PEA Soup is very pleased to host this discussion on relational morality with R. Jay Wallace (University of California, Berkeley) and Stephen Darwall (Yale). These two leading voices each examine the topic, providing both complementary and contrasting views, which should prompt a lively discussion below.

Morality and the Other  

By R. Jay Wallace 

I still remember my excitement when I first read Stephen Darwall’s The Second-Person Standpoint. I’d assigned it in a graduate seminar I was teaching at Berkeley in the spring of 2007, and I was immediately taken by the way the book framed basic issues in ethical theory. There was something captivating in the idea that morality is fundamentally interpersonal—a matter not of the internal consistency of the agent’s will, or of the impersonal goods that might be advanced through the agent’s conduct, but of how the agent relates to other (moral) persons. It was exciting to think about how moral norms might look if we tried to incorporate this interpersonal dimension into our understanding of them, at the most basic level of interpretation. 

Another thing that fascinated me about The Second-Person Standpoint was the connection Steve makes between moral requirements and practices of accountability. I’d long been interested in these practices, and in my first book (Responsibility and the Moral Sentiments) I offered an interpretation of them in broadly Strawsonian terms, emphasizing the way in which the reactive emotions are connected to the stance of expecting or demanding that others comply with moral standards in their treatment of us. What I had not quite understood, and what Steve makes a compelling case for in his book, is that there is a tight connection between the role of moral norms in our accountability practices and the agent’s obligation to comply with them. We are subject to obligations to comply with moral norms, in some sense, because other parties are entitled to address moral demands to us in the way that is characteristic of interpersonal accountability. 

As impressed as I was by these general themes in The Second-Personal Standpoint, however, I was also a little uncertain how to understand the idea of a second-personal reason that was central to Steve’s development of them. His own discussion seemed to suggest two different ways of thinking about the relational aspect of interpersonal morality. One model, which I have called the voluntarist conception, applies fairly literally to Steve’s example of the sergeant creating an obligation for their troops through the issuing of a command. The idea is that obligations derive from the addressing of a demand by an authority figure, who creates deontic reasons that wouldn’t otherwise exist by exercising a kind of normative power. But there were also hints of a very different picture in The Second-Person Standpoint, a relational interpretation, as I have called it, which understands moral requirements to be obligations that are owed to other persons. On the relational model, obligations are normative considerations that are directed to other individuals; they correspond to claims that those individuals have against the agent to compliance with them, and their violation is not merely wrongful, but something that wrongs the claimholder in particular.  

It seemed to me in reading The Second-Personal Standpoint that some version of the voluntarist model was probably Steve’s favored way of thinking about moral obligation, and he has confirmed this impression in more recent work. Thus he does not deny that there are some obligations that are directed to other individuals. But what makes these relational considerations moral obligations is that they can be made sense of in broadly voluntarist terms, as demands that representative members of the moral community are authorized to address to the agent.  

This approach, to be sure, departs from the literal voluntarist paradigm in certain important respects. That paradigm traces obligations to the actual exercise of a normative power by an authority figure, where it is taken as given that the obligation would not have obtained had the command not been issued. (The students in your course are under a requirement to submit a final 10-page paper by a specific date only because you are their teacher, and because you have in fact made this a requirement in setting up your syllabus.) On Steve’s picture, by contrast, it is not required that anyone actually address a moral demand to you in order for you to be under an obligation to comply with it. What matters, instead, is the hypothetical fact that the representative member of the moral community could legitimately demand that you comply with the requirement. 

I think this is an interesting picture, and agree that moral requirements are ones that could (in some sense) be addressed to agents by other parties, in the ways that are characteristic of our accountability practices. But I think Steve is committed to thinking that norms become obligations for agents because of their suitability to be addressed hypothetically by the representative moral person, and this seems less plausible to me. The elements in the voluntarist picture that make it promising, as a model for the generation of obligations, include the ideas that there is an individual authority figure; that their authority invests them with a normative power; and that they create obligations through the free exercise of that power. But these ideas are either entirely missing, or present only in very attenuated form, in Steve’s appeal to the merely hypothetical demands of an arbitrary representative of the moral community.  

For this reason (among others), I have come to favor a different interpretation of the basic idea that morality is fundamentally an interpersonal phenomenon. According to the approach defended in my book The Moral Nexus, morality is a domain of directed obligations and the claims constitutively connected to them; it is a matter of what we owe to each other, just insofar as we are persons whose interests stand to be affected by each other’s agency. As noted above, I discerned hints of this relational picture in The Second-Personal Standpoint, but it took me a while to see that it might provide the key to understanding the fundamentally interpersonal dimension of morality. Moral norms link us in a normative nexus with other individuals who have claims specifically against us. It is in this way essentially about our relation to the other, and it differs from standards of merely personal attainment and from demands (if there are any) that have their source in impersonal value.  

One impetus for thinking this is that directed duties represent an independent and plausible model for a normative obligation. We understand debts that are owed to other parties to generate deontic constraints on agency, and their functioning in this way has something to do with the fact that they constitutively implicate two different individuals in a complex of directed duties and claims. So an appealing account of the special “reason-giving force” of moral norms seems to fall out of the relational interpretation of them, fairly directly. 

Against this, it might be thought that we need a different model of moral obligation to make sense of the fact that it is sometimes okay not to live up to the literal terms of what we seem to owe to each other; we can permissibly break a promissory commitment, for instance, if a medical emergency arises that couldn’t have been anticipated in advance. Philosophers sometimes speak of situations of this kind as ones in which a moral right or claim is permissibly infringed, and that way of talking implicitly invokes a fundamentally non-relational notion of moral obligation and permission. But the relational interpretation will favor a different way of understanding the cases at issue. The unexpected emergency doesn’t make it all right to infringe a claim that remains fully in place; rather it shows the limits of the claim, and it is permissible to fail to honor to the terms of the promise precisely because you do not literally owe it to the promisee to live up to them in these exceptional circumstances (though you might owe them back-up obligations of explanation and compensation). 

What about the connection of obligation with our accountability practices? I agree with Steve that this is an important aspect of morality, and one that is neglected in most prominent treatments of the subject. The way I would put the point is by saying that it is part of our conception of moral norms that they function both to define obligations in the perspective of agential deliberation and to provide a basis for relations of interpersonal accountability. Our understanding of something as a moral obligation is thus connected to our sense that its violation would reasonably attract blame on the part of others. 

Steve would make sense of this idea by appeal to the fact that considerations become obligations through their suitability to be addressed to the agent by the representative member of the moral community. This is what, on his approach, explains their status as obligations in the first place. On the alternative picture that I favor, it is the relational character of moral requirements that explains their status as obligations in the perspective of agency. But that same feature of them also explains why they provide a reasonable basis for relations of interpersonal accountability. Relational requirements are connected essentially to claims, and claims in turn are considerations that individuals are entitled to assert on their own behalf in their interactions with each other. Thus you are apt to resent it when I have wronged you, and this is a response to my recognitional failure to acknowledge the moral claims you hold against me. 

This approach, to be sure, incorporates a different conception from Steve’s of the relations of interpersonal accountability that structure the moral domain. On my picture, which is relational all the way down, the primary locus of accountability is the assertion of moral claims on one’s own behalf, which can be understood as a way of addressing demands to other parties that they honor the claims one has against them. On Steve’s account, by contrast, the paradigmatic expression of accountability is the (hypothetical) assertion of claims by a representative person on behalf of the entire moral community.  

I of course agree with Steve that there is such a thing as third-party blame or indignation. But the relational approach interprets this as an essentially vicarious activity, which is to be understood against the background of the idea that morality is structured in terms of claims that individuals are entitled to assert against agents in their own person. Third party blame is not the primary case of interpersonal accountability, but something we do on behalf of other individuals whom we understand to have privileged complaints, insofar as agents fell short of demands that were owed to those individuals in particular.  

This essentially relational conception of interpersonal accountability seems to me both plausible and attractive, by comparison with the more impersonal conception that Steve seems to favor. In addressing demands to agents, we are not fundamentally speaking as representatives of the entire moral community. Who elected us to this role, anyway? By what right do we presume to represent everyone? Rather we are speaking primarily for ourselves, and secondarily for other individuals who have claims on their own behalf that have been flouted or ignored. Among other things, this approach provides a helpful framework for thinking about the ethics of blame—the ways we go astray, morally, in holding other members of the community accountable for their actions. Asserting claims on behalf of another is an inherently fraught activity, and it is open to moral objections when it undermines the individual we are endeavoring to speak for, or otherwise impairs our relations to them.

What is Relational About Morality? 

By Stephen Darwall 

There are two quite different ways to think about morality and relationship.  One way, which I associate with Jay Wallace, sees morality as a “nexus” of relations that give rise to “directed” duties or obligations that are owed by an obligor (the obligated agent) to an obligee (the being to whom they are obligated).  Relationship here is something approaching a logical notion.  It need not involve any relating between the relata.  As far at the idea of directed duty goes, I might be obligated to the environment even though it is incapable of relationship in the sense of mutual (second-personal) relating. 

A second approach to thinking about morality emphasizes this second sense of mutual second-personal relationship.  This is the approach I pursue in The Second-Person Standpoint and later writings, including, most relevantly for present purposes, “Bipolar Normativity.”  I argue that deontic morality is relational in this second, second-personal sense, in its nature.  It is a conceptual truth, I contend, that an act is morally obligatory if, and only if, it would be blameworthy to perform the act without excuse.  What brings in second-personality is that Strawsonian reactive attitudes like blame are second personal in the sense that they involve implicit address.  Not actual address, necessarily; their implicitly addressing character consists in the fact that reactive like blame are naturally expressed in language with the second-person pronoun, e.g., “You shouldn’t have done that.  What were you thinking?” 

Strawson says that reactive attitudes are held from the “participant” stance within relationship and that the deontic ones implicitly address demands to their objects.  It is their addressing, relating character that leads me to call the participant stance “second personal.”   This does not necessarily mean “second party.”  Guilt is a second-personal attitude though there is no second party. 

Blame, resentment, and guilt mediate mutual accountability.  They implicitly address their objects and expect reciprocating attitudes in response.  They come with RSVPs.  They hold their objects accountable and expect mutual accountability in response. 

If I am right in my claim that deontic morality involves accountability, therefore second-personal interaction, conceptually, it will follow that deontic morality is about relationship in the second of the two senses I distinguished.  A second-personal approach seeks to account for the phenomenon of directed obligation and duty as a consequence.  The idea of moral obligation “period,” as I say, is distinguishable from the concept of a directed obligation owed to someone.  As a conceptual matter deontic morality might exist—hence moral obligations—even if no obligations were directed.  They would not be owed to anyone.  But they would still entail accountability, but the accountability would not be to any individual; it would rather be to the “moral community” (a presupposed transcendental ideal rather than any actual community) or to any person as its “representative.” 

In this opening post, I would like to lay out what I take it that Jay and I agree about, where I think we disagree, and why I prefer my approach.  

What do I think Jay and I agree about? 

First, I agree with Jay that we have many directed duties, so owed by obligors to obligees.  (I, but not Jay, hold these are always necessarily accompanied by pro tanto moral obligations period.  If I make a promise, I am both obligated to you and obligated period.) 

Second, we agree that directed duties come with a special standing—what I call “individual authority”—that obligees have to hold their obligors personally accountable in a way that makes it fitting for them to feel resentment and demand compensation, but also to forgive.  My theory offers a conceptual or, perhaps, metaphysical explanation of this. 

Third, Jay and I also agree that contractualism offers a promising normative theory of deontic morality.  For Jay, that means an explanation of the moral nexus of directed duties.  For me, it means that plus moral duties period, those not owed to anyone in particular or to any collective.  The details of how this would work in both instances would take a while to lay out. 

Fourth, we agree that among the directed duties are those that are owed to each and every person (or human being or . . . ).   

Fifth, we agree that this last situation differs from what I call moral obligation period.  That’s because duties that are owed to individuals involve their individual authority, and moral obligation period are not owed to anyone, not even to the moral community.  The point is not just that they are not owed to any particular individual; duties owed to everyone do satisfy that condition.  Though they are not owed to the moral community, moral obligations period involve accountability to the moral community and to any person as a representative member having representative authority. 

Why does this last make a difference?  Individual authority involves various things one can do as an individual—e.g., consent, waive a correlative right, seek compensation or not, forgive, and so on.  If moral obligations period were owed to each and every person, everyone would have the relevant individual authority to do some or all of these things.  That is a very different idea than moral obligation period. 

So what, then, do Jay and I disagree about?  

Some of our disagreements have been implicit in what I have already said. 

First, I think Jay is skeptical about all of the following: 

-moral obligation period 

-representative authority 

-blame conceived of as an attitude that is held from a representative, second-person standpoint 

-the moral community, or a representative, conceived as a transcendental idea that is presupposed by deontic moral concepts and judgments 

Second, because he is skeptical about these, I think is tempted to see my view as         

-a form of voluntarism. 

Third, I am skeptical that morality can be adequate conceived as a nexus of directed duties. 

Finally, why do I prefer my view? 

There are a number of matters having to do with the normativity of moral obligation that I think a second-personal view can explain that a “moral nexus” view cannot. 

One thesis that it would be nice to be able to capture is what we might call, following Butler, the supreme authority of moral obligation, that is, the claim that there is always conclusive normative reason to do what we are, all things considered, morally obligated to do.  This thesis about the normativity of deontic morality is sometimes called rationalism. 

The claim that morality has supreme authority in the above sense is not: 

-the claim that moral reasons always override normative reasons of other kinds 

-the claim that a directed obligation always overrides other reasons 

-the claim that a pro tanto moral obligation always overrides other reasons. 

It is the claim that there is always conclusive normative reason to comply with all-things-considered moral obligations. 

I think there are a number of considerations related to this for preferring a second-personal view. 

First, it is not at all clear how to distinguish between pro tanto and all-things-considered duties in a moral nexus framework where there is no such thing as moral obligation period.  This is because neither pro tanto duty nor all-things-considered duty is a directed notion.  This distinction only makes sense with respect to moral obligation period. 

Second, I argue in “Making the Hard Problem of Moral Normativity Easier,” that what explains the supreme authority of deontic morality is the distinctive feature of moral blame, conceived of as an attitude that is held not from any individual’s standpoint but from the standpoint of the moral community or a representative person.  My idea here is that blame, guilt, and resentment all have the same reciprocating RSVP demand-addressing structure.  Where they differ is that resentment is what Strawson calls a “personal” reactive attitude, held from the perspective of a victim perhaps vicariously, as when one identifies with the victim.  Guilt is also personal, but it is felt from the perpetrator’s standpoint.  And blame (Strawson uses ‘indignation’), I argue, is felt from the third-party perspective of the moral community or a representative person.  The content of these attitudes is identical: “You shouldn’t have done what you did and you need to take responsibility for it.” 

Strawson calls this last attitude “impersonal” and says it is a “vicarious” version of personal resentment.  I think this is unhelpful in two ways.  First, blame is not so much impersonal as impartial; it is definitely it not Sidgwick’s “point of view of the universe.; it is an implicitly addressing, so second-personal attitude.  And second, a truly vicarious version of resentment would be expressed from the victim’s perspective: “You shouldn’t have treated me that way and you need to hold yourself responsible to me.”  Blame is addressed from a third-party, second-personal perspective: “You shouldn’t have treated them that way and you need to hold yourself responsible as one accountable person among others.”  The demand it addresses comes from the perspective of the moral community or a representative person. 

