view we are awakened to ourselves, and thus achieve reflective distance from our own inclinations, by recognizing that we are being recognized, and thus recognizing the other reciprocally.

2.3. On Sui Generis Mutuality

Thus far we have considered the ‘recognition’ component of mutual recognition, and asked what we recognize each other as. In what is sense must this recognition be mutual? If we are awakened to ourselves through recognizing others as recognizing us, doing so cannot simply be a matter of both recognizing each other symmetrically; there must be mutual recognition of the fact that we have recognized each other.

Consider a variation on John Perry’s famous case of the shopper, who, both leaving a trail of sugar behind himself and searching for the person spilling sugar all over the store, eventually sees himself in a mirror and takes himself to be looking at “the messy shopper going up the aisle on the other side,” unaware that it is himself that he sees (Perry 1979: 12).³⁰ Similarly, I might spy, across Schenley Park, one of my friends, apparently looking not at me but into a restaurant. She notices me in the reflection of the window, but because of the reflection’s angle cannot see that I’m looking at her. We both recognize each other, but neither of us is aware of being recognized.

Our symmetrical recognition does not constitute the shared experience of one another that characterizes properly mutual recognition. Properly mutual recognition, as opposed to merely symmetrical recognition is (forgive me!) shared, not paired. Mutual recognition cannot be constructed by stacking individual recognitive states one on top of the other, because it is essentially self-conscious of its own mutuality. The kind of recognition we are after is not just a matter of both parties having matched recognitional attitudes towards one another, but a matter of standing in a mutually recognitive relation.

The implicit conceptual or propositional content is not just that another is a human, a person, an equal—mutual recognition places the other with respect to oneself. Each of us does not only recognize what the other is like but also how they stand to us and thereby we to them. They are not (only) the content of our awareness: they share with us a recognitional nexus. Not every experience of another has this structure, but it is a structure that a satisfying picture of agency as a socially embedded phenomenon must have room for.³¹ A central part of how we understand ourselves is as among others.

One’s practical relation to another and the other’s practical relation to one are not independent but two ways of specifying the same self-consciousness relation-to-other. Consider the shared experience of a sunset.³² Two friends are looking at a sunset together. What’s happening is more than that one is looking at the sunset and the other is also looking at the sunset. It is part of the experience for each that it is shared; it is an experience of looking at a sunset with a friend. But this in turn is not just a matter of one having an experience that includes in its content the fact that another person is also having that experience. The experience is of looking at a sunset with another who is also having the experience of looking at a sunset together.

Recognition is thus not the apprehension of a fact about another but a fact about us, the apprehension of which partially constitutes its object. When one experiences a mutually recognitive interaction with another there are not two symmetrical recognitional states but a single, shared experience of recognition, in which one’s recognition of another is a part of the other’s recognition of them, and vice versa.³³

2.4. Mutual Recognition and Alienation

We are now in a position to understand the foregoing not as the logical construction of mutual recognition, but as its reconstruction. In other words, mutual recognition is not the product of adding layer on top of layer of normative attitudes, but it can nevertheless be explicated partly by decomposing it into these elements. In doing so, we do not see the ingredients that, when combined, yield a mutually recognitive relation, but the key aspects of a relation that is by its nature sui generis.

I’ve introduced this picture of mutual recognition as a plausible candidate for a theory of agency, on the hypothesis that it might give the constructivist a way to chart a course between our two forms of alienation. But if mutual recognition is our picture of agency, don’t we encounter a puzzle: to be an agent must we be mutually recognized as one?

This apparent paradox—that we must recognize one another as agents in order to be agents—is meant to be virtuously circular. The idea is not that we assemble these relations piece by piece, but rather that by the time we are capable of reflecting on our own reflective capacities we find that we already implicitly recognized one another. However it is that the process happens, once we achieve full-blown agency—the capacity to achieve reflective distance from our inclinations, to see them as (potential) reasons for action—we find ourselves as each one agent among many.³⁴

To recognize others as standing in this relation to one involves granting them the authority to interpret, grant significance to, and determine one’s actions, and, conversely, recognizing their behavior as action like our own, which depends on our recognition in the same way that our action depends on theirs.³⁵ Agents, as such, recognize other agents as co-constitutors of a shared space of reasons, as co-determiners of the norms that structure a shared practical reality (what Hegel calls ‘ethical life’—Sittlichkeit). We orient ourselves in normative space with respect to one another as poles of authority. Mutual recognition thus structures all of our practical possibilities, and not just our deliberation and action with respect to a certain object.

