Stephen Darwall's The Second-Person Standpoint has quickly become one of the most discussed recent books in ethics. It richly deserves that honor, since it offers a thorough, intriguing and deeply thoughtful reconception of the entire sphere of morality. And what it has to say is quite new. More than any recent book, except perhaps Robert Adams's Finite and Infinite Goods, Darwall's book challenges the meta-ethical presuppositions of practically every going moral theory. Darwall remains, as he has been throughout his career, strongly committed to Kantianism. But his Kantianism is now one in which a certain relationship – the relationship between an I and a you – lies deeper than even the categorical imperative. If he is right, we will need to read Kant himself in a different light.

Since the book has already been widely reviewed, I shall give just a brief description of its contents. I'll then turn to two large issues I'd like to raise about it.

Darwall begins by drawing our attention to the distinctiveness of the second-person standpoint. From the third-person standpoint – the view from nowhere – the world simply appears to us, as if from no perspective at all; this is the standpoint appropriate to discovering facts and doing speculative philosophy. By contrast, the first-person standpoint gives each of us our ‘take’ on the world: I look out on the world and see it in relation to my feelings and interests and potential decisions. This is the standpoint that seems best fitted for deliberation, the appropriate one from which to exercise practical reason. In the second-person standpoint I stand always and immediately in relationship to another rational being; from there, solipsism is impossible and dialogue is the characteristic form for thought. Darwall wants to suggest that morality properly begins from, and can only be justified within, this second-person standpoint – that it is not an outgrowth of first-personal practical reasoning but a distinctive mode of such reasoning that makes sense only in relation to a you. (We can of course, and ultimately do, address ourselves from a second-personal standpoint: the idea is precisely that we obligate ourselves as part of our relations with others.) This is not an obvious route for a Kantian to take, and indeed Darwall contrasts his position with Kantianisms both of the third person (Nagel's) and of the first person (Korsgaard's). Darwall does not believe that these alternative approaches can succeed: the arguments they offer for morality are, he thinks, in the end unconvincing. Our basic moral concepts – obligation, accountability, respect, dignity and freedom, among others – form a circle, he says, in which each is linked to the others, none can be derived from any more fundamental principle, and all make sense when, and only when, they are seen as presuppositions of the second-person standpoint.

The book introduces all these thoughts in its opening chapter. It then explores a series of topics related to the overall argument but not directly necessary to it. We get a more detailed examination of respect and accountability, showing how and why they are best understood as second-personal phenomena, followed by second-personal accounts of the reactive attitudes; of the way Adam Smith understands sympathy; of some well-known experiments in moral psychology (the Millgram experiments, especially); and of a debate between Hume and Reid over justice. This takes up most of the book, but in its final chapters it returns to make a direct case for the claim that Kantian morality in general, and contractualist versions of it in particular, can be defended only by way of second-personal premises. I am not quite sure why the book takes such a winding path to its conclusions, but I presume the idea is to show the wide range of moral phenomena that the second-person standpoint illuminates. At any rate, the effect of composing the book in this way is to give one a sense of the pervasiveness of the second-person standpoint in the moral sphere, and one comes away with not just an argument to remember but a feeling that the book's thesis has a certain majesty: that it has great scope and is capable of being employed, and expanded, in many different ways.

I am inclined to think that Darwall's thesis is largely correct: a great deal of morality can indeed be best explained from a second-personal standpoint. My own view of the moral terrain is an eclectic one, and I am skeptical of any attempt to ground all moral phenomena in any single feature of our thought or practice, but I am at least as persuaded by Darwall's proposal for such a grounding as by any other. The two issues I want to raise concern ways in which I think the implications of his emphasis on the second-personal standpoint could be fruitfully expanded. I'll ask first why we should not extend the second-personal standpoint into the epistemic realm; after that, I'll suggest that a truly second-personal view of morality might not be quite as Kantian as Darwall suggests.