Guilt is internalized blame.  If it is genuine, it implies not just that one has violated a moral obligation, but also that one did not have, and could not have had, sufficient normative reason to do it. 

But why is the latter true?  I follow Williams in holding, as he argues in “Internal Reasons and the Obscurity of Blame,” that blame presupposes that the blamed person did not have sufficient reason for acting as they did.  If they had such reason, then there action would have been justified.  But it is conceptually impossible for an attitude both to be blameworthy and justified.  But if, then, it is a conceptual truth, as I argue it is, that an act is morally wrong, all things considered, if and only if it is an act of a kind that would be blameworthy lacking excuse, it then follows that there cannot be sufficient reason to violate all-things-considered obligations and do moral wrong. 

None of this can be explained, I think, on a “moral nexus” view.  If I make you a promise to meet for lunch, then I have given you some claim to my keeping it.  But imagine a Kamm-like situation in which I can some great good (say, save someone’s life in a way I am not obligated to) at the cost of breaking my promise. Although my promise gave you a claim (or something weaker, like a claim to claim, see below),  I would not be doing anything blameworthy despite the fact that I didn’t keep my promise.  Though you still have a claim or something claim-like that requires me to do something to make up my not keeping my promise to you, I have not done anything that is blameworthy even though I lack an excuse.  I can’t have an excuse, after all, since what I did was justified and therefore not wrong.  Only wrongdoing can be excused. 

Now I believe that Jay would prefer to say about such cases that although you would have had a claim but for the justifying consideration, you no longer do when the justifying consideration comes along.  He is, of course, permitted to stipulate a meaning to ‘claim’.  However, I think we need something like ‘claim’ to keep track of the fact I still owed and owe you something.  Imagine that I hadn’t informed you or tried to make it up to you.  Moreover, I can’t see how Jay will be able to avoid the idea of culpability (lacking excuse) in trying to justify his thesis that you end up with no claim, even though you would have had it otherwise.  What does the work in keeping the books, I think, is the idea of culpability.  And blame, again, involves representative authority, not the individual authority that is in play with directed duties. 

Presumably, Jay would want to rely on Scanlonian contractualism to sort out these matters.  But despite his title—What We Owe to Each Other—directedness does not enter into Scanlon’s theory.  The ideas he seems to be working with are moral obligation, right, and wrong period. 

There is a final matter I would like to mention concerning the normativity of moral obligation.  Some people object—I think I have heard Jay make this objection—that blameworthiness cannot do the work I want it to for Euthyphro-like reasons.  We think that something is culpable because it is wrong, not that it is wrong because it is culpable without excuse.  Now up to this point, I have only been relying on the biconditional, that is, on its being conceptually necessary that an act is wrong (violates a moral obligation) if, and only if, it is an act of a kind that it would blameworthy to do without excuse. 

I am, however, strongly attracted to the view that wrongness consists in an act’s being of a kind that it would be blameworthy to do without excuse.  Here is why.  It is the only way I know of explaining why the deontic moral notions—duty, right, wrong, etc.—are normative notions.  Something is a normative notion, I think, if, and only if, its being instantiated entails the existence of normative reasons, or its being the case that someone ought, to have some attitude or to act in some way in the instantiated situation (or, equivalently, I think, that the attitude or action would be fitting in the technical sense that has been in play in metaethical discussions for the past century or so).  Without loss of generality, we can assume a reasons-first position like the one Parfit holds in On What Matters.  Now Parfit also makes the very objection I’ve just referred to and says that if we hold that something’s being wrong consists in its being blameworthy lacking excuse, we will then be unable to say that something is blameworthy because it is wrong.  I’m not sure that is right, but suppose it is.  Parfit holds on those grounds that wrongness is an irreducible notion.  But if that is true, it then turns out that wrong is not a normative notion by Parfit’s own lights, since it does not entail normative reasons.   If, however, we hold that wrongness consists in there being reason to blame lacking excuse, we get the benefit being able to demonstrate wrongness’s distinctive normativity.  It entails reasons, in the first instance, for blame and, in virtue of that, for not doing what is blameworthy. 

These are some, at least, of the reasons I prefer a second-personal account of deontic morality to the moral nexus account.  The former sees morality as relational in the sense of being concerned with relatings.  The latter seems relational in a thinner, less philosophically pregnant, sense. 


By R. Jay Wallace 

Steve’s initial post helpfully lays out some of the reasons he rejects the relational account of the moral that I have tried to develop and defend (and that I summarized in my post). In this response I want to look at three questions that seem to divide us. 

Relating and Relationships 

Steve rightly notes that the relational aspect of my view is very different from that implicit in the story he favors, which emphasizes “mutual (second-personal) relating”. He asserts that the relationship defined by what I call the moral nexus is “something approaching a logical notion”, which might not involve any individuals at its two ends—a duty might conceivably be owed to something, like the environment, that is not even capable of mutual accountability. 

This is misleading, however. It is part of the relational conception I favor that the nexus of directed obligations and claims links two moral persons with each other. Granted, I allow that there will be different ways of interpreting the manifold of individuals who are potential bearers of claims against each other. This might well include some individuals who are not capable of fully reciprocal “mutual accountability”, such as mentally infirm persons or even animals. (Does Steve deny that such individuals might be members of his “moral community”, given their incapacity for “mutual accountability”?) But I don’t think there is a plausible interpretation of the manifold of persons that would assign claims to something impersonal such as the environment. We of course have obligations not to despoil the environment that sustains our common life, but these are owed to other moral persons, not to the trees and to the atmosphere, considered in themselves. 

It is true that the moral nexus, so construed, links persons who might not do much concrete “relating” to each other, in practice. Individuals have moral claims against each other, even if they do not stand in any thick causal or social relationship—indeed, I might have obligations to individuals who do not yet exist and whose lives will not even begin until long after I am gone.  

Steve appears to disagree about this, asserting, for instance, that deontic morality conceptually involves “second-personal interaction” [emphasis mine], whereby moral demands are addressed to the agent. But this seems to me misleading. What is conceptually entailed by deontic morality, on his conception of it, is that a representative member of the moral community would have reason (of some kind—more on this below) to address demands to the agent. But there need not be any actual addressing at all, or what Steve would call “second-personal interaction”. The “relating” that figures in his view thus seems to me no less “logical” or “thin” than the nexus at the center of my account. 

Moreover, there is a sense in which Steve’s account seems to me not to involve a relationship of any kind between real moral persons or individuals. As he sees things, we are accountable ultimately to the “moral community” (as a “presupposed transcendental ideal”), not to the real citizens who individually inhabit it. Granted, real persons can address demands to us, as agents, on behalf of the community in which we all together participate. But in doing this, they are not speaking for themselves, but serve rather as impartial representatives of a kind of moral abstraction. What ultimately matters is not the “interaction” between real persons through which one of them (actually or potentially) addresses demands to the other, but the ideal of accountability to the transcendental community, which is not well-thought of as a relationship of any kind. 

Steve and I are both influenced deeply by Strawson’s conception of moral accountability. In his post, Steve invokes Strawson’s notion of the participant stance, remarking that we take this perspective when we address demands to each other. I of course agree with him that we address demands to each other through our susceptibility to the reactive attitudes. But on my approach, this is a way for two individuals to relate to each other, asserting claims on their own behalf, which constitutively involves “participation” in normatively-structured interpersonal relationships.  

The representative of the abstract moral community, by contrast, does not in this role really participate in the kind of relationships I think Strawson had in mind; rather, they are stepping back from such relationships to adopt an impartial point of view that abstracts from their interactions as one person to another. There is, at any rate, a significant difference between our views when it comes to the interpretation of this notion of participation in human relationships, and the view I favor seems significantly more interpersonal than Steve’s, not less. 

Pro Tanto Obligations and Claims 

Steve asserts that there is no way to make sense of the distinction between pro tanto and all-things-considered obligations on a moral nexus framework of directed obligations and claims. I think he is right about this, and am happy to accept this consequence of my position. I am in fact skeptical about the very idea of a pro tanto obligation. The notion of pro tanto normativity is familiar from the domain of reasons, where (for instance) the fact that a course of action would be valuable along some dimension might still count in favor of doing it, even if there is more reason to choose some alternative action. But deontic notions do not operate in this way.  

Take Steve’s example in which the opportunity to realize some superlative moral good makes it the case that I am no longer strictly obligated to keep my promissory commitment. I suppose he thinks that the promissory commitment represents a pro tanto obligation in this scenario, which obtains despite the fact that I am not under an all-things-considered obligation to keep the promise. But what does this really mean? Are we supposed to think that, insofar as I promised, I am to that extent still obligated to keep my word, which would seem to be the deontic analogue of a pro tanto reason? I do not believe that deontic notions really function in this way, as considerations that contribute to normative reasoning in their deontic guise. 

As Steve anticipates, I would say about his imagined case that the promisee no longer has a strict claim against me to promissory compliance under the circumstances he has described. But he thinks this won’t work, arguing that “we need something like [a] ‘claim’ to keep track of the fact I still owed and owe you something. Imagine that I hadn’t informed you or tried to make it up to you. Moreover, I can’t see how Jay will be able to avoid the idea of culpability (lacking excuse) in trying to justify his thesis that you end up with no claim, even though you would have had it otherwise.” 

On this point, I would say that the exceptional opportunity to do good might well undermine the promisee’s strict claim to fulfillment of the original terms of the promise. But it would not necessarily undermine all claims against the promisor; in particular, it is plausible to think that there will remain claims to timely warning and to compensation in the case in which I decide to take advantage of the exceptional opportunity to realize beneficent ends. I would be culpable for failing to honor these residual claims if I was in a position to meet them, even if I was not culpable for the failure to honor the original promissory commitment itself. The idea that that original promissory claim remains in place, as a ghostly “pro tanto obligation”, is thus entirely dispensable. (For the record, I think Scanlon would agree with me about this, but I won’t argue the case here.) 

Rationalism about the Moral 

Steve argues, finally, that his view has resources to make sense of the supreme authority of deontic morality that my relational account lacks. The account he favors emphasizes the conceptual connection between blameworthiness and the agent’s reasons. Steve follows Williams in thinking that blame (in the absence of excuse) presupposes that its target had sufficient reason to follow the moral norms they in fact flouted. He argues, further, that we can exploit this conceptual presupposition to explain the authority of moral norms if we hold that “wrongness consists in there being reason to blame lacking excuse”, as he himself believes. If an action is wrong according to this definition, then it will follow from its being blameworthy that the agent must have had sufficient reason not to perform it. 

I am inclined to agree with Steve and Williams about the normative presuppositions of reactive blame. Steve thinks a vindication of moral rationalism falls out of this idea if we combine it with the claim that wrongness consists in being blameworthy without excuse. But I’m not so sure. A true skeptic about the supreme authority of deontic morality might grant Steve his account of what wrongness consists in, but deny that any actions are really wrong in that stipulative sense. Alternatively, they might follow Williams in agreeing that we sometimes have good reason to blame people when they act wrongly, but offer a “proleptic” account of why this might make sense despite the literal falsity of its normative presupposition. 

Be that as it may, I reject Steve’s story about what deontic wrongness “consists in” on independent (Euthyphro-like) grounds. His account suggests that acts are morally wrong because they are blameworthy, whereas it seems to me that acts are blameworthy because they are morally wrong. Steve acknowledges this concern, but prefers his constitutive approach because of its advantages when it comes to the question of supreme authority.  

To see what is at stake here, it will be helpful to spell out his constitutive account a little more explicitly. The account holds that wrongness “consists in there being reason to blame lacking excuse”. Blame in turn is understood to involve the second-personal addressing of demands to an agent, where that is implicit in the “Strawsonian reactive attitudes”, such as resentment and indignation. So our reasons for blaming people should be reasons for the reactive attitudes.  

But the constitutive story Steve favors is radically revisionary on this score. Our reason for resenting someone is ordinarily taken to be, at least in part, that they have acted wrongly (indeed, that they have wronged us in particular); this is the consideration that is taken to make the Strawsonian attitude apt or fitting. But of course, that consideration is not yet available to be invoked within the account Steve is offering us. What makes the act wrong in the first place is the fact that there is reason to respond to it with a reactive attitude, and whatever that reason is, it looks to me to be a reason of the wrong kind. 

Even if there is a (non-deontic) reason of some kind for the representative of the moral community to blame the agent, is this really enough for the constitutive account? Maybe there is, in a given case, more reason to refrain from blaming the agent—perhaps they are a violent narcissist who will just lash out and kill the hostages if we impartially blame them for their action. It surely wouldn’t follow that it wasn’t wrong for the narcissist to take the individuals hostage and threaten to take their lives.  

But I think I am still unclear about the details of Steve’s appeal to hypothetical address. I understand how a recognized authority—the platoon sergeant in his own canonical example—might make it the case that an action was wrong by exercising their normative power to create obligations through the issuing of a command. But what seems to do the work in his account is not the actual addressing of a demand to the agent, but the fact that the representative of the community would have reason to do so, and this doesn’t look like it is enough. 

Finally, on the issue of rationalism, I favor the relational approach because I think it promises to shed some light on the distinctive normative features of the deontic moral right. The relational account offers a substantive characterization of the moral right, holding that to be morally right is to be something that the agent owes to another individual or individuals, just insofar as they are persons with moral standing. Directed obligations of this kind intelligibly function within deliberation as practical requirements or presumptive constraints of agency, which is one way of characterizing their normative authority. They also provide a normative basis for relations of interpersonal accountability between real individuals, insofar as they give those individuals claims that they have authority to assert on their own behalf, and the violation of those claims provides a reason (of the right kind!) for reactive blame. 


By Stephen Darwall 

Jay has helpfully laid out three areas of remaining disagreement, which I will take up under his headings. 

Relating and Relationships 

It may be that there is less distance between Jay and my view than it initially appears.  Jay says that the moral nexus of directed obligations “links two moral persons with each other.”  This goes beyond a more abstract notion of a moral nexus between a being who has the directed obligation and one whom the obligation is owed.  It adds the idea that the beings between whom the directed obligation holds are both moral persons.  There is, of course, the further question of what a “moral person” is.  I suspect Jay and I agree in holding that one concept of moral person is Locke’s “forensic” notion of a subject of moral obligations, a being who, as Locke puts it, is “capable of a law.”  As I interpret that idea, it is a being capable of mutual accountability, hence a being with what I call “second-personal competence.”  But that leaves open what is necessary to be an obligee, that is, a being to whom directed obligations are owed.  I don’t know that Jay has said anything about that. 

I think Jay and I agree here also that obligees can include beings who lack second-personal competence.  I’m inclined to agree that we can owe things not just to other second-personally competent beings but to “mentally infirm persons” and “animals” who lack this competence.  As I see it, for that to be the case, it must also be the case that second-personally competent beings must hold themselves accountable for  organizing themselves to act as trustees for those who are not able to claim their own rights.  (Like the Lorax in the Dr. Seuss book who “speak[s] for the trees.)  So Jay and I are both inclined to extend the moral nexus beyond the class of moral subjects.   

My thought, then, is that we both need a story about what such directed obligations consist in.  That is where, as I see it, the idea of accountability and “second-personal interaction” enters in.  Directed obligations consist in the relevant accountability relations that are constituted by the relevant individual authorities. 

I say that I am inclined to believe that we have directed obligations to non-second-personally competent beings.  (I argue that we do in “Animal Value and Right.”)  I am certain, however, that we have moral obligations period with respect to non-second-personally competent beings.  That also, I argue, must consist in the relevant accountability relations, only this time they involve representative rather than individual authority. 