Conceiving of agency in terms of mutual recognition puts us in a position to resolve the tension between the demands of normative and social alienation. It addresses social alienation by placing sociality at the ground floor of agency, so that self-conscious relations with others are not something that a constructivist theory will need to go on and explain. As I will argue in the next section, when appealed to in the first phase of a constructivist metaethical theory it will provide the resources to address normative alienation by explaining how the normative order produced by the constructivist theory is properly connected to the capacity we have as agents to respond to it.

3.1. Deriving Morality from Mutual Recognition?

If being an agent already involves implicitly recognizing others as having some basic authority to co-constitute a shared form of life it plausibly follows that we owe them some basic form of respect, from which a recognizably moral order could perhaps be derived. This is the general shape of a view like Darwall’s, though he identifies his view as ‘contractualist’ rather than ‘constructivist.’ His conception of the second-person standpoint (that occupied by every agent as a condition on their freedom, and characterized by a circle of interdefinable notions of accountability, authority, and recognition) plays the role of the first component of a constructivist theory elaborated in §1.4: an account of the practical standpoint. His claim that occupying that standpoint “commits us . . . to the equal dignity of free and rational persons,” (2006: 23) and that “those [second-personal reasons] connected to moral obligation and the equal dignity of persons are what we are committed to whenever we relate to one another second-personally at all,” (2006: 11) functions as the second component: an account of the determining/grounding/explanatory relation between [1] and a normative order. In other words, according to Darwall, to be the kind of creatures we are we must stand in mutually recognitive relations with others, from which it follows that we must respect their equal dignity, which is the substantive basis for modern liberal morality.

Darwall attributes the first claim to Fichte, so the sake of convenience I will refer to any theory that appeals to post-Kantian agency and what it entails as Fichtean constructivism. I do not take the following concerns to be decisive, and for all I’ve said the core argument of this paper could be construed as offering support for Darwall’s project, with which I have considerable sympathy. However, I am not persuaded that Fichtean constructivism is the way forward. While adopting the post-Kantian theory of agency addresses the dilemma I discussed in §1, in relying on the same strategy as the Kantian for grounding morality in agency it risks inheriting some of the Kantian’s other problems.

As I suggested above, much of the skepticism philosophers have regarding the Kantian constructivist project may reflect an implicit concern with what I’m calling social alienation: that morality involves others in a way that agency as self-legislation cannot explain. However, Kantian constructivism faces a number of powerful objections that target its broader explanatory strategy, rather than the details of its account of agency. Broadly speaking, the worry is that in placing on a theory of agency the explanatory burden of accounting for the determinate content of a normative order it stretches its explanatory resources too thin. The worry I want to deal with here is that this objection will apply to any constructivism, no matter its theory of agency.

This skepticism is anticipated by Hegel, in his infamous ‘empty formalism’ objection to Kant (see Hegel 1821: §135R). Street (2012) echoes Hegel in arguing that, setting aside the various moves in Korsgaard’s ‘regress argument’, pure practical reason is just not well-suited to deliver anything so concrete as a substantive morality. If the worry about social alienation targets the Kantian’s account of agency ([1] in the schema above), the empty formalism objection and its more recent descendants target the Kantian account of how agency is related to morality [2]: that the latter is entailed by the former. Call this a deductive account of the agency-morality relation.³⁶

3.2. Developmental Constructivism

Before I elaborate the alternative, developmental account of the agency-normativity relation, it will help orient the discussion to keep in mind the possible combinations introduced by contemplating, for [1], the two conceptions of agency (viz. individualistic and social) and for [2], the two accounts of the agency-normativity relation I plan to distinguish:

Deductive constructivist theories, like the Kantian and Fichtean, hold that determinate normative facts can be derived from an account of agency. A more modest way to conceive of the agency-normativity relation is in developmental terms: that agency develops into or is actualized in ways of being an agent that are not entailed by the formal features of abstract agency, but which nevertheless count as ways of being an agent only insofar as they realize the abstract form.