(1) Darwall tells us several times that there is no analogue of the second-person standpoint in epistemic matters. Epistemic authority, for instance, is not second-personal. ‘In theoretical reason-giving’, he says, ‘the authority to enter into the reason-giving relation to another person ultimately derives from the putative authority's access to facts that hold independently of that relation. . .Second-personal reasons of the kind we are interested in, by contrast, are relational all the way down’ (pp. 59–60; cf. pp. 287–9). Again, Kantian autonomy is a ‘distinctively practical’ form of freedom, dependent on the second-person standpoint, ‘that lacks any analogue in theoretical reasoning’ (pp. 31–2). There are ‘varieties of freedom. . .which even a theoretical reasoner must presuppose’, but ‘nothing analogous to Kantian autonomy of the will seems to be among them’ (p. 281). Darwall rejects Kant's own attempt, in the third part of the Groundwork, to argue for autonomy from the need for freedom in the theoretical realm (pp. 215–16, 224–5, 281). There is not and cannot be a self-given principle like the categorical imperative, he says, in theoretical reasoning. Theoretical reasoning must be ‘object-guided’, responsive to reasons that depend on the properties of the objects of belief, which is to say that it must be heteronomous; it cannot give itself its own fundamental law.

Well, to begin with, Kant himself explicitly proposes an analogue of the categorical imperative in the theoretical realm. In a remarkable short essay called ‘What is Orientation in Thinking?’, he explains the basic rule of thinking that he elsewhere calls the maxim of enlightenment – ‘think for oneself’ – as follows: ‘To make use of one's own reason means no more than to ask oneself, whenever one is supposed to assume something, whether one could find it feasible to make the ground or the rule on which one assumes it into a universal principle for the use of reason’ (as translated by Allen Wood and George di Giovanni, in Religion within the Boundaries of Mere Reason and Other Writings, 1998). If everyone followed this test, Kant says, superstition and enthusiasm would fall away; we would no longer accept beliefs for which we could not, at least in principle, give our own reasons.

This is a passage that Darwall does not address, and it is worth dwelling on a bit before returning to his own position. Two of its features are particularly worth noting. First, Kant does not say that the test he proposes would lead us to knowledge. Rather, it would lead us to enlightenment, which is for Kant both a condition for knowledge and a morally important state in itself: a constituent of moral and political freedom. So, whether or not knowledge intrinsically involves a certain orientation towards objects, and hence heteronomy, there is at least a mode of believing so heteronomous that nobody who aims either at knowledge or at morality ought to engage in it. And the principle ruling out that sort of believing seems to have precisely the same form as the categorical imperative: indeed, it looks like a directive meant to unite practical and theoretical reason, and explain why they are and must be combined, as Kant stresses in his Critique of Practical Reason.

Second, the test Kant proposes here is directed to what we can accept, and to the rational basis for accepting it, not, immediately at least, to the truth of what we accept. We are, that is, to ask ourselves first and foremost not, ‘Is this proposition likely to be true?’ but ‘What happens if I shape my beliefs in general in accordance with the type of evidence or argument being offered for this proposition?’ In context it is clear that the sort of case Kant has in mind is one in which a person asks us to accept a religious belief on the basis of his authority – on the basis of his position in a religious hierarchy or supposed access to special supernatural experiences we do not have ourselves. Here we are asked to make a choice, with effects that ramify beyond our cognitive activities, about the entire method we use to investigate the world. We are not simply, with such a method in place, evaluating a particular proposition for truth or falsehood.

With these thoughts in mind, we can return to Darwall's philosophical case for keeping the theoretical firmly distinct from the practical. Darwall makes two main points: that theoretical reason ‘must be guided by the very world that provides [beliefs about the world with] their objects’ (p. 281), and that the principles that structure such reasoning can therefore only be very thin ones, concerning the formal relations among beliefs rather than their content (pp. 215–16). Practical reason, by contrast, can and should be wholly responsible to its own structuring principles: generated out of its own groundwork, which for Darwall is the second-person standpoint.