Pro Tanto Obligations and Claims 

What is at issue between Jay and me here?  We are agreed about the kind of case I describe that once the exceptional opportunity arises, I am no longer obligated (all things considered, according to me) to keep my promise, and that I have various residual obligations to compensate, etc., if I will or do not.  The question is what explains these residual obligations?  I say, they arise because of the pro tanto obligation to keep the promise that was created when I promised.  I am not clear what explains them on Jay’s view.  Perhaps what we can both say is that my promise obligated me to keep my promise, unless exceptional opportunities arose, etc.  Perhaps we could add: in which case, I am obligated to give due warning, compensate, etc.  However, if we think of what we are doing when I fail to keep the promise as compensating, I don’t see how to avoid the idea the legal/moral idea of injury in the sense of a violated, or at least infringed, claim. 

The point can be illustrated with Feinberg’s famous cabin case in which I break into your cabin during a very dangerous blizzard to keep myself alive.  What I do is fully justified though I have still violated, or at least infringed, your property rights and so owe you compensation.  Jay would say, I think, that you no longer have a claim to my not breaking in in these circumstances.  But if we say that, I don’t see how we can think you are owed compensation. 

My larger points with the language of pro tanto and all things considered is, first, that it is only when all-things-considered obligations are violated that culpability enters, second, that blame is from the perspective of representative authority, and third, that it is blameworthiness that assures conclusive normative reasons, as we appreciate when we feel guilt and make the moral demand of ourselves, thereby assuming the role of representative of the moral community. 

Rationalism about the Moral 

Jay says he can agree with my claim that justified blame entails insufficient normative reason for doing wrong, hence conclusive normative for acting as duty requires.  He doesn’t say that he agrees with me that perspective of the attitude of blame is a third-party representative standpoint and not that of vicarious identification with the victim.  This is important, I think, because it is only on that assumption that the presupposition of insufficient reason to have done wrong goes through.  I can fully accept that you have reason, from your perspective, to resent something I did, and so I must listen to your complaint, without agreeing with your complaint and internalizing the attitude as guilt.  It is blame that is internalized as guilt, where both involve the implicit addressing of a demand from a position of representative authority. 

Jay notes, however, that agreeing with me so far still leaves a thoroughgoing moral skeptic free to argue that there just are not any actions that are wrong in my stipulated sense, that is, that are blameworthy lacking excuse.   I agree that this is a possible position.  I argue in “Making the Hard Problem of Moral Normativity Easier,” however, that the moral skeptic is now in a difficult dialectical position.  I assume that the moral skeptic we are talking about is not a normative skeptic across the board.  Presumably, they are prepared to accept other normative notions, such as the credible (what there is normative reason to believe), the desirable (what there is normative reason to desire), and so on.  This then would put them in the dialectical situation of having to argue that there can be normative reasons and oughts with respect to other attitudes, like belief, desire, esteem, and so on, but that there cannot be with respect to the reactive attitude of blame.  What could justify that?  If, however, there sometimes are normative reasons for blame, then it will follow that deontic moral notions—culpability, wrong, right, duty, and so on—are instantiated. 

Alternatively, Jay says, the skeptic might accept, as Williams does, that there are “proleptic” reasons to blame.  Such reasons, however, are reasons “of the wrong kind” for the attitude of blame.  They do not bear on blameworthiness and hence on moral right and wrong.  Of course, the skeptic is unlikely to be moved by this.  The real problem for them, as I see it, is the one mentioned in the previous paragraph. 

Related to this, Jay says that even if we agree that there is some reason to blame someone that might be insufficient to justify blaming them, since if we did, they might lash out in response.  Again, this would be a reason of the wrong kind for not having the attitude of blame.  Whether they would lash out or not is irrelevant to whether their act was blameworthy and morally wrong. 

This brings us to the Euthyphro problem.  Jay and I both want to say that an act is blameworthy because it is wrong.  And Jay thinks that I cannot say that if I also say that act’s being wrong consists in its being wrong lacking excuse.  This is like an objection Scanlon faces to contractualism that acts are not wrong because they violate norms that no one could reasonably reject.  Scanlon’s reply, as I understand it, is to say that what makes acts wrong according to contractualism is that they have more specific wrong-making features such that no one could reasonably reject principles requiring people not to instantiate them.  Contractualism aims to provide an account of what makes those specific feature wrong making.   

I argue in The Second-Person Standpoint that my second-personal can be seen as a deeper foundation for contractualism, based on conceptual requirements of deontic moral concepts.  Ultimately, what being a wrong-making feature consists in its being a blameworthy feature, lacking excuse, that is, that there exists reason of the right kind for holding people accountable through moral blame for not instantiating the feature. 

The costs of not holding this but holding, as Parfit does, that wrongness is not itself analyzable in terms of reasons (of the right kind) for blame, but is rather an irreducible moral notion, is that one thereby loses any explanation of how deontic moral notions are normative concepts in the sense of entailing normative reasons for attitudes and actions.  On my analysis deontic moral notions are normative for blame in the first instance and, in virtue of that, normative for action. 

41 Replies to “What is Relational About Morality? – with R. Jay Wallace and Stephen Darwall

  1. Hi both,

    First, let me thank you both for amazing posts about a foundation issue in moral philosophy. I’m on the road, and so this is even less well thought out than usual. And I haven’t worked on these issues much. But I had one initial thought.

    As I understand it, Steve issues one kind of challenge to Jay’s view like this: suppose that there are moral obligations grounded in the value of things that don’t have a perspective or a point of view – natural beauty, for example. Obviously, we don’t owe it to a thing of natural beauty not to destroy or damage it in the way that we owe it to a person not to destroy or damage them. But, as I understand Steve, there is a challenge about whether we can capture this in an appropriate way without appealing to accountability practices as a ground of duties. Jay, if I am right, responds that what is fundamental is that people can assert claims on their own behalf, and not that they will be able to hold them accountable for breach.

    I wanted to suggest a third idea, which I at least initially think might be better than either of these views. The idea is this. Amongst the entities that have value, where that value can potentially ground moral status, only some have moral status. To have moral status, as I understand it, is to have moral significance as the bearer of value, rather than only in virtue of the value that one bears. A person, for example, may have a life that goes better or worse. But a person is also morally significant as the bearer of value; a person matters as such. When an entity has significance as the bearer of value, we have reason to instantiate value that it bears for its sake – to make its life go better for its sake, for example. But we also have a reason just to hope, for its sake, that things go better for it.

    We now have the basis for a contract between duties that are owed to entities and duties that are not owed to entities. Natural beauty is valuable, and that value can ground duties, but there is no bearer of value that is morally significant independent of the value of natural beauty. We don’t hope for a flower’s sake, for example, that things go better for it. So there are no duties owed to flowers, even if it is wrong to destroy them because they are beautiful. This is in contrast with humans, who matter in and of themselves, independently of the value instantiated in their lives. We can owe them obligations because they matter for their own sake. This view explains the difference between directed duties and non directed duties without appealing either to accountability, or the issuing of demands in one’s own voice. And I find it appealing, for one reason, just because it does not rely on these ideas, which strike me as less fundamental; but also because it offers a more natural explanation than either Steve or Jay’s views, I think, for the idea that we owe duties to non-human animals, and that these duties are intuitively directed in just the same way as our duties to humans are directed. But they cannot issue demands or hold others to account.

    I wonder if this view has been considered as an alternative to those you both outline, and if you had reasons for rejecting it in favour of your views?

  2. Hello everyone,

    Thank you very much, Steve and Jay, for this illuminating exchange. It helped my understanding of your views a lot and is full of interesting observations.

    A bit like Victor, I was wondering what you thought about an alternative to both of your views. The third alternative I have in mind plays a role in (some of) care ethics and also in Thaddeus Metz’ work on relational ethics.

    All three views can be summarised as relational theories of ethics. But there seems to be an important difference. The difference, I think, is the following. Your views can be described as focusing on the form that moral relations among persons take. In JW’s case, it’s directed duties. In SD’s case, it’s what is presupposed in practices of accountability among persons. The third view, by contrast, focuses on the quality of relationships, we might say, and it isn’t limited to second-personal relationships. It can also include relationships to beings that are not (yet) persons in the full sense, such as young children, or beings who might no longer be persons in the full sense, say because of illness. It can also include multi-pronged relationships, as in a community, or relationships to animals or plants.

    I wondered how you think about this alternative interpretation of relational ethics. Does your own view extend to cover it? Or is there a more fundamental difference?

  3. Many thanks to Fabienne and Victor for those helpful comments. Very briefly:

    I agree with Victor that many things without moral status are bearers of value, and that there are significant reasons for action that derive from that fact. I even think there might be practical requirements, of a kind, connected to such impersonal values, which come apart from moral status. My idea, however, is that these are not specifically moral reasons and requirements, precisely in virtue of the fact that they lack the kind of relational structure that is characteristic of interpersonal morality, and that helps to explain some of its significant features (e.g. its connection to practices of interpersonal accountability).

    This allows for a kind of pluralism about the normative domain, while hanging onto the idea that there is a conceptual unity to the sphere of the moral, which has to do with what we owe to other individuals who have moral status. There are many philosophers tempted by pluralist conceptions of the moral domain itself (Thomas Nagel and Johann Frick, to mention just two salient examples). They think that some requirements are deontological, and have something like the relational character I describe in my book; but they think that another part of impartial morality cannot be characterized in these terms, but is sensitive e.g. to aggregative considerations that are hard to articulate within a relational conception of the kind I favor.

    Since we all agree that there are some reasons and requirements that are not properly part of the morality of what we owe to each other, an interesting question is whether there is just a terminological disagreement here. Some (such as Frick, perhaps) might think that the specifically moral requirements are not simply consequences of the ascription of impartial value to things of various kinds, but have to do with people who have moral standing. It’s just that they cannot all be characterized in terms of directed obligations and the claims that correspond to them. I see the attractions of this kind of view, especially when it comes to accommodating our intuitions about some hard cases for traditional deontology, but remain troubled by the theoretical disunity within the broader moral domain on this interpretation of it, It seems just to stipulatively mash together different kinds of person-involving considerations that don’t really cohere very well, and that function very differently in normative practice (e.g. in relation to practices of interpersonal accountability).

    Maybe we just have to accept some theoretical disunity of this kind, at the end of the day. But I’d first like to understand better what a more thoroughly relational account of interpersonal morality might look like, to see how far it can be extended to the full range of normative phenomena that we think of as raising issues of interpersonal morality.

    Fabienne points toward a different dimension of pluralism, I think (or at least this is how I’d understand the kind of view she is pointing towards). The relational requirements of interpersonal morality are fairly abstract, and apply to individuals who do not standing in thick or socially salient relationships of interaction and attachment to each other. It is enough if one party has interests that stand to be affected by the agency of another. (For the record, I am willing to include young children, the mentally infirm, perhaps even animals, within the broader “manifold” of moral persons who can potentially be bearers of relational moral claims, even if they are not capable of reciprocal moral agency. not sure about individual plants, though!)

    But of course thick social relationships matter normatively. Some of this can be accommodated within interpersonal morality, insofar as people who stand in thick relationships to each other are vulnerable to each other’s agency in a way that mere strangers are not. As a result, our moral obligations to them might be more exigent. But I think there are also things that we owe to each other, not simply as moral persons who can affect each other, but (e.g.) as friends, family members, fellow citizens, and so on. These are sui generis directed obligations, and they might even conflict, in some cases, with the obligations of interpersonal morality (a theme, e.g., in influential contributions by Williams, Wolf, and others).

    Of course, one might think that morality itself is grounded in a thicker kind of relationship between individuals than merely being susceptible to each other’s agency. (Some, including perhaps Michael Thompson, might even think that a thicker relational basis is a precondition for the application of bipolar or relational normative concepts.) That points toward a different kind kind of relational view, one that is very much worth exploring. But it would be less cosmopolitan conception than the one I was trying to defend in my book.

    One thing that is very fruitful about this whole debate, in my view, is that it brings out some of the different ways a conception of morality might be described as “relational”.

  4. Hi All,

    I have a question about the back and forth on the Euthyphro. On Steve’s view wrongness consists in there being reason to blame lacking excuse. This view is appealing because it enables us to explain the normativity of wrongness in a way that Parfit, for example, cannot.

    Jay has a two part worry about this. He says that Steve’s “(1) account suggests that acts are morally wrong because they are blameworthy, whereas (2) it seems to me that acts are blameworthy because they are morally wrong.” (numbers added)

    On the first point, it seems to me that Steve’s response is sound. His account does not actually entail that acts are morally wrong because they are blameworthy – rather, it entails that acts being wrong consists in their being blameworthy. This does seem like when people criticize Scanlon for giving a bad account of the wrong-makers when he was actually aiming to give an account of wrongness (and its importance) not the wrong-makers. Any way, I wonder what Jay thinks about this part of Steve’s response.

    I want to hear more about Steve’s view on Jay’s second worry. He fleshes it out like this: “Our reason for resenting someone is ordinarily taken to be, at least in part, that they have acted wrongly (indeed, that they have wronged us in particular); this is the consideration that is taken to make the Strawsonian attitude apt or fitting. But of course, that consideration is not yet available to be invoked within the account Steve is offering us.”

    First, I wonder what Steve thinks about it (unless I missed this in the exchanges above). How would he explain wrongness being a reason to blame for example?

    On reflection I am worried about this on Steve’s account. Here are my rough thoughts: If wrongness consists in there being reason to blame lacking excuse, then it is hard to see how an acts being wrong (itself) can be a reason to blame someone. Take a case in which the harm I caused you is a reason to blame me lacking excuse. Steve thinks I wronged you and that the wrongness consists in the fact that there is a reason to blame me (which is true because the harm is such a reason). The harm I caused you is a reason to blame me for what I did. But it is hard to see how, on Steve’s account, my acting wrongly (or wronging you perhaps) itself is a reason for you to blame. After all, my wronging you just consists in the fact that there is some lower level reason to blame me. It seems like all the warrant to blame is generated by the harm (which provides reason to blame), and none is added by the abstract fact that there is some reason to blame me.

    Perhaps this problem can be avoided if Steve moves away from (or clarifies that he does not endorse) a strong reading of the “consists in” claim. Steve’s talk of consisting in implies (to me any way) that there is nothing more to its nature (this is the strong consists in claim)…and then we struggle to see how it can provide a reason to blame. But he might instead say, for example, that being wrong essentially involves or partially consists in there being reason to blame lacking excuse. The idea would be to allow that there may be more to say about the nature of wrongness than that it involves there being reason to blame lacking excuse.

    As with debates about whether wrongness can provide a reason for action without generating fetishism worries, it seems like the thing to do is to say *more* about the nature of wrongness. For example Scanlon could say that for an act to be wrong is both (a) for there to being reason to blame lacking excuse and (b) for the act to fail to embody appreciation for the value of relations of mutual recognition. Or Steve could say that for an act to be wrong is (a) for there to being reason to blame lacking excuse and (b) for the act to embody lack of respect for the dignity of persons. Then Steve has a more plausible account of how wrongness can be a reason to blame and not just consist in the fact that there is some reason to blame. He can say in the harm case, for example, that you have two reasons to resent my action: first, the harm is a reason to resent it; second, the fact that it is wrong itself provides a reason to blame and we can explain that by appeal to the fact that being wrong in part consists in failing to respect your dignity as a person. If it just consists in there being a reason to blame however, it seems to me like Jay’s second worry is a live one.

    Regardless of whether this is off the mark, thanks for the interesting exchange!