An example of a developmental relation of this kind might be found in the explanation a capacity offers for its actualization: my speaking English is (at least partly) explained by my capacity for language-use, a capacity that might have been actualized in—developed into—my speaking Spanish or Mandarin, in different socio-historical conditions. For a closer, though perhaps less obvious comparison, consider Kant’s claim in the Doctrine of Right that pre-political rights are merely provisional, and the ‘universal principle of right’ requires us to bring about a civil condition wherein an authority exists to make those rights fully determinate and enforceable.³⁸ At some level of abstraction the same idea is at the core of St. Thomas Aquinas’s legal philosophy, where it appears as the ‘determinatio’ principle. Natural law requires a determinate rule in some domain, but positive law is required to make a rule determinate. This is because the general principle can be realized in any number of different fully determinate and adjudicable rules, and as far as natural law is concerned there is no one in particular that exclusively expresses justice.³⁹

What Kantian provisional right and the Thomist determinatio have in common is that they propose a normatively fundamental level at which a nascent but indeterminate and provisional normative order is established, to be realized concretely in a set of practices, norms, relationships, and institutions. Exactly how a rightful society is to be ordered is not laid out by abstract right or natural law as such, providing for a deductive pathway from first principles to determinate answers in specific cases, but neither do reasons depend purely on individual desires or whims: the reasons a person has are grounded (at least partly) in the form of social life they participate in, and such reasons would be unavailable—or indeterminate, which comes to the same—in abstraction from it. Someone living in a rightful condition who wants to reform the scheme of property rights may legitimately do so, but so long as the extant scheme is within the range of what the Universal Principle of Right requires, they cannot simply ignore it and govern their behavior according to a scheme they take to be better.

Of course, neither Kantian provisional right nor the Thomist determinatio is a developmental constructivist theory in the sense that I’m after—they both begin with normative concepts or principles that are themselves unconstructed, and then propose a developmental process by which those concepts or principles are realized within a particular domain. They could, however, be understood as examples of what Street (2008) calls ‘restricted’ constructivist theories, which determine the normative facts within a specific domain, on the basis of some set of ‘grounding’ normative judgments that are not constructed.⁴⁰

In tracing the movement from the restricted developmental constructivism of Kantian provisional right and the Thomist determinatio, to the unrestricted developmental constructivism of the Hegelian, we can look to Street’s own approach. She holds that the content of the practical standpoint is supplied from without—in the first instance, by an agent’s normative attitudes, but these in turn have a causal-historical basis—and that what it is to be an agent is to be a particular, concrete agent, essentially embedded in an evolutionarily, socially, historically conditioned context. To be an agent is to be a realization of abstract agency, which develops into different concrete forms in different human creatures as their evaluative attitudes are systematized.

The process by which the grounds of construction determine normative facts is deductive in the sense that the procedure of scrutiny involves applying logical inference rules, but given the wide scope of the normative requirements of consistency the resulting set of normative facts is not determined by the procedure. Rather, there are any number of different ways a given creature can realize Humean agency by developing the abstract form of Humean agency—rationally coherent normative judgments—into a concrete set of reasons (or correct normative judgments, as the case may be). Street herself does not have the deductive/developmental distinction in view, but the contingency, path dependency, and historical nature of individual agency on her view are the hallmarks of developmental constructivism.

The Hegelian view is that mutually recognitive social relations as such, like the standpoint of a valuing creature as such, have no particular content, and do not determinately establish the social formations in which they are realized; rather, mutual recognition develops into different concrete forms of social life, which in turn provide the actual content required to fix the normative facts.

Recall from §2.4 that mutual recognition involves recognizing another as the co-constitutor of the normative scheme that informs a way of life. The claim, then, is that we do not plug this foundational status into a procedure of scrutiny that lays out by a series of implications the answers to questions about rights and duties and so on, but look instead to how it is realized in concrete social formations, and to what rights and duties are bound up in those. In other words, we do not stand in bare recognitional relations with others—not as a constitutive feature of agency nor as a separate matter—but we do, necessarily, stand in some recognitional relations of particular kinds, kinds given their texture from the forms of life in which they, in turn, are embedded.