But surely the distinction is too sharp here. Of course beliefs should be world-guided but our reason can and must autonomously determine what counts as being world-guided and the principles that flow from this determination can be quite substantive, as Kant himself argued in the Critique of Pure Reason. Putative empirical evidence that doesn't conform to causal laws will be rejected, and that is not simply a formal point (it certainly isn't just a matter of logical consistency: supposed evidence for magical forces will for instance be ruled out even if a magical picture of the world violates no logical rules). The bald say-so of others, and any appeal to experiences that only specially equipped human beings (‘geniuses’, spirit-seers) can have, will also be ruled out, especially on the version of Kant's epistemology that we get in the ‘Orientation’ essay. We don't choose whether to believe a particular proposition or not, but we do make certain choices about the sorts of things to count as facts – think of the rules in a court-room about admissible evidence – or about what to believe when neither position on an important question can be established definitively by reason or empirical evidence, as in the case of the religious issues Kant was addressing. Moreover, to some extent moral and political concerns come in, and should come in, when we determine how the fact-gathering institutions in our society should be run. These institutions will inevitably have a good deal of power, so choices about how to structure them should not be made simply on the basis of what would make them most effective as fact-gatherers. That these choices, and our choices, as individuals, about whether to heed a representative of a particular fact-gathering institution, should be made as much as possible to foster individual autonomy is surely just what reason demands of us.

If we look in the opposite direction, moreover, we may question whether practical reason can be as wholly independent of the world as Darwall suggests. Second-personal relations alone cannot constitute morality. Our second-personal relationships will, rather, have to provide us with a way of evaluating and responding to the facts in the world around us. Darwall returns often to the example of one person stepping on another's foot, contrasting how the moral landscape looks if we take the problem here to arise simply from the existence of a pain in the world, on the one hand, with how it looks if we take it as arising from a demand you might make of me to remove my foot from on top of yours. The point is a good one, and I agree that morality enters the picture most clearly when we place your pain in the context of a relationship between you and me. But the pain must remain in that picture. If you ask me not to step on your foot, it is only because I know that that hurts that I see the reasonableness in your request of me not to do it – and that I, therefore, grant you second-personal authority over me and oblige myself, in the course of feeling obliged by you, to remove my foot. If you were instead to issue the command ‘Move your foot!’ quite arbitrarily, when my foot was nowhere near yours and doing nothing that I could reasonably see as disturbing you, I would ignore you, or accuse you of arrogance. So the facts of the world are going to have to play a role in morality even once we've got our second-personal structures of reciprocal recognition up and running. If it is heteronomy for our reasons to be world-guided, then they will be heteronomous even here. And if the responsiveness to facts our reasons need even here is not enough to make them heteronomous, then they can be autonomous even in the theoretical realm.

I suspect that the fundamental difficulty here is one that Kantian accounts of morality often face: how to get concrete content into their very formal, abstract principles. If notions like accountability, respect and recognition alone are supposed to constitute the moral realm, it will be virtually impossible to determine what we ought to do in specific cases. A cognitive or emotive stance by which I enter into relationships of reciprocal recognition tells me no more about what I ought to do to show my respect for others in practice than does the most formalistic readings of the categorical imperative. If, on the other hand, we can somehow build sensitivity to empirical facts into Darwall's structures of reciprocal recognition, much less a role for the historical and cultural circumstances that shape our concrete conceptions of ourselves, we may wind up with something that will look heteronomous to Kantians. This is a dilemma that Kantian systems generally face, and Darwall does not appear likely to do better with it than his predecessors have done.

(2) My second worry also has something to do with content, but is best brought out by a note on style. This is a very third-personal study of the second person. One might have expected Darwall to write a book like Kierkegaard's Philosophical Fragments, which explicitly addresses the reader at various points, and throughout problematizes the relationship between author and reader, or perhaps, like Martin Buber and Emmanuel Levinas, two prominent earlier spokespeople for the importance of the ‘you’, to employ a cryptic, poetic style which gestures at the difficulty of capturing what goes into the standpoint that constitutes ‘you’ from outside that standpoint. Instead, like most philosophers in the anglophone tradition, Darwall generally discusses his subject as if from the vantage point of a disembodied reason: as if he occupied the view from nowhere, the third-personal stance from which, in morality, he wants to pry us away. Indeed, he shows no anxiety about the possibility that something about the second person may go missing when it is considered from the perspective of the third person, which arouses the suspicion that he is less interested in second-personal address itself than in the presuppositions for such address. The presuppositions of second-personal address, like any other set of concepts and claims, can be studied third-personally; the nature of second-personal address may not be so readily accessible.