  5. Thanks for that helpful comment, Brad. I take your first point, and agree with you (and with Steve) that he doesn’t need to say that what makes actions wrong is the fact that there is reason for the representative of the moral community to blame (without excuse). This is an account of what it is to be wrong, and the wrong-making feature could just be (e.g.) whatever gives the representative person reason to blame.

    If we modify the constitutive claim to say that an act’s being wrong essentially involves there being reason to blame, I think that would help. This is in fact a commitment that Steve and I share, insofar as we agree that it is part of our concept of moral wrongness that an act’s being wrong entails that there is a normative basis (of some kind) for blame and accountability reactions.

    I agree with Brad that the way forward is to say *more* about the nature of wrongness. I do this by giving a substantive characterization of wrongness that explains the normative connection between wrongness and blame. Thus I definitely do not follow Parfit in thinking of wrongness as an unanalyzable primitive notion. The relational account puts forward an interpretation of what it is for an action to be wrong: it is to be an act that you owe it to other individual(s) not to perform, just insofar as they and you are moral persons. (Parfit would say this might be a high-level property that makes acts wrong in his indefinable sense; but I think this is what moral wrongness consists in.)

    This is a pretty thin account, to be sure, though it is consistent with a pluralism about the kinds of more specific considerations that might make actions morally wrong (which include effects on the welfare of other individuals, on their reliance interests, on their interest in being treated fairly, etc.). But my thought is that there is enough structure in the thin account of wrongness to explain both why wrongness is a deontic requirement on agents in the perspective of deliberation, and why it is a basis of accountability relations.

    To spell out the latter thought, here is the simple idea: flouting a directed moral requirement is a matter of wronging another party. But an individual who is wronged has a reason for reactive blame (in the form of resentment), indeed that is (at least part of) the canonical account of what makes resentment apt or warranted. The attitude of disregard for the directed moral requirement is also a recognitional failure in relation to the bearer of the claim, a failure to take them seriously as a bearer of claims, or to honor or acknowledge those claims. It is thus a way of disrespecting the dignity of the claimholder.

    (We have to be careful here: the recognitional failure that gives the other party reason for resentment is not the wrongness of an action per se, but the attitude of disregard for directed moral standards that leads the agent to act wrongly. This possibility is latent in the distinction, which I accept for at least many cases, between what Scanlon calls permissibility and meaning. But the distinction is also a challenge for Brad’s thought that wrongness might consist, in part, in an act’s embodying a lack of respect for the dignity of persons.)

    I am not sure whether Steve will want to take on board some of Brad’s constructive suggestions. A challenge will be to accommodate them to the commitment he affirms at the end of his reply, where he writes that “deontic moral notions are normative for blame in the first instance, and in virtue of that, normative for action”. The “deontic moral notion” that is normative for blame can’t be moral wrongness, I don’t think, insofar as this consists in there being reason to blame. This fundamental idea—shared e.g. with the accounts that Mill and Skorupski offer of moral obligation—looks like it is itself an account of the nature of wrongness, and I have trouble seeing how it might be combined with the different kinds of accounts that Brad gestures toward, and that Scanlon and I would favor. But perhaps I am just missing something here.

  6. Thanks to Profs Wallace and Darwall for the very helpful discussion (and to PEA Soup for hosting).

    I think the two interpersonal or relational dimensions of morality under discussion here are both extremely important. The first dimension is the relation between Arlene and Bret when Arlene owes a moral duty to Φ to Bret. The second dimension is the relation between Arlene and Bret when Bret has the moral standing to hold Arlene accountable for violating her duty to Φ. A moral theory that did not have a good account of the direction of our directed duties or of the contours of our accountability relations would fail to capture core features of our moral lives. So I think all of this stuff is exactly what we should be thinking about.

    I am less persuaded that either of these two dimensions, important as they are, can be fundamental to morality in quite the way that has been proposed–specifically, that we can unpack the idea of moral obligation either in terms of direction or in terms of accountability. In both cases, I think we smuggle in the notion of obligation and so leave it unexplained. I find this unfortunate, because I don’t have an alternative account of what it is for an act to be morally obligatory.

    1. Jay claims above that “it is the relational character of moral requirements that explains their status as obligations in the perspective of agency.” A feature of obligation, as opposed to merely having a reason for action, is that the agent lacks discretion whether to perform the action or not. In The Moral Nexus, Jay claims that:

    “[T]he relational aspect of directed obligations helps us to make sense of this contrast. In a case with the inherently relational structure I have described, one’s reasons for doing something are constitutively connected to claims to performance on the part of another person. The values in which these reasons are based are not purely monadic; they do not exclusively concern the agent, but essentially implicate the person to whom the agent is related, whose own normative situation will be altered if the agent fails to respect the value that is at issue. … But when these features are present, it seems natural that the agent would lack the unilateral discretion to discount the normative consideration that seems to be present with some other kinds of reasons for action” (49-50).

    My worry with this argument is that it seems to me that some non-obligatory reasons for action can be directed towards other people in much the same way that duties can be directed towards people. If so, it isn’t the directedness of the directed obligation that explains its obligatory force. Suppose Arlene has a reason to Φ for Bret’s sake (and only for Bret’s sake), but it’s not true that she has any obligation to Φ. (Perhaps she knows he could use a hand moving his belongings to a new apartment, or perhaps she should visit him in hospital after his surgery.) If Arlene doesn’t Φ, she won’t have *wronged* Bret, of course, because she has no obligation to Φ. But she will have done something bad relative to him: she’ll have disappointed him, or let him down, or *failed* him in some way that falls short of wronging. This suggests to me that Bret is essentially implicated in Arlene’s reason to Φ, and yet, since the reason remains non-obligatory, Arlene retains discretion to act on the reason or not, depending on the balance of considerations that confront her.

    2. I think a similar kind of worry arises for thinking about moral obligation in terms of accountability relations. We hold people accountable for all sorts of things beyond their obligations. We chastise, encourage, guide, and goad people relative to various kinds of normative standards without thinking that they have an obligation to conform to those standards. Bret might exhort Arlene to help him without demanding she do so, and he might appropriately have some measure of mild disdain for her for failing to help him without *resenting* her exactly. So being worthy of these “low-intensity” reactive attitudes is not what violating an obligation (without an excuse) consists in. For that, we require “high-intensity” reactive attitudes: blame, resentment, and indignation, or suchlike. But what does the difference between the low-intensity and high-intensity reactive attitudes consist in if not the antecedent idea that the underlying action either is or is not obligatory? If it is a conceptual truth that “an act is morally obligatory if and only it would be blameworthy to [fail to] perform the act without an excuse” this may well be because being blameworthy, as opposed to being merely mildly disdain-worthy, consists in having had an obligation to not perform the relevant action. And then we need an account of what having this obligation consists in that does not involve circularity.

    So I take that the appeal to Scanlon’s footnote 21 distinction between what makes an act wrong and what its wrongness consists in won’t really help avoid the Euthyphro problem here, because it seems to me that the same problem arises if we just ask about the latter question.

  7. Many thanks for those very interesting challenges, Simon. I’m not entirely certain what to say about them, but here are some preliminary and tentative thoughts:

    1. About the “direction” of some non-obligatory reasons for action: I’m not entirely certain that this is the same kind of phenomenon we encounter in directed duties, which we understand to be owed to another party. There are lots of reasons that connect to things being good for another individual in some way; that looks like what might be going on with Arlene and Bret. It would be good for Bret if Arlene helped him move or visited him in the hospital, and this is certainly a reason for Arlene, which involves or implicates Bret pretty centrally. But it just looks to me like an ordinary value-based reason, where the value in question happens to be located in another individual’s life. A reflection of this might be the fact that there is not really a normative “nexus” between Arlene and Bret in virtue of the facts so far described. There is just the value for Bret, and the reason that that generates for Arlene. Reflecting on this, I’m inclined to think that we find “direction” only in cases that are deontic, where there is a duty or obligation that is owed to the other party, and a claim or entitlement on the other side of the normative relation.

    Against this, you might say that there is something normative (but not deontic) on both sides of the Arlene/Bret relation: there is the reason for Arlene to Phi, given by the value of her Phi-ing for Bret. But that same value might give Bret reason to hope that Arlene will act so as to promote it. So the two parties are linked in a nexus of reasons that are each connected to the value for Bret of Arlene’s doing Phi.

    But I think this is misleading in important respects. Values always potentially generate ordinary (monadic) reasons for action and emotion for parties that are differently situated with respect to them. In our case, the benefits to Bret from her Phi-ing give Arlene reason to do that, and they give Bret reason to hope that Arlene Phi’s. But they give reasons of the same kind, to hope Arlene Phi’s, to third parties who happen to care about what happens to Bret. They in this way proliferate across the community of interested or affected parties, in ways that are not disciplined by relationships between a single individual and another. This doesn’t look sufficiently analogous to the directed obligation case to be fruitfully thought of as normatively-weaker analogue of direction.

    To see this, it might help to think about an extreme scenario. Maybe Arlene’s Phi-ing could benefit another individual plant in some way, and we might think that that gives her a reason to Phi. But the same value doesn’t give the plant a reason to care about Arlene’s Phi-ing, since it isn’t capable of that attitude; and perhaps there are no other reason-responsive individuals in the picture who do or would care about the fate of the plant. In that case, Arlene’s plant-related reason would remain fully in place, but it would coexist with the absence of reasons for concern on the part of anyone else. This suggests to me that the normative consideration at issue, Arlene’s reason, isn’t really constitutively linked with the reasons of other individuals, in anything like the way a directed obligation is connected to the claim of another individual. In the latter, deontic case, you really couldn’t have the directed phenomenon without someone else’s have a corresponding claim, but the analogous point doesn’t carry over to other-regarding reasons.

    Of course, if Bret and other parties care about whether Arlene will benefit Bret by Phi-ing, that itself is something significant that Arlene should take into account. Those reasons belong to other individuals, so to speak, and she certainly lacks discretion to deny them or to discount them in her thinking, per se. But I think it is consistent with this that she might still discount her own reason for Phi-ing in this case, so long as we are clear that the effects of her failure to Phi on other parties will not be sufficient to give rise to a directed obligation towards them.

    I think this aspect of the case is perhaps obscured slightly by bringing in the idea that Arlene’s failure to Phi will be a “failure” relative to Bret, or “disappoint” him, locutions which suggest a residual claim or entitlement to Arlene’s Phi-ing, after all. To make things clearer, let’s imagine a case where what Bret is reasonably well-situated, but would still be benefited in some way if Arlene, a complete stranger, were to transfer $1,000 into his account. I grant that there might be some reason for Arlene to do this, given its value to Bret. I also think that Bret might have reason to hope that Arlene will accede to huis request for the money, and hence be disappointed if she refuses the wire transfer. But that does nothing to undermine the thought that Arlene retains discretion to act on the reason or not, depending on her other reasons and the rest of what matters to her.

    2. The point about low-intensity reactive attitudes is interesting, and I agree with Simon that there are continuities between those reactions and the more blamey reactions that figure centrally my account (and I think in Steve’s, as well). Simon mentions exhortation and disdain, to which we might add such reactions as annoyance or irritation, as well as some of the stuff that philosophers gesture at when they talk about the phenomenon of “answerability” (which is about having attitudes that a subject might in principle be asked to justify or give reasons for). Annoyance or irritation or disdain sometimes shade imperceptibly into resentment, without these reactions being very distinctly separable in our social experience.

    Having said that, I also believe there to be a principled distinction between these reactions, even if it isn’t always salient to us phenomenologically. The higher-intensity ones really do seem more deontic in character, structuring our social relations in characteristic ways; this connects to Strawson’s tendency to associate such reactions as resentment and indignation with the notions of a (normative) expectation or a demand. A reflection of this is the fact that our understanding of something as being “not okay” is linked to our awareness of it as something that another party might reasonably be angry with our doing (where the high-intensity reactive attitudes are well-understood as refinements of social anger).

    So I think the thing to say is just that the connection between wrongness and reactive attitudes only involves the high-intensity, blamey kinds of reactions. There are lots of different things you could do or ways you could be that you might understand to attract annoyance or irritation, or that someone might conceivably exhort you (for their own reasons) to change or give up. What is distinctive of moral wrongness is not that it gives others reasons for low-intensity reactions of these kinds, but that it is constitutively linked to reasons that others might have for high-intensity reactive blame. At any rate, it seems to me that both Steve and I might have to say something like this, about the special significance to moral practices of a subset within the larger category of broadly reactive attitudes. Nor does this move seem question-begging or merely stipulative, since it is part of our understanding of the high-intensity reactions that they distinctively structure our social relations in deontic ways, marking out actions as “okay” and “not okay”.

    Of course, Steve and I will make different uses of this point. He seems to think that wrongness just is the property of there being reasons for representative moral persons to respond with high-intensity reactive blame. I, by contrast, believe that wrongness consists in an action’s being prohibited by principles that specify what we owe to other individual moral persons; and I accept this, in part, because I think it helps us to understand why an act’s being morally wrong provides others with reasons for the high-intensity reactive attitudes.

  8. Many thanks to Victor, Fabienne, Brad, and Simon (and Jay, of course) for all of your comments and questions. Being a little late to the dance, I will try to take them up in the following order. In this post, I will respond to Victor and Fabienne, since they are circling around a similar idea (at least, as I see it) and then respond to Brad and Simon in separate posts.

    Combining Victor’s and Fabienne’s points, it seems to me that there are two separate ideas here that can be distinguished. One is being a “bearer of value” in the sense Victor introduces–i.e., not just having value, but being a being for whom things can have value and for whose sake things can be hoped and done–and the other is being an obligee–i.e. a being to whom things can be owed.

    Victor wonders whether duties to nonhuman animals can be grounded in the former, and Fabienne wonders how these ideas, along with what Jay and I have to say can relate to relations that are not being persons in the Lockean sense and so don’t mobilize mutual accountability.

    An idea I have been working out in some papers on nondeontic reactive attitudes like love, gratitude, and trust and a forthcoming book titled THE HEART AND ITS ATTITUDES is that there is another kind of valuing attitude that we can have that recognizes a different kind of intrinsic value that nonhuman animals and human beings who lack second-personal competence can have, namely love (or care) of a potentially reciprocating kind. Love in this sense is an “attitude of the heart”: it opens the heart to another being in the hope that their heart will be open in return (if only if receiving one’s heartfelt address). Other attitudes of the heart include hope (for a being’s sake), faith, grief, sadness, and remorse (agential sorrow, in Yuan Yuan’s term) rather than guilt. These attitudes are genuine reactive attitudes (Strawson includes gratitude, love, and hurt feelings) even though they are not deontic and do not mediate mutual accountability. They rather mediate heartfelt connection and appreciation of the distinctive value of being worthy of love. They come with RSVPs, etc.

    So this is a way of valuing beings who are bearers of value in Victor’s sense that is not deontic and so does not entail directed duties.

    To get the idea of directed duties to nonpersons, we have to introduce the idea of accountability since there is no such thing as a duty that doesn’t involve accountability. I can’t give the argument here, but I argue that a second-personal framework like mine can indeed ground directed duties to nonhuman animals and human beings that are not moral agents in “Animal Value and Right” in the Korsgaard Festschrift (NORMATIVITY AND AGENCY). The argument makes use of some points that Korsgaard makes in FELLOW CREATURES by putting them into the second-personal frame (and argues that her own arguments can’t work because they don’t get inside the circles of deontic (second-personal) concepts.