We recognize one another as friends and family members, as colleagues and co-citizens, as church members and sports fans and as participants in shared projects. And, as importantly, we recognize ourselves as friends and family members, as colleagues and co-citizens, as church members and sports fans and as participants in shared projects. And we do so through being recognized by others. To pick at one of the less immediately plausible examples: I may privately enjoy Taylor Swift or K-Pop, but I’m not a part of the ‘Swiftie’ or K-Pop community if no one else in the community knows that I am. If I prefer to keep to myself and enjoy Taylor Swift or K-Pop on my own, that’s fine—not everything I do, not every part of how I understand myself—is mediated by the recognition of others (or at least not directly). But a lot of it is, and the Hegelian thought is that I can’t be an agent at all without being bound up in some recognitive communities: even my private enjoyment of Taylor Swift presupposes my involvement in recognitive communities at some level of generality: linguistic, political, economic, etc., if not anything more specific.

This suggestion echoes Korsgaard’s use of the ‘practical identity,’ here seen not (just) as a conception of oneself through which one deliberates, but as a conception of oneself through which one understands oneself among others, and as understood by them. Some if not all of one’s practical identities make essential reference to involvement in shared projects or practices, membership in communities, and so on. They are given their determinate content in virtue of being situated within networks of recognitive relations, which, in turn, bear the traces of their own peculiar histories, rather than being derivable in full from first principles.

On this much Korsgaard would presumably agree, though it would not play the same role in her theory.⁴¹ Where the Hegelian story departs from Korsgaard’s is in the relationship between more local, bounded practical identities, like fandom or family, and more general practical identities, like humanity or personhood. For Korsgaard it follows deductively from having any practical identity at all that one has the practical identity of humanity, from which it follows in turn that one must respect the humanity in others, and this is the source of our moral duties and rights. This feature of Korsgaard’s view more or less captures the Kantian commitment to the significance of pure practical reason—practical reason as such, and not as realized in any contingent creature or context—as the source of moral reasons.

For the Hegelian, however, this explanation is backwards: we do not occupy maximally general practical identities as a logical precondition of occupying more specific ones, but as the hard-won and still-incomplete result of a historical process through which more parochial communities become knitted together. And, further, we do not derive specific duties and obligations from the very idea of occupying maximally general practical identities, but find rights and duties woven into those collectively constituted standpoints through the process that created them, which, because it is open-ended, leaves those rights and duties subject to continual negotiation. What it is to recognize one another as co-members of, say, humanity, is determined by what it is for us, here and now, to see ourselves as members of this community, as a precondition of being, not agents as such, but the agents that we in fact are.

3.3. Mutually Recognitive Forms of Life

That is the Hegelian view in broad overview, but it will help to work through a few examples at different levels of generality to see it in action. The core claim is that Hegelian constructivism identifies 1. forms of social life, 2. in which mutual recognition is realized, 3. and through which normative facts are fixed, and holds that to be an agent is to stand in social relations that realize abstract mutual recognition, which develops into different concrete forms in different human social groups as their practices, norms, relationships, and institutions are systematized. So, consider one extremely local institution, which comprises a variety of relationships, norms, and practices: the courtroom. The roles and attending duties and obligations are, in a courtroom, formalized and explicit, which makes it a helpful example, though of course in many contexts they will be much less so.

A judge is entitled to decide whose turn it is to speak and when an objection will be upheld, to instruct the jury or, in some cases, to direct them to find one way or another. A judge is not permitted to decide what the law is, and in certain circumstances a judge is not permitted even to decide how a law should be interpreted: when a new judge takes over for a previous judge in the middle of an ongoing litigation, the second judge is bound by the first judge’s interpretation of any point of law that has been previously applied, for example in deciding a pre-trial motion. Likewise a witness is entitled to testify as to events they have observed, but forbidden to repeat hearsay, or to testify as to the character of a defendant for the purpose of proving conduct.

What makes a witness a witness is that they have been called to testify by an attorney, and are recognized as a witness by the judge and jury. Someone who wanders into a courtroom while a trial is in progress, makes it to the stand before being forcibly ejected by the bailiff, and begins to relate their personal experience of events that happen to be relevant to the case, does not thereby become a witness. Their story is not testimony unless they have been authorized to testify by those others in the courtroom who recognize them as a witness, and whom they recognize as judge and attorney. A judge cannot pass a sentence if no one else recognizes them as a judge.⁴²

These are roles that one contingently, temporarily, and to varying degrees voluntarily occupies, so they are not attractive candidates for basic, identity-constituting roles for the agents that occupy them. What they are, however, are roles easily seen as embedded in (some region of) a form of life, roles that realize mutual recognition in parochial ways, and roles that generate determinate normative facts. The point generalizes. We have varying degrees of control over whether we are judges or attorneys, students or teachers, friends or family, members of the neighborhood association or of the online Swiftie community. We have some control, though considerably less so, over which legal orders we are bound by, which countries we are citizens of. But to whatever extent we continue to occupy and, insofar as it is not equivalent, to identify with these roles, they represent ways that we stand in recognitive relations with co-members, and they answer (some of) the questions about what we must, may, and may not do.