The reason for that is something Darwall never mentions throughout the course of this long book: that taking up the second-person standpoint requires us to open ourselves to the possibility that a ‘you’ might tell us something we would never have come up with on our own. How do I enter into a true relationship with you? Well, one thing my taking up such a standpoint presupposes is that I see you as similar to myself in many ways, and as having equal standing to address me. There is no second-person address unless you and I reciprocally recognize one another, see one another as having some sort of equal dignity, hold one another, and ourselves, accountable for what we do, and assume that we are in principle capable of understanding one another. Darwall lays out this side of the second-person standpoint extremely well. But another presupposition of the second-person standpoint is that I see you as different from me. Otherwise I will simply project myself into you. And that will not be a relationship, just a solipsistic fantasy of one (a temptation to which Fichte, one of Darwall's favorite sources, sometimes succumbed). I will not, in that case, really take up the second-person standpoint at all, just extend my first-personal view into you: I will address you as if you were another ‘I’, not an independent being. There is a delicate line to be trodden here. If I don't assume that I am fundamentally similar to you, there will be no bridge between us, no basis for a relationship, but if I assume that I am only similar to you, there will also be no relationship, no way for us to recognize each other as distinct persons. Hence the great emphasis, in Levinas and his followers, on seeing the you as an ‘Other’ – capitalized, or preceded with the word ‘ecstatic’, to bring out the emphasis – whose expectations of me cannot be wholly anticipated by any of my own thinking, including my philosophical thinking. The second-person standpoint opens one to experiences and demands that cannot be captured in any theory, philosophical or otherwise, given the third-personal nature of theory itself.

I'm not sure what implications this point might have for the content of morality, but it could have quite drastic ones. A truly second-personal morality might, for instance, take different forms in different second-personal relationships. If I properly recognize you only when I both see a deep sameness between us and see you as a distinctive person who may demand or expect of me something different from what ‘any person in general’ would demand, then the content of my moral relationship with you may depend on your distinctive features. Adultery may be acceptable – morally acceptable – in the context of one I–you relationship but not in another. A degree of irresponsibility may be permitted by the moral rules structuring one I–you professional relationship but not by the rules of a different one. One pair of friends, in mutual respect, may tolerate a considerable amount of lying on each other's part but resent ever being scolded; other friends, equally respectful of one another, may welcome mutual scolding but view any dishonesty as incompatible with further friendship. If morality arises out of the demands of the second-person standpoint, it may need not only to be constructed but to be constructed anew, and differently, between each new pair of mutually recognizing persons.

I suspect Darwall would be appalled if he thought anything like this might follow from his position. He fervently opposes moral relativism, and the moral principles at which he wants to arrive, here as in his earlier work, are at least roughly Kantian ones: a single set of rules or virtues fostering what should count, universally, as free and equal relationships between persons. It is just that he now wants to derive that morality from the presuppositions of the second-person standpoint, rather than from a theory of practical reasoning. But the presuppositions of the second-person standpoint do not give us the whole content of that standpoint, and we will not, I think, wholly appreciate what it has to offer morality if we do not allow its content to reflect the uniqueness of each person, and, consequently, of each I–you relationship. The second-person standpoint essentially involves an Other – someone different from the I or the neutral voice of the theorist, and something that goes beyond what any theory can contain – and if the openness or indeterminacy of that possibility makes one uneasy, one should stick with the first or the third person instead. I hope that Darwall's important new book will encourage the philosophical community to embrace that openness, as an essential element of morality, rather than shutting it off.