  9. I would like to respond now to Brad’s points/questions.

    First, thanks, Brad for such a sympathetic and comprehending response. I think you’ve pretty much gotten the way I am thinking of it right. You have a remaining worry that I can’t say that the fact that an action is wrong is itself a reason for blaming the agent. I agree with that. As I lay out in “But It Would Be Wrong,” (in MORALITY, AUTHORITY, AND LAW) and “Making the Hard Problem of Normativity,” I am a buck-passer about blameworthiness, but not about wrongness WITH RESPECT TO ACTION. The fact that something is in itself conclusive reason not to do it (by virtue of the argument I give about the conceptual connection between wrongness and blameworthiness lacking excuse and blame presupposing conclusive reason not to do the blameworthy action) but it is not itself a reason for blame. The reasons for blame will be the wrong-making considerations (which because wrongness and blameworthiness lacking excuse are the same property) are simultaneously blameworthiness-making considerations.

    I think we all have to be buck-passers with respect to the relevant attitude-worthiness–being blameworthy is not a reason to blame, being desirable is not a reason to desire, etc.–but that is consistent with wrongness itself being a (conclusive) reason not to do the wrongful action. But wrongness is not itself a reason to blame, since wrongness just is blameworthiness lacking excuse, and being blameworthy can’t be a reason to blame.

  10. I would like to respond now to Brad’s points/questions and also Simon’s.

    First, thanks, Brad for such a sympathetic and comprehending response. I think you’ve pretty much gotten the way I am thinking of it right. You have a remaining worry that I can’t say that the fact that an action is wrong is itself a reason for blaming the agent. I agree with that. As I lay out in “But It Would Be Wrong,” (in MORALITY, AUTHORITY, AND LAW) and “Making the Hard Problem of Normativity,” I am a buck-passer about blameworthiness, but not about wrongness WITH RESPECT TO ACTION. The fact that something is in itself conclusive reason not to do it (by virtue of the argument I give about the conceptual connection between wrongness and blameworthiness lacking excuse and blame presupposing conclusive reason not to do the blameworthy action) but it is not itself a reason for blame. The reasons for blame will be the wrong-making considerations (which because wrongness and blameworthiness lacking excuse are the same property) are simultaneously blameworthiness-making considerations.

    I think we all have to be buck-passers with respect to the relevant attitude-worthiness–being blameworthy is not a reason to blame, being desirable is not a reason to desire, etc.–but that is consistent with wrongness itself being a (conclusive) reason not to do the wrongful action. But wrongness is not itself a reason to blame, since wrongness just is blameworthiness lacking excuse, and being blameworthy can’t be a reason to blame.

    Regarding Simon’s objections that we can chastise, goad, etc. people outside the framework of mutual accountability IN THE SENSE OF DEONTIC MORALITY and that we have to advert to the independent idea of moral duty to get inside that framework, I of course agree that these phenomena exist. They are not, however, second-personal (or Strawsonian reactive attitudes) and so are just orthogonal to what I am talking about. I can chastise my dog for peeing on the floor as a kind of discipline hoping (stupidly, as it happens) to goad her into more house-friendly behavior without anything about mutual accountability going on. Similarly, I argue (in “Contempt as an Other-Characterizing Hierarchizing Attitude”) that contempt and disdain are third-personal rather than second-personal attitudes. Advice also doesn’t necessarily involve anything second personal–giving it, like any speech act, is second personal, since it has an addresser and an addressee–but the advice given isn’t. You give someone a reason in this sense by pointing to a reason that would be there anyway independently of its capacity to be given in advice.

    All of these phenomena, as interesting and important as they may be, are just orthogonal to the Strawsonian, second-personal framework. Still, you might worry that there are many places where we blame and invite others to take responsibility (and feel guilt) that are outside what we normally think of as morality, and so we need an independent idea of moral obligation to ground MORAL accountability. Here I am inclined not to quibble over words–we can use ‘moral’ any way we like. However we use it we will need the idea of culpability; that is what is driving my argument.

    Finally, the problem with an independent idea of moral obligation that isn’t analyzed in terms of reason to blame is, as I argue above and against Parfit in “Doing Right By Wrong,” is that it gives us no way of understanding deontic moral ideas like wrong or moral duty as NORMATIVE ideas.

  11. Hello you two! I very much enjoyed reading your exchange amongst each other as well as with my fellow commentators so far. I myself have two comments, the first comment dividing into two parts, one addressed at Jay and the other at Steve, the second comment addressed exclusively at Steve.

    Jay and Steve

    I am interested in the difference between you two as to how to analyze the sort of situation in which you promise to meet me for lunch but then end up rescuing a drowning person instead, thereby missing our lunch date. I wonder whether the disagreement between you two here might not be described as a case of the one’s modus ponens amounting to the other’s modus tollens. What you two seem to agree on in the first place is that you were acting permissibly here, at least insofar as you then make the appropriate amends to me, say by apologizing, notifying me as soon as possible, offering a make-up date, etc. More interestingly, what you also seem to agree on is that the domain of directed or bipolar obligation is hostile to the very notion of aggregation from what holds pro tanto to what holds all things considered. However, you then draw different conclusions from those two shared premises.

    Jay, given that on your view all obligation is directed, the formal opposition between bipolarity and aggregation entails that there is no such thing as pro tanto OR all-things-considered obligation, i.e. the category of aggregation as such does not apply here (although there may well be such a thing as pro tanto and all-things-considered reason by your lights). Given that you acted permissibly here, it follows that the situation cannot be described as one of your having broken a promise and so violated an obligation to me in the first place on your view. Rather, your promise to me had an (implicit and informal) qualification covering those sorts of cases built into it.

    Steve’s worry about this picture seems to be that, if it really is the case that you didn’t wrong me at all, it isn’t clear why you would owe me amends here. Your reply as I understand it is that your obligation to me comes not merely with the qualifier in question but also with an obligation to make amends whenever the qualifier comes into effect (perhaps in its turn governed by its own qualifier). So, when you perform the rescue but then don’t make amends (where the qualifier in turn governing the amends making does NOT apply), there you do wrong me.

    I wonder whether this reply might not unduly flatten the moral structure at work here. More specifically, it seems to me to be internal to the notion of making amends that it respond to the breach of an obligation, which is precisely the thing you wish to deny. Put another way, there seems to be a difference between the situation presently at issue on the one hand and a genuinely conditional promise on the other. Say, suppose that you promise me to meet me for lunch if your old friend isn’t in town and to schedule a make-up lunch date with me otherwise, and suppose that your old friend is in town. On your view these cases would be alike in that you do not violate any obligation to me by not meeting me for lunch today, which is fine as far as it goes, but it does seem that scheduling a make-up lunch date with me counts as making amends in the one case but not in the other. So I wonder how to account for the apparent difference here by your lights.

    (I have to admit that there may well be less of a difference between these cases than I am pretending for argument’s sake, in that you may well be inclined to act in very much the same way in both cases. Say, it would be completely natural for you to apologize to me for needing to reschedule even in the case of the conditional promise. Perhaps the difference here could be saved by arguing that that kind of apology isn’t really owed but is really a matter of supererogation, but, again, I’m not sure.)

    Steve, you on the other hand may well take the exclusion of aggregation from the domain of the bipolar as a strike in favor of understanding the domain of obligation as exceeding the domain of the bipolar, precisely because this generates space for the concept of pro tanto obligation, which then comes in handy in explaining the normative structure of the case at hand: your promissory obligation to me to meet me for lunch was merely pro tanto and is being outweighed by the superior moral good of rescuing the person from drowning, so you aren’t blameworthy for breaking it, but you do owe me whatever you owe me residually given that your promise to me wasn’t simply SILENCED by the superior good. (You could by the way imagine cases of silencing of this sort here — perhaps it turns out that whatever you promised me to do was going to assist me in committing a grave moral wrong, say — and here presumably you don’t owe me anything residually. So all this fits well with your view.)

    However, I wonder whether there is not a problem lurking here. More precisely, I wonder whether you may not in the end be forced to give up on the second premise I take to be shared between you and Steve, that is: the formal opposition between bipolarity and normative aggregation. After all, I take it that your promissory obligation is a bipolar obligation all along. This is not per se to take issue with your thesis that every bipolar obligation is grounded in “obligation, period.” Perhaps this is so. The point is rather that the bipolar nature of your promissory obligation to me isn’t merely accidental to it. What I take it this means is that in the aggregation of your promissory obligation and the superior moral good — with the latter winning out — the former is being aggregated in its character as bipolar obligation: “On the one hand I owe it to Micha to show up for lunch, on the other hand there is this person here who is drowning, preventing a drowning matters more than keeping my promise to Micha, therefore I shall save the person from drowning, missing my lunch with Micha,” and straightaway you act by saving the person drowning (or rather, straightaway after having had more than one thought too many). So here you do seem to aggregate over bipolar obligation. Or am I missing something here?


    In your reply to Jay you write the following:

    “I can fully accept that you have reason, from your perspective, to resent something I did, and so I must listen to your complaint, without agreeing with your complaint and internalizing the attitude as guilt.  It is BLAME that is internalized as guilt, where both involve the implicit addressing of a demand from a position of representative authority.”

    I am a bit puzzled by that proposal by what I take to be your own lights. I took resentment and blame qua fellow reactive attitudes to both carry the presupposition that there was conclusive reason speaking against that which constitutes their object, i.e. the action or attitude on whose account their author is being resented or blamed.

    I therefore take it that the kind of “acceptance” of my resentment towards you available to you here has to turn on either one of two thoughts on your part. Either your thought has to be that I am not aware of all the facts yet. Perhaps you have not yet informed me of the superior moral good that was being served by your breaking your promise to me, say. Or your thought has to be that we simply disagree on the normative merits of the case, that is, perhaps I don’t see the moral good that was being served by your breaking your promise to me as all that superior. In that case the relevant form of acceptance will be one of your agreeing to disagree with me.

    However, all of this might conceivably also hold in a case of blame in your own, “representative of the moral community” sense of the term, and by your own lights at that, I would have thought. That is, when you blame me for despoiling the environment, I might “accept” that attitude of mine without internalizing it as guilt, on the strength of the exact same two thoughts: either I think that you don’t have all the facts yet (what you took to be littering is really my setting up an art installation), or I think that there is nothing wrong with littering (after all, no one is getting hurt here, right? (I might think)).

    The point being, my sense is that, whatever their actual difference, resentment and blame do not in fact differ in the way you are suggesting here. I was wondering whether you agreed with me here. I should say that I don’t think that the issue here runs especially deep. Rather, I take myself to just be nit-picking.

  12. Thanks for that very thoughtful set of questions, Micha. Let me pick up the specific part that was addressed (second-personally!) to me. About Steve’s promise case, I want to say that we don’t need a ghostly pro tanto obligation that remains in place to explain the salient features of it. Even if the promisee has no strict claim to performance of the act originally promised, there might still be some back-up claims in place, e.g. to timely notification and the like. Your write:

    “it seems to me to be internal to the notion of making amends that it respond to the breach of an obligation, which is precisely the thing you wish to deny. Put another way, there seems to be a difference between the situation presently at issue on the one hand and a genuinely conditional promise on the other. Say, suppose that you promise me to meet me for lunch if your old friend isn’t in town and to schedule a make-up lunch date with me otherwise, and suppose that your old friend is in town. On your view these cases would be alike in that you do not violate any obligation to me by not meeting me for lunch today, which is fine as far as it goes, but it does seem that scheduling a make-up lunch date with me counts as making amends in the one case but not in the other. So I wonder how to account for the apparent difference here by your lights.”

    I think I agree with you about the notion of “amends”, which seems to presuppose a breach of a moral obligation (indeed, on my account, the flouting of a directed obligation). It is for that reason that I didn’t describe the back-up claims of the promisee in the case in these terms, but spoke instead about a residual claim to “compensation”. But what grounds this claim, if not the breach of a pro tanto obligation that remains in place?

    My thought is that the source of the residual claims will be the interests of the promisee that grounded the original promissory commitment, and that remain the place after the exceptional opportunity has later emerged to do something superlatively valuable. Maybe these include interests in assurance of some kind (to take one prominent possible story, broadly of the kind that Scanlon has proposed for making sense of promissory obligation), as well as interests in the promisor’s performance of the promised action. The idea would be that these interests might not suffice to ground a claim to promissory fulfillment under the circumstances described. But they remain in place, and continue to have significance for the promisor. That agent, after all, deliberately induced the promisee to rely on them to do what was promised, and the failure so to act affects the promisee’s ongoing interests in reliance and performance. Given these facts, it would (as we might put it) not be reasonable for someone in the promisor’s position to reject a principle that requires that they compensate the promisee for the failure to perform a valued action that the promisee was counting on them to perform. (Note that compensation for such harms is not the equivalent of performance, which is why the original promissory commitment cannot be understood as a directed obligation either to perform or to compensate.)

    Micha presses me to explain the difference between this case and a different one in which my promise is conditional from the start. The actions required of the promisor in the two cases might be identical, but (he suggests) only in the original case, in which the promise was not explicitly conditional from the start, would we be tempted to speak of something like making amends. Since I reject the language of amends for the original case, this challenge is perhaps especially telling. Am I not treating all promissory commitments as having a latently conditional structure, in ways that elide distinctions we wish to draw in moral practice?

    I do indeed think that promissory commitments are always implicitly conditional, in a way. They need to be to accommodate the important fact that circumstances can change in morally significant respects, in ways that we couldn’t have been expected to anticipate at the time the promissory commitment was undertaken. There are moral and personal emergencies, and all parties will want some leeway to adapt their plans in response to them (an aspect of the case that Scanlon accommodates by incorporating an “in the absence of special justification” clause into the principles he proposes to explain promissory obligation). Still, I think there is a difference between the case in which this conditional aspect is merely implicit in the original promissory exchange, and the different case in which the promisor goes out of their way to make it explicit. In the latter case, they are highlighting the condition that would undermine the promissory commitment from the start, and encouraging the promisee to take the condition into account in their own planning for the future. This seems to me to be relevant to our understanding of the exchange between the parties, and potentially to undermine the promisee’s claim to compensation for losses.

    The idea of apology in these contexts is interesting. I agree with Micha that promisors might well describe themselves in both cases as feeling the need to apologize for not being able to e.g. meet the friend for lunch. On my view, this isn’t a literal apology for wrongdoing in either case, since (by hypothesis) there is no directed obligation that has necessarily been breached through the promisor’s action. What we describe as an apology is really a different phenomenon, a way of acknowledging the costs that were imposed on the promisee by the way things have played out, and affirming their ongoing significance. The apology serves to assure the promisee that we continue to take their interests seriously and to express regret that we weren’t able to accommodate them as we originally expected to be able to do.

  13. Hi Jay and Steve, thank you so much for this great exchange! As you know I’m much indebted to and inspired by your work.

    As I’m sitting on the train, let me here try to organize and share some thoughts inspired by the pro-tanto and all-things-considered distinction that was already under discussion and how it relates to both Jay’s and Steve’s respective views on the connection between morality and accountability/blame. As I see it, Jay and Steve are working with different notions of claims. On Jay’s view, claims are to be understood as outputs of a particular kind of practical deliberation (a Scanlon-style contractualist deliberative procedure), whereas on Steve’s view they are to be understood as inputs into practical deliberation.