We may abandon our friends, quit our jobs, relinquish community membership, or flee our countries—less dramatically we may not identify with our citizenship status, and we while still legally bound by the laws of the state we may then have no more than instrumental reasons to obey them—but we cannot cease to be anyone to anyone and still be who we are.⁴³ Severing every severable tie to a recognitive community would, first, leave one with a radical underdetermination of reasons, and, second, reveal additional layers of identity, no less constituted by recognitive communities, albeit likely in less obvious ways. A hermit, for example, is not a person who lives in a solitary state of nature, prior to and untouched by society, but someone whose understanding of their self is mediated by a host of social schemas and norms, and to some extent self-defined against participation in ordinary social life.

What’s left when the more obviously contingent and more straightforwardly voluntary community memberships are stripped away is, for any of us, personhood. But, importantly, on the Hegelian view I am proposing here, being a person is not a maximally general moral status that serves as the foundation for everything else, but membership in a historically constituted and still-incomplete community.

This last claim has two components: first, that to be a person is really to be a person in relation to other persons. As Thompson puts it, “The judgement X is a person . . . is essentially a ‘de-relativization’ of the prior bipolar judgement X is a person in relation to Y . . . ‘Recognizing someone as a person’ is registering her as a person in relation to yourself; it is the appropriation of such a proposition in the first person” (2004: 353). Second, what fills in the determinate normative content of the personhood relation is a socio-historic process. I don’t have the space here to go into detail regarding the first, and Thompson does so elsewhere, so I will leave it for now as dogmatically proposed. As to the second, the thought is that the concept of personhood is one arrived at through generalizing what was first a religious and then a civic notion—equal standing within a community, without a specified social role that set one apart from another. Not by discovering what lay beneath more parochial recognitive communities all along, but by working through the contradictions generated by their coexistence within a single, shared social world.

By way of contrast, consider the form of social unity proposed by the Analects: a moral code consisting of proper performances associated with parochial roles, tied, in turn, to basic moral relationships, like parent-child, or subject-sovereign. This is not an armchair anthropological claim about whether the form of life of the Zhou dynasty included the concept of personhood (much less whether there were persons living there and then—there were), but a conceptual point about the possibility, or at least the provisional coherence, of a society structured exhaustively by such relationships, without a more general, basic (though still relational) social position available. In that context, someone who ceased to occupy any of the enumerated roles would in an important sense cease to be anyone. There would be nothing that it was proper for them to do.⁴⁴

For us, here and now, produced by and ensconced in the modern liberal order, to be an agent at all is to be many things to many others, including, most basically, to be persons to other persons. But this is not a logical consequence of agency as such, it is the way agency-as-mutual-recognition is realized in our form of social life. Personhood is the most general way that mutual recognition is realized in our form of social life, and while its generality leaves plenty of normative indeterminacy for more parochial communities to fix, it provides a baseline of conduct toward one another that we recognize as morality. For the Hegelian, this is a socio-historical achievement, the development of mutual recognition into an advanced social form, and not the logical consequence of practical reason as such. “Hegel’s account of the social and historical structure of recognition permits him to reconcile the sense in which we (all) make the norms we bequeath and the sense in which we (each) are made by the norms we inherit” (Brandom 2019: 264).

Mutual recognition—an abstract form of constitutively social normativity—suffuses and animates the social formations, practices, habits, relationships, institutions, and all the rest of the ‘ethical substance’ that composes a form of life. To be an agent is to stand in relations of mutual recognition with other agents, and what is normative for agents is not simply a matter of how they do or how they must reason as individuals. To be an agent is always to be a particular kind of socially and historically conditioned agent, one whose actual agency depends on the social conditions in which one’s agency is realized.

If the sense of ‘construction’ at work in deductive approaches to constructivism involves an ideal construction in thought, the term takes on a more literal meaning in the developmental context: for the Hegelian, we determine what it means to recognize one another, what recognition consists in, by actually constructing forms of life for ourselves in historical time.