    Now, I think I side with Steve that claims should be taken to function as inputs into practical deliberation. For one I also think we need to understand claims as inputs in order to make sense of the fact that we can be held accountable by and owe an individual an apology, explanation, or some sort of compensation when, say, breaking a promise (with or) without adequate excuse. After all, I take it, that is precisely what we are doing when apologizing to another, and, moreover, doing so sincerely: to acknowledge that we have violated a directed obligation that has its normative source in the valid claim of the other person in particular. Whereas Jay shares this view when it comes to the flouting of a (directed) moral obligation (that is an obligation all things considered), he seems to have a different story in mind when it comes to a case where an exception has been triggered to, say, no longer keep the promise to do something more valuable, or to keep another promise, or what have you. While Jay thinks that we owe the person to whom we are no longer keeping our promise some amends, this is not because we are still obligated to her, but because of something else that gets triggered in virtue of that person’s interest. For, obligation functions as an all-things-considered notion on Jay’s view and there is no longer any obligation in place. But then, and going back to the idea of a sincere apology and what we’re trying to do with it, it seems as though we are no longer able to apologize to that person, given that we are under no obligation. But that strikes me as counterintuitive. Instead, I think it would be better to say that, following insights from private law, the promissory obligation and with it the claim remains not only in place but normatively still very much alive, despite the fact that it can be permissibly infringed. And this is what explains that we owe an apology or some other sort of compensation even in cases where, all things considered, we were obligated/had more reason to not keep the promise.

    Let me make use of an example to make it clearer what I’m thinking here. Suppose I borrow 100 dollars each from Anne and Bob and promise to repay each of them the following day. As it turns out, on the next day I only have 100 dollars in my wallet and all the nearby ATMs are malfunctioning. What should I do? Let’s suppose that Bob needs the money for some urgent medical treatment, whereas Anne is well off and won’t suffer any setbacks if I repay her a bit later. So, all things considered I should pay Bob back first. At the same time, I still owe money to Anne (I’m still obligated to her), and she can hold me accountable. In fact, there is a real sense of culpability on my part and failing to exhibit some acknowledgment of that fact on my part, say, by feeling guilty to Anne, would amount to a great deficiency on my part. That is to say, at the very least I should apologize to Anne and, eventually (rather sooner than later) pay her back the money.

    I worry that on Jay’s view we cannot really explain this, given that there is, after going through the relevant contractualist deliberation, only one obligation in place, namely that to Bob. And given that claims function as outputs of this kind of deliberation, only Bob, but not Anne has a claim. But then, again, I can’t really apologize to Anne, if what we mean by apology here is the acknowledgment of the infringement of a valid claim and directed obligation. After all, there is none.

    All of this to say that thinking of claims as inputs might be explanatorily more adequate. What do you think about that Jay, especially the case where the obligation to Anne still seems to be very much ‘on’? (I guess I’m also drawn here to Thomson’s view on these issues and her thought that we can only determine what we ought to do all things considered against the prior question of what claims people have against us)

    Now, even though I agree with Steve about that, I’d like to disagree about the nature of accountability and its connection to morality. I think here I’m in full agreement with Jay’s view . It turns out, at least on my reading of Steve’s views on the relation between bipolar and monadic normativity, that Steve will actually also not be able to make sense of the common sense intuition that in the case just described I still owe an apology to and can be held morally accountable by Anne. For Steve seems to deny that I’m culpable in that case and should feel guilty. After all, I have done nothing wrong, period and am not blameworthy. I’m blameworthy and should feel guilty only if I violate my all things considered obligation or ‘moral obligation, period’. But then, again, it seems puzzling why I should apologize to Anne if what I’m trying to do here is to acknowledge to Anne that I have done wrong to her and feel guilty to her for it.

    Thinking about sincere apology again not only in the case of permissible infringement but for an all-things-considered violation (or normative injury in Jay’s language) might be of help here.

    I think Jay and Steve are agreed that an apology is itself bipolar in that it needs to be addressed to the person whom one has wronged. I think that Jay thinks that it’s furthermore bipolar in that the practice of apology in fact presupposes a bipolar understanding of the primary obligation whose violation gives rise to the need for an apology in the first place. And I agree. That is precisely what I think; in order to sincerely apologize I need to acknowledge that what I have done was not merely wrong, period but a wronging of the other in particular. To successfully offer an apology, then, one must a) address the person being owed an apology and do so sincerely by b) acknowledging and recognizing the addressee as the victim of one’s wrongdoing, i.e. as the person whom one has wronged in particular. At the same time, apologies require warrant. After all, we don’t just have to apologize for just anything. What warrants an apology is the fact that what I did was wrong to you and that I subsequently deserved to be blamed and held accountable by you. Again, that apology presupposes accountability becomes clear, moreover, once we realize that what one does in apologizing: to acknowledge one’s wrongdoing for which one can rightly be blamed and held accountable by feeling guilty and reciprocating the resentment of the other person. And it is here that Steve invokes Pufendorf’s point, the thought that in genuinely obligating others and holding them accountable we must presuppose that they can blame and hold themselves accountable for the very same reason, and thus from a perspective that they and we share. Importantly, Steve understands the relevant perspective to be the impartially disciplined perspective of the moral community. As such, I should accept your blame and feel guilty only if I realize that I have done something ‘wrong, period’ and as such something for which anyone can hold me accountable. But I see no reason why this must be so. Nothing about Pufendorf’s point itself, as far as I can tell, says that this shared perspective must be the all-inclusive perspective of the moral community. Instead, we can conceive of the relevant standpoint as bipolar and thus shared between two individuals, like you and I, who are implicated in a given normative nexus. As such, I also think that there is no reason to think that blame and guilt cannot be inherently bipolar, with the paradigmatic expressions being resentment and ‘guilty to’ or bipolar guilt. In fact, I think, that they must be understood in these bipolar terms in order to make sense of accountability practices such as sincere apology. It does not suffice to simply feel guilty,period and thereby reciprocating the blame of the representative members of the moral community if I want to apologize to you. Instead, I must feel guilty to you; I must acknowledge your blame, that is your resentment, as warranted against the background of a valid claim that you have against me. I must, as it were, exhibit bipolar guilt – guilt that reciprocates blame on the basis of your valid claim against me. For, you will only accept and give uptake to my apology if it expresses my acknowledgment that I have done something that I shouldn’t have done to you in particular and which you, and primarily you, were in a position to demand of me in the first place. Why, Steve, could we not understand Pufendorf’s point in inherently bipolar terms?

    Thinking of the practice of sincere apology then, and here I’m in full agreement with Jay, seems to show that moral accountability does not track wrongness, period, but wrongings. And going back to the conflict case above, granting the intuition that I owe an apology to Anne even though all things considered I should first repay Bob, shows this furthermore, I believe. If I should be offering an apology to Anne despite it also being true that, all things considered, I did the right thing in returning the money to Bob, this suggests that accountability, as a warrant for sincere apology, is not conceptually tied to the non-bipolar concept of ‘moral obligation, period’. Accountability, it turns out again, does not track ‘wrongness, period’ but instead wrongings. To explain the connection between morality and accountability, then, it is indeed sufficient – and one might add, imperative – to appeal to bipolar facts. And I agree with Jay (and his reading of Strawson) that resentment is the paradigmatic expression of blame, with indignation being importantly on behalf. Whether or not a wronging is a wrong, all things considered, is of importance but must be determined independently of its relation to blameworthiness/moral accountability.

    As I’m about to post this, and thinking once more of the conflicting promises case, I’m thinking that some of what I’ve said here is maybe also a roundabout way of suggesting that Steve’s understanding of claims as inputs (which I agree with) actually, and somewhat surprisingly, commits him to a Jay-style positional understanding of blame (which I also agree with). For, the very idea of conflicting obligations that Steve is himself committed to seems to entail the possibility that even though I ought all things considered comply with my promissory obligation to Bob, I can be held morally (!) accountable and feel bipolar guilt and culpability for my failure to comply with my promissory obligation to Anne. For what I have done, even though permissible, has wronged Anne. I might (obviously) be wrong about all of this here, but thought that this is a nice way to end my comment, finding and actually feeling myself in agreement with both of you on a lot of these important issues.

  14. Thanks very much for that intervention, Jonas, which is rich and interesting. You offer a path of reconciliation between Steve and me, suggesting that the kind of relational conception of accountability practices that I favor might be married to the different conception of pro tango directed obligations that Steve defends and that I reject.

    At the risk of seeming ungracious, however, I think I am just going to resist the invitation to meet Steve on this middle ground. One reason for resisting the invitation is that I’m not sure I see what real interpersonal work Steve’s notion of permissibility does if we agree that our accountability practices are structured in bipolar or relational terms. If my wronging someone who has a claim against me provides full warrant for blame, understood in reactive terms, then it seems to me sort of irrelevant that what I did was really permissible in the all things considered or “second-personal” sense. I might remind myself of this aspect of my normative situation, in the spirit of private consolation, but it seems to be socially inert, unable in particular to provide me any protection from the opprobrium of my moral peers. Whereas it is an important point of agreement between Steve and me that the deontic notions that are important to deliberation and agency track and mesh with our conception of the circumstances that warrant moral blame.

    Steve retains the connection between these notions by centering the conception of all things considered obligation that is defined any his understanding of second-personal blame, issued by the representative person on behalf of the moral community; directed obligations and claims are contributory inputs into reasoning about these deontic notions. I retain the connections by taking the relational conception of obligation to be fundamental, and treating this as connected normatively to practices of accountability that are essentially relational in their structure (involving the assertion of claims by and on behalf of claimholders). And those seem to me to be the main options if one wishes to retain the tight connection between deontic moral obligation and moral accountability. (Granted, many moral philosophers would be happy to dispense with this big idea, holding that moral reasons for action are one thing, and reasons for blame are another, and the two are only contingently connected to each other. This is perhaps the dominant tendency in contemporary moral theory, in fact; but it is an idea that Steve and I both reject, and I’m happy to have him as an ally on at least this basic point!)

    Given all this, I think I am just going to have to stick with my position that genuine bipolar moral claims are never permissibly infringed. I know this is a very common way of talking in the rights literature, and it is influentially defended by Thomson in particular. But the alternative I favor has its own venerable history. It is a form of “specificationism” about rights or claims, where the task of moral reasoning is not to determine whether it is permissible in some further sense to infringe a right or claim, but rather to determine the “contours” of the claim (as Scanlon has put the point).

    I’ve already said a lot about how a relational theorist will think about the kinds of cases in which it might appear that a claim is permissibly infringed, and I don’t want to repeat myself. But to turn briefly to Jonas’s case: I think it almost certainly was the right call to pay Bob the $100 rather than Anne, given that I couldn’t under the circumstances keep my promises to both. But it is consistent with this that I still have a directed moral obligation to Anne as well, so that I am flouting a genuine (non pro tanto) obligation I still owe to her, which provides a basis for resentment, apology, and amends. That at least is one possible way of thinking about the case. Perhaps the situation is one in which cash is both vitally important and in short supply, and in leading both Anne and Bob to rely on me to give them $100 tomorrow, I was undertaking a commitment to see to it that I have the resources on hand to deliver when the time comes (and in a way that doesn’t depend on the fickle local atm machines). If that is right, however, then it is indeed a “great deficiency” on my part not to have been in a position to give Anne her money as well as Bob, and I’m fully on the hook for that failure, interpersonally.

    Under another possible developments of the case, however, it might better be understood to have involved the kind of local contingency that could not have reasonably been anticipated when the promises to Anne and Bob were originally undertaken, and that is sufficient to render Anne’s claim to literal fulfillment inert. (In this variant, it might have been unreasonable for Anne to expect me to plan for the extremely unlikely eventuality that all the local atm machines are on the fritz when the time for payment arrives, especially so given that it is common knowledge that she will suffer only minor inconvenience if payment is delayed.) Even in this scenario, however, Anne retains interests that I need to take into account. She certainly has a residual claim against me to compensation for any financial losses she might have suffered as a result of my failure to give her the money at the appointed time (though I wouldn’t myself describe this as making amends).

    Jonas is right that in this variant of the case, there is no real room for a literal apology on my part, given that there was no wrong to Anne that was committed through my failure to fulfill the terms of the promissory undertaking. But there is room for the different kind of apology that was mentioned in my exchange with Micha, which functions to acknowledge the continuing significance of the interests of the party whose original claim is no longer in force. Thus in the second variant of Jonas’s case, I think it would be very natural for me to say something to Anne like: “I’m really sorry that I wasn’t able to give you the money that I promised, and I’ll do everything I can to make it up to you”; this registers the significance of the inconvenience she has suffered, and expresses regret that I wasn’t able to avoid subjecting her to it. But I don’t think it would make sense to say: “I’m really sorry for the wrong I did to you”, nor does it seem reasonable for Anne to resent me once the facts of the case are carefully explained.

  15. Thanks to Steve and Jay (and everyone else!) for these great posts, and more generally for providing inspiration and fodder for my thinking for the past 15 years. Jay’s describing his first encounter with The Second-Person Standpoint took me down memory lane to 2008, when I was an overeager undergrad in Steve’s seminar on moral obligation. We worked through SPS and it lit my brain on fire! Most of my philosophical preoccupations ever since can be traced back to that heady semester, guided through the years by Steve’s ongoing and generous mentorship. And though I’ve had fewer opportunities to talk with Jay, I’ve enjoyed and profited from engaging with his work, especially (though not only) through studying The Moral Nexus.

    Okay, on to philosophy. Apologies if the comments below miss some of what’s happened in the last few hours of comments – I wrote them on the plane, and more discussion has happened since which I haven’t yet been able to read. (Sorry, Jonas!)

    I have a couple different comments, both themed around the relation between normative reasons and directed / moral obligation. More generally, I’m interested in the distinction between what I like to call *input* normative concepts, which feed in to and determine what one ought to do (roughly analogous to premises), and *output* normative concepts, which depend on the overall balance of the inputs (roughly analogous to conclusions). I see two ways in which this theme bears on the Jay/Steve dialogue above:

    (1) Overridden promises and relational inputs (primarily for Jay)

    I’m piggybacking on Micha’s comments here – he already said several things I wanted to say!

    Whether we call it a ‘pro tanto obligation’ or not, it seems clear to me that there is a pro tanto normative reason to keep my promise to meet you for lunch that does not disappear when the promise is overridden. It would be strange if our reason to keep our promises disappeared every time it was outweighed – that would make it very unlike other normative reasons. (The tastiness of a donut remains a reason to eat it even when it is, tragically, outweighed). I think Jay’s right that promises have implicit exception clauses, and maybe “don’t worry about our lunch date if you can save someone from drowning” is one of them. If that’s true, then maybe in the Kamm example your reason to keep your promise really does disappear. But I am skeptical that there is an implicit exception clause for every possible way I could be justified in breaking a promise.

    This is most vivid to me in lower-stakes cases, where we might feel genuinely conflicted about whether to keep a promise. (I think it obscures things in the Kamm case that it’s so obvious that you ought to break your promise – it would be strange, even monstrous, to think “what about my promise to meet X for lunch?” when faced with a drowning person). Suppose you promise to meet a graduate student at 2pm today, but your child is throwing a nuclear-level temper tantrum and your spouse is feeling too sick to care for them on their own … what should you do? The phenomenology here is one of weighing up reasons: your reason to keep your promise vs. your reason to take care of your child and spouse. Even if all things considered you ought to break the promise, that is so in spite of your countervailing reason to keep it.

    If the promise had an explicit exception clause, however, no such reason would apply. Here I agree with Micha, though I’ve historically used a slightly different example: suppose I ask my wife to promise not to interrupt me while I’m working, unless our guest arrives, in which case I want her to let me know. Once the guest arrives, there is *no reason whatsoever* for my wife not to interrupt me, or at least no reason based in the promise. (This seems clearer to me than in Micha’s rescheduling lunch case, where there is reason to keep the lunch date if possible). This is very different from a case where a promise is overridden (or Feinberg’s cabin case), where the reason to keep the promise remains in the normative balance.

    Suppose we’re agreed on this: that promises (and other similar phenomena – eg the property right to one’s cabin, the bodily right not to have one’s foot stepped on) – can continue to ground normative reasons for action even when one is justified all things considered in breaking/infringing them. If we agree on this, and we agree that there are residual duties of explanation and compensation, what is the remaining debate to have about this case?

    There’s a danger of falling into a merely terminological dispute: do we describe this case by saying that the promissory right remains in force, and is just overridden and so permissibly infringed (as Judy Thomson and Joel Feinberg do), or do we go with Jay in saying that there is no promissory right or claim at all, precisely because it is permissible to break the promise? I worry that without further elaboration these are just different labels on the same normative landscape. I also think there are entrenched traditions in philosophy of using the terms ‘right’, ‘claim,’ ‘wronging,’ ‘directed duty’ etc. in both of these ways, and that ordinary language doesn’t cut finely enough to favor one over the other.

    The non-terminological, substantive question at stake, I think, is whether there is something distinctively directed or relational about the normative input provided by promises (and other overridden ‘rights’ such as the property right in Feinberg’s cabin case), or whether relationality only comes on stage on the output side, after all the reasons have been weighed, in the form of what Jay calls ‘strict’ claims and Steve calls ‘all-things-considered’ obligations. So when Steve and Jay are talking about whether there are pro tanto directed obligations, I interpret them as asking whether there is a distinctively relational or directed kind of normative input, of the kind that can be weighed against other normative reasons in determining what one ought to do all things considered, and thus can in some cases be outweighed. Both of them seem skeptical that there is such a thing – as I understand it, the main difference is that Jay takes this to be a feature of his view, while Steve sees it as a bug.

    However, I’m of the view that there is plausibly something directed or relational going on at the input side. The relationality will of course manifest differently: I agree with Steve and Jay that I do not warrant your resentment when I justifiably break my promise to you. But here are two reasons to think there’s *something* relational going on.

    First, only some of my normative reasons are capable of generating directed obligations I owe to you. My promises to Steve can make me obligated to Steve, but can’t make me obligated to Jay (at least in the normal way). In the Kamm case, my reason to keep my promise clearly bears a different relation to the promisee than my reason to save the drowning person does. Both affect what I owe the promisee, but my promise can serve as a ground of a directed obligation to the promisee, while my reason to save the drowning person can only *defeat* my obligations to the promisee, not generate them. What makes some reasons apt to ground my obligations to you and others not? I think it must be something about the normative relation – you might even say the moral nexus – between us.

    Second, a large part of the way Jay motivates and helps us get a grip on the notion of directed obligations and claims is by pointing at cases of non-overridden promises, bodily rights, etc. Jay returns to the case of promises frequently in The Moral Nexus, and he makes this sort of appeal in this post:

    “One impetus for thinking this is that directed duties represent an independent and plausible model for a normative obligation. We understand debts that are owed to other parties to generate deontic constraints on agency, and their functioning in this way has something to do with the fact that they constitutively implicate two different individuals in a complex of directed duties and claims. So an appealing account of the special “reason-giving force” of moral norms seems to fall out of the relational interpretation of them, fairly directly.”

    When the reasons based in promises etc are not overridden, they ground this rich suite of normative relations, as Jay correctly observes and nicely elucidates. If this is right, though, and promise-based reasons do not disappear when they are overridden, then it would be surprising if promises’ relationality disappeared when they were overridden. When not overridden, promises “constitutively implicate two different individuals in a complex of directed duties and claims” – but once a drowning person shows up, promissory relations lose all of these features and just become one humdrum monadic reason among many? If that is Jay’s view – I’m not sure, it very well might not be! – it strikes me as weirdly discontinuous.

    Why not say instead that when the drowning person shows up, my promise to you continues to ground one normative relation between us (the relation that generates a pro tanto reason and secondary duties) but ceases to ground another normative relation between us (the relation that would make it fitting for you to resent me for breaking my promise)? In other words, why not say that there are two distinct but intimately related relational phenomena here: ‘strict’, all-things-considered claims of the sort that warrant accountability, and pro tanto reasons based in promises, bodily and property rights, etc. that also involve a distinctive normative relation to a person? And these aren’t merely two disparate normative relations, because plausibly the former, output relation is grounded in the latter, input relation.

    I’ve been mostly directing these comments towards Jay, but I think they offer a rebuttal to Steve’s argument against Jay as well. Steve writes “neither pro tanto duty nor all-things-considered duty is a directed notion. This distinction only makes sense with respect to moral obligation period.” But why can’t we make a distinction between pro tanto directed obligations and all-things-considered directed obligations? Pro tanto directed obligations are, I suggest, the normative inputs that ground all-things-considered directed obligations when not overridden or otherwise defeated by other inputs, just as pro tanto moral obligations are the normative inputs that ground all-things-considered moral obligations when not overridden or otherwise defeated by other inputs.

    The pro-tanto vs. all-things-considered distinction isn’t exclusive to moral obligation. It also applies to the all-things-considered ‘ought’, which I (and I think Steve?) take to be distinct from moral obligation: I think I ought to exercise tomorrow, but I don’t think I’m morally obligated (or obligated to anyone) to do this. There are pro tanto normative reasons and the all things considered ‘ought’, just as there are pro tanto moral obligations and all things considered moral obligations, and just as (I suggest) there are pro tanto directed obligations and all things considered directed obligations. The general distinction is, again, between inputs and outputs: there is an output normative status (the all-things-considered concept), and there are the normative inputs that ground and explain which actions have that output status (the pro tanto concept).

    So, contra Steve and Jay for different reasons, I would suggest that we can make perfect sense of the idea of a pro tanto, overridable, permissibly infringeable, but still distinctively relational normative concept. Call it a ‘pro tanto directed obligation’ if you like; my preference is just to call it a normative reason that bears some sort of special relation to two persons that makes it apt to ground obligations owed by one to the other.

    (2) The relation between moral obligation and normative reasons (primarily for Steve)

    I’m with the objectors against Steve on the Euthyphro problem (sorry, Steve!). I think that moral and directed obligation are in the first instance normative for action, and secondarily normative for blame and resentment. One piece of evidence for this: when we are wondering whether someone is to blame, we focus on their reasons for action (though of course we also ask whether they had an excuse).

    Above Steve argues that we need to constitutively connect moral obligation with reasons for blame in order to explain why it’s a normative concept at all: “Something is a normative notion, I think, if, and only if, its being instantiated entails the existence of normative reasons, or its being the case that someone ought, to have some attitude or to act in some way in the instantiated situation.” Moral obligation is thus a normative notion because it entails reasons for blame.

    I agree that normative concepts have to be connected up with some sort of normative reasons, but I’m not convinced that we can’t explain moral (and directed) obligation’s normativity by saying that they entail normative reasons *for action*. (I agree that they entail normative reasons for blame too – but I think that’s explanatorily downstream from their connection to normative reasons for action).

    But I see the entailment relation between moral obligation and reasons for action differently from Steve. Saying “X entails Y” leaves open whether X depends on Y or Y depends on X. Steve thinks that moral obligation entails reasons for action because it generates them: the fact that an action is morally wrong adds a conclusive reason against doing it. While my view is that moral obligation entails reasons for action in the sense of being constrained by them: if I have sufficient reason not to phi, then phi-ing can’t be morally obligatory. Moral obligations entail reasons for action not because they generate them, but because they depend on them.
    I think Steve’s entailment requirement for a concept to be normative has to allow for this ‘output’ kind of entailment. Consider the concept of a normative reason itself. This is clearly a normative concept, and it meets Steve’s requirement: the claim that I have normative reason to phi entails that I have normative reason to phi. But that doesn’t mean that the fact that I have normative reason to phi generates a new, additional, extra normative reason for me to phi! That would be bad: reasons would pile on top of themselves ad infinitum.

    So my question for Steve is: instead of saying that moral obligation consists in there being reasons for blame, why not say that moral obligation consists in a certain kind of fact about one’s reasons for action? Then moral obligation would entail reasons for action, not by generating them but by being constituted by them. In other words, why not analyze moral obligation (and directed obligation, for that matter) in terms of an obligatory action’s relation to certain normative reasons? (As you might be able to tell, I like this idea).

    I think Steve is skeptical that such an analysis could work, for two reasons. One, he wants to hold on to the idea that moral obligation provides reasons for action rather than just reflecting them (as he argues in “But It Would Be Wrong” and elsewhere). My view is that this might be true, but the reason generated by moral obligation must necessarily be redundant, because something’s being morally obligatory requires that one had decisive reason to do it as a prior matter of fact. But whether or not I’m right about this (Steve and I have been exchanging emails about it), it can’t be that moral obligation must generate normative reasons in order to count as a normative concept at all – otherwise the concept of a normative reason wouldn’t count as a normative concept.
    Second, Steve simply doesn’t think the project of analyzing moral obligation in terms of reasons for action can be made to work. He has argued that at least two reductive proposals won’t work. First, moral obligation isn’t merely the concept of being supported by decisive normative reason, because (a) there are things I have decisive reason to do that I’m not morally required to do (like exercise) and (b) more deeply, the concept of a normative reason isn’t conceptually connected to accountability in the right way. Second, moral obligation can’t be merely the concept of having most moral reason, because that would rule out the possibility of supererogatory actions (which are most favored by moral reasons but not obligatory).

    I agree that neither of these simple analyses work, but I think that just shows that the relation between normative reasons and moral obligation is more complicated. (In a forthcoming paper I defend the following account: an action is morally required just in case the moral reasons in its favor are enough on their own to outweigh all of the normative reasons, moral and otherwise, in favor of any alternative).

    Whatever account we adopt, suppose we have in hand a true biconditional of the form “action A is morally obligatory for X iff A bears relation R to X’s normative reasons for action.” Steve’s deeper worry would still apply: moral obligation is constitutively connected to accountability, while normative reasons are not. True enough, but I think there very well might be a distinctive *subset* of normative reasons that are constitutively connected to accountability. In my view, that’s exactly what moral reasons are. That is to say, I think the correct account of what distinguishes moral reasons from other kinds of normative reasons for action will have to be an account that explains how moral reasons are constitutively connected to interpersonal accountability (whether we think of that accountability as primarily Steve-style or Jay-style).

    But if we have in hand an account of the relation all and only morally obligatory actions bear to moral (and other) reasons for action, and we have in hand an account of how moral reasons are distinctively connected to accountability or relationality, then why not say: for an action to be morally obligatory *just is* for it to bear the relevant relation to moral reasons (i.e. the distinctive subset of one’s reasons that are connected to accountability in the right way)? Then moral obligation will entail normative reasons, and will be normative in the first instance for action, and because it is not constituted by the fittingness of blame, it will be able to explain it. What’s not to like?

    I am also optimistic about the prospects for analyzing directed obligation in terms of an action’s relation to an agent’s normative reasons. The analysis would be divided into two parts: (1) characterize the distinctive subset of my reasons that are fit to generate obligations I owe to you; and (2) characterize the relation those reasons must bear to an action in order for me to be obligated to you to do that action. And my hunch is that the distinctive relationality would actually be located in (1): to figure out what normative inputs can generate obligations I owe to you, we have to look at the normative relation between us. This connects with my first comment – this is why I think there must be distinctively relational or directed inputs. (2), on the other hand, need not have anything to do with the relation between us in particular – it can just be a general normative relation (in fact, I think it’s the very same relation as between moral obligation and moral reasons).

    If this approach is at all promising, the next step would be to characterize the normative inputs that are apt to generate directed and moral obligations. What distinguishes the normative reasons that can generate obligations I owe to you from all the other reasons for action I have? What distinguishes the normative reasons that can generate moral obligations (i.e. moral reasons) from all the other reasons for action I have? What are these distinctive bodies of reasons, and how are they relational or second-personal, such that they can ground the fittingness of directed and/or impersonal accountability? This is where I think the theoretical action is. To understand the relationality of the output phenomena of moral and directed obligation, I think we need a theory of the relationality of the normative inputs on which these outputs depend.

    Apologies for the almost-a-conference-paper length of these comments – I got carried away! Take it as a testament to how exciting and stimulating I find Jay’s and Steve’s work.

  16. Many thanks, Brendan, for those extensive and probing reflections, which I am still trying to absorb. There is a lot to think about here. On some points, I may not have much to add to comments I have already made on earlier contributions to this discussion. But let me focus on some new issues you have brought up concerning the extent of promissory commitments.

    I find pretty interesting the concession that in true emergency cases, or perhaps in the Kamm case that Steve deployed, the claim to promissory fulfillment might in fact be extinguished altogether. These are the paradigm examples of situations in which many wish to say that there is a relational right or claim in place that it is permissible for the agent to infringe. So if this general model is apt for these cases, why not agree with me (and many others!) in thinking that permissibility tracks the contours of relational rights and claims, rather than representing an independent normative consideration that gets layered on top of them?

    Brendan frames his answer to this question in terms of a distinction, which I like, between normative notions that are inputs to determining what one ought to do, and normative notions that represent the outputs of this kind of reasoning. My approach goes together with thinking that the notion of a directed obligation is an output notion within morality, indeed it is the generic notion of what it is morally right to do. Steve, by contrast, thinks of directed normative considerations as inputs to determining what it is morally obligatory to do, in some different (and to me still elusive) sense.

    I also agree with Brendan when he says that there is a further and more generic output notion, which is the notion of what one ought (all things considered) to do. This might (not sure!) be the same as Parfit’s notion of the indefinable normative right. The fact that an action is morally wrong, in either my or Steve’s senses, is an input to reasoning about what one ought in this indefinable sense to do. On my account, the specific input notion is that of a directed obligation, and this consideration generally functions, qua obligation, as a presumptive constraint, determining what is to be done simpliciter. But the view allows for the possibility that the morally right might be in conflict with obligations of other kinds—including perhaps directed obligations that one owes to people one loves, or maybe just very weighty reasons that are grounded in considerations of impersonal value. These kinds of scenarios might well be understood to involve practical dilemmas, where it remains the case that one is under a moral obligation to do X, but there is a conflicting obligation, grounded in considerations of a different kind, to do not-X. One ought all things considered to do X, insofar as it would be morally wrong not to; but it is also the case that one ought all things considered to do not-X, in virtue of considerations of a different kind. (Note that, on the picture I am trying to sketch here, the directed moral obligation is a contributory or input notion in relation to the all things considered ought. But it is not contributory qua pro tanto obligation that might be permissibly infringed. It is just an obligation, which functions to determine what is to be done, and which happens to conflict with obligations of other kinds.)

    The “fetishism” objection came up at one point earlier in this (now very long) conversation, and I think it is well-understood as applying to the general output notion of what one ought all things considered to do. The objection is typically brought up in connection with the notion of the right, and I think it makes sense if we interpret that notion in Parfit’s indefinable sense, or as the all things considered ought. To suppose that one’s reason for doing X is the fact that it is right in this sense is precisely to mistake an output notion for an input notion; to say that X-ing is right in this sense is to say that there are considerations of other kinds that make it the case that X is what one ought to do, and it is those input considerations that are one’s reasons. But for all that, the notion of the moral right could be an input notion of the relevant kind—so that the fact that it would be wrong to break your promise is the reason why you ought (in the indefinable, output sense) not to do it.

    To return to the cases of permissible promise-breaking: I agree with Brendan that the different camps to which Steve and I belong on these cases each have their own, internally-coherent vocabulary for describing them. So it can look as if it is a merely verbal dispute, without broader philosophical import. But I think the issues I’ve just been discussing help to bring into focus at least one thing that is at stake in this disagreement. The two positions go together with two very different models of moral obligation. If you think, with Steve and Thomson and Feinberg, that relational claims that are fully in place might be permissibly infringed, then you need a notion of moral obligation that is distinct from that of a directed obligation. I think some of the conventional rights literature is a little short-winded on this score, appealing to casuistical intuitions about obligation without really explaining the notion of obligation they are operating with. (This comment obviously does not apply to Steve, who as we have all by now seen has his own elaborate account of moral obligation in terms of reasons for representative moral blame.) If, by contrast, your basic notion of moral obligation is that of a relational moral requirement, that will give you a principled reason for thinking that the notions of permissibility and obligation have to track the contours of relational claims, so that there cannot be genuine directed obligations and claims in place that it is permissible to infringe.

    Brendan mobilizes a different set of considerations that might be at stake in the larger debate here, concerning the way relational normative considerations seem to figure as inputs to moral reasoning about permissibility. In the helpful case he presents involving the promise to the student and the family meltdown, I agree that the promise figures in your reasoning, generating reasons to meet the student that seem to be both present and to pull against the conflicting reasons generated by your child’s tantrum and your partner’s illness.

    Of course, nothing in the relational approach commits one to denying this. It holds that the directed moral obligation is not a consideration that functions as a normative input into reasoning about what it is morally right to do. But even if that is the case, it is still true that one made a promise to the student, and that might continue to have normative significance as you reason about what to do. But Brendan argues that this is not plausibly thought of as a merely monadic normative reason for the agent, but that it must itself have its own implicitly relational structure.

    For one thing, the agent’s promise is apt to ground directed obligations only to some people, but not to others, so it seems that it must already link the agent with the other party before the reacted obligation that it potentially grounds is in place. For another, it would be weirdly discontinuous to think that an input reason could ever give rise to a deontic nexus with the elaborate structure of a set of directed obligations and claims unless it already had some of that same relational structure embedded in it. For these kinds of reasons, he invites us to “say instead that when the drowning person shows up, my promise to you continues to ground one normative relation between us (the relation that generates a pro tanto reason and secondary duties) but ceases to ground another normative relation between us (the relation that would make it fitting for you to resent me for breaking my promise)?”

    Of course, I agree with Brendan that the promise and its uptake does “ground” the secondary directed duties and claims (to timely warning, compensation, etc.). But it doesn’t follow from this that the grounding consideration itself has relational structure that parallels that of a directed duty. This goes back to an issue that came up in my exchange with Simon, who had similarly suggested that we might sometimes stand in relations of directed but not deontic normatively to other individuals. I expressed some skepticism about this idea, and the skepticism carries over to the suggestion that Brendan seems to be making here.

    Maybe it will help to express my own take on these considerations by appeal to moral contractualism, which Steve and I are both fans of, but which I see as a way of thinking about the sort of reasoning that takes normative considerations as inputs and generates, as outputs, directed moral obligations and claims. Reasoning of this kind is always, in a certain sense, relational or comparative. It considers the reasons that agents have for objecting to principles of prohibition, and compares them to the reasons that other parties have on their own behalf for objecting to principles of permission. These reasons are just ordinary monadic reasons, grounded in personal interests of the parties (as I call them), without any inherent relational structure. What brings them into relation with each other is the way they are incorporated, as inputs, into the deliberation that goes to determining what the agent owes it to other parties to do. They are set against each other, as potential bases for objecting to principles of prohibition and permission, with an eye to identifying principles that nobody could reasonably reject.

    So, in the generic promising case, if we have induced someone to count on us to do X through a promissory undertaking, it will follow that there is generally someone, the promisee, who will have very strong personal reasons for rejecting principles that permit us to defect. These reasons are significantly weightier, considered as objections to principles of permission, than the reasons that third-party beneficiaries have for objecting to principles of permission. So in reasoning about what it is okay to do, I will need to pay special attention to the position of the promisee, and to understand the way their personal interests might be affected by my compliance or defection from the promise. In standard cases, these effects on the interests of the promisee will make it reasonable for them to reject principles that permit my defection, even taking fully into account my personal reasons for rejecting principles of prohibition. But the balance tips in the kinds of cases we have been considering, where there is an unexpected opportunity to rescue someone at great risk to myself, or I am experiencing the combined crises involving the emotional state of my child and the health of my partner. Under these changed circumstances, it becomes reasonable, as determined by this essentially comparative procedure of moral reflection, for someone in my position to reject principles of prohibition. But it is not reasonable, given the personal interests at stake on the promisee’s part, for me to reject principles that require me to give timely warning and to offer compensation.

    This is just a very schematic account of how the considerations that are generated by the promise might have a place, as normative inputs, within reasoning about the contours of promissory moral obligation, even if they do not themselves already have an intrinsically relational structure. The alternative account that Brendan sketches at the end of his contribution, by contrast, seems to treat as inputs to determining what it morally right the agent’s “moral” and other reasons for acting. The considerations that are “apt” to give rise to directed obligations fall into the subset of the agent’s reasons that already have built into them some kind of latent relational structure.

    This is a very intriguing picture, which I admit I don’t understand very well, and need to think about some more. But I am struck that its success at explaining some of the distinctions Brendan wants to make sense of depends on our ability to draw independent distinctions within the broader class of the agent’s reasons, between those that are moral and those that are non-moral, and between those that have some kind of relational structure built into them and those that do not. I don’t find that I have a very firm grip on these kinds of distinctions, when I just think about the different things that might give an agent reason to do things. And I worry that projecting them onto the agential normative inputs will in the best case merely relocate the problem, and in the worst case end up explaining the somewhat obscure in terms that are only more obscure. The contractualist framework, by contrast, takes as inputs the ordinary personal reasons of different parties (not just of the agent!), and it brings them into a procedure of reflection that, in virtue of its essentially comparative nature, is always implicitly relational—even in cases where the output is not an obligation on the part of the promisor to fulfill the original terms of the promise.

    P.S. I like both Micha’s and Brendan’s examples of promises with explicit escape clauses, and think that in both of them, there pretty clearly is no wrong if the explicit condition is satisfied and the promisor either doesn’t show to lunch or interrupts Brendan’s work. Brendan rightly observes that there is nevertheless an intuitive difference between them, which is that in his case, there is no reason generated by the promise for his partner not to interrupt his work, whereas there is a reason to keep the lunch date even when the old friend arrives in town.

    I agree that there is this difference, but don’t think it is explained by differences in the ways in which the two promises generate reasons. What is pertinent, rather, are differences in the interests that lead the promisee to elicit or to welcome promissory assurance. In Brendan’s case, he positively wants to be interrupted when the guests arrive, and affirmatively desires assurance that his partner will act accordingly under that condition. In Micha’s case, by contrast, it is natural to assume that what the promisee desires is simply assurance that the promisor will meet them for lunch; they do not, in addition, affirmatively wish to be assured that the promisor will not lunch with them if the old friend shows up. The explicit escape clause functions here, as I noted earlier, to alert the promisee of a non-obvious limitation on the assurance the promisor is able to provide, and to invite the promisee to take that limitation into account in their own planning for the future. If the old friend shows up in town on the day of the proposed lunch, the promisee will still have an interest in having lunch with the promisor, and this may give the promisor a reason of some kind, one that manifests, perhaps, in regret that they are now unable to meet the promisee for lunch. But that isn’t a reason generated by the promise itself, but can be traced to the nature of the interests on the promisee’s part that led the promisor to offer assurance in the first place.

  17. This is for Micha:

    Micha, thanks so much for your thoughtful points and question. I agree that resentment is justified only if blame is justified. So if resentment is justified, then it will follow that, because blame is justified, there was conclusive normative reason not to have done the blameworthy act. My thought was (and is) that the assurance of the conclusive normative reason goes through culpability and that blame is a third-party (representative) attitude, whereas resentment is felt from the victim’s point of view. I can share that point of view vicariously, but I still recognize that it is not my point of view, it is the victim’s. Blame, however, comes from a perspective that every person can share as a representative person. So they make the demand of themselves and that assures the normative reason.

    Anyway, that’s the way I’m thinking of it. Thanks again for the question and feel free to try again if I seem not to be getting it.

  18. For Jonas,

    Thanks so much for this, as always, Jonas. I agree of course that apologies only make sense in cases of directed obligation and are to the obligee whose obligation has been violated. I also agree that an apology is called for in cases where where what I would think of as a residual claim has been violated in a way that is justified, all things considered. The person has still be injured even if justifiably. Whether we say they have been wronged or not seems to me largely a semantic issue concerning whether we want to maintain the entailment from wronged, to wrong (period) or not. If we do, then we should say they haven’t been wronged, though their claim gave rises to an injury. If we do not, then we can happily say they have been wronged.

    On the question of whether accountability tracks wrongings or wrongs (period), I say there is personal accountability, which tracks wrongings, and accountability to the moral community (and everyone as representatives, but not individually) and that tracks wrong period.

    Hope this is responsive, and thanks again.

  19. For Brendan,

    Thanks so much for the kind remarks!

    Just quickly, first, on pro tanto directed obligations — I have a meeting in a couple of minutes–I agree that the pro tanto obligation is generated by a directed obligation, I’m just denying that its pro tantoness, as it were, is directed, since we are weighing the contribution it makes to what is one’s moral obligation period, all things considered.

    I’ll get to your other questions in a bit.

    Thanks and let me know if I missed your point.

  20. I’d like to open with a note of (second-personal/ relational) gratitude to both of you, Jay and Steve, for this wonderful and illuminating exchange and for your hard work to make ethics take more seriously the possibility of relational or second-personal normativity. More personally, your work has been transformative for me. I can’t imagine working on these topics at the level of conceptual sophistication that we currently have without your work. Thank you.

    Lots has been said by you and others in this wonderful exchange, and I apologize I’m arriving a little late. At this point, I can’t and shouldn’t attempt to track everything – so I’ll also express my apologies to the other participants in this wonderful exchange for my inability to track what they’ve said. Instead, I’ll focus on what I take to be two foundational questions, one is the fundamentality of relational properties, the other is rationalism in ethics.

    One striking feature of your exchange concerns the nature and so fundamentality of relations, i.e., what exactly makes it the case that some entity counts as a relation in your sense and what is the metaphysical status of such a property? What I find striking about your work is that not only do you have different conceptions of relationality, but also that each thinks that the other’s conception is not adequate because somehow only ‘notional’ relationality. Thus, Jay’s ‘relational’ account focuses on relationality as directedness of duties. And Steve worries that making relationality about directedness makes relationality ‘something approaching a logical notion’. This is not a real relation because directedness need not involve interaction, mutual second-personal accountability. For his part, Steve’s account of relationality in terms of accountability faces a mirroring worry from Jay. If relationality amounts to nothing more than bare accountability, as Steve himself admits, this is a relationship that could be instantiated between an agent and a transcendental idea, i.e., the idea of a community, rather than actual other individuals. The same concern Steve raised against Jay can now be raised against Steve’s account: why isn’t this something that approaches merely a logical notion of relationality?

    And so, I’ve wondered for a while, and still wonder: what exactly is relational about your relational versions of morality?

    One way I’ve found it helpful to flesh out this question is in terms of an old metaphysical distinction between monadic and relational properties. In particular, I have in mind an old disagreement between Leibniz and Kant on the structure of the world and, in particular, on the relationality of causality. Leibniz’s doctrine of pre-established harmony says, contra occasionalism, that substances are real causal agents, but their causal activity (and their resulting properties) are purely monadic. That is, the properties of a substances make no essential reference to any relations to other substances. If a stone is heated by the sun, the stone’s heat is not a relational product of the sun’s heating; it’s a property of the stone itself that mirrors, through pre-established harmony, the similarly self-affecting heating activity of the sun. Strictly speaking, the sun doesn’t heat the stone; the stone is not heated by the sun. Already in his earliest pieces of writing, Kant vehemently opposes this rationalist monadicism. Like Leibniz, Kant too rejects the occasionalist notion that that finite substances are not true agents. But unlike Leibniz, Kant rejects the monadic idea that a substance’s properties are purely monadic. When Kant writes the first critique, nearly 20 years later, he’ll go on to characterize the very structure of all temporal and spatial relational in irreducibly relational terms. Kant’s worry is that if only monadic properties are fundamental, the world shatters into as many discrete unrelated parts as there are substances, making not just the unity of the world (and their laws) unintelligible, but also the very possibility of action and interaction unintelligible. Causal relations, spatial relations, temporal relations, Kant argues, are irreducible to monadic properties (an argument that is beautifully rehearsed by Eric Watkins on his book on causality and Ray Langton’s 1998 book on Kantian humility).

    I apologize for this detour into early modern metaphysics, but the debate between Leibniz and Kant on the fundamentality of monadic properties seems, at least to me, to be potentially illuminating here. As I see things, the form of Kant’s objection to Leibniz is that Leibnizian relations are, to put it in Steve’s wonderful terms, ‘something approaching a logical notion’. It’s not that Leibniz denies relational properties. Rather, Leibniz makes relations purely notional, because, Kant worries, the only fundamental properties are monadic.

    Now my worry is that Kant’s worry applies to both of you. Although Jay’s account explains relationality qua directedness, Steve’s worry is that directedness is not relational enough. But what does that mean? One possibility is that Jay’s relationality is merely notional because not fundamental. And that seems correct.

    Jay’s account superimposes relationality as directedness on what seem to me ultimately monadic properties, namely, the interests of individuals. Just as Leibnizian substances have properties perfectly intelligible independently of relations to other substances, Wallacian (!) agents have interests that are perfectly intelligible independently of relations to other agents. The fundamentality of monadic properties (i.e., the value of individual interests) opens the account to the Kantian charge that relationality as directedness is not just supervenient, but epiphenomenal, unreal, notional, merely logical.

    Steve’s account superimposes relationality as second-personal accountability on what seem to me ultimately monadic properties, namely, the representative authority I have via membership of a notional, transcendental community. Just as Leibnizian substances have properties perfectly intelligible independently of relations to other substances, Darwallian (!) agents have a representative authority that is perfectly intelligible independently of any actual interaction to other real agents. The fundamentality of monadic properties opens the account to the Kantian charge that relationality as accountability is not just supervenient, but epiphenomenal, unreal, notional, merely logical.

    And so I’m left wondering: if both accounts make what Leibniz and Kant would have called monadic properties fundamental, then what exactly is relational about your relational moralities? And why, after all, do we need relations?

    One way to sharpen this point is to transition from the question of fundamentality of relations to the question of rationalism. And here, I’ll echo things Simon said about the topic.

    If in Jay’s account, what’s fundamental is monadic properties pertaining to the interests of individuals, then one wonders what explanatory work, if any, is done by relationality as directedness. We can sharpen the point by Jay’s affinity to contractualism. Scanlon’s contractualism, as you both acknowledge, has no appeal to (maybe because there’s no need to appeal) to relational notions. A healthy dose of realism about reasons combined with the interests of individuals means that what we owe to each other is really about what we owe, period – as Steve would say. But then, it’s not clear that relationality as directedness (an epiphenomnal relational structure superimposed on the value of interests) is much needed to get an account of rationalism. The Leibnizian would be pleased here: the account is driven by the monadic value of interests.

    Curiously, the same structure can be extended to Steve’s account. If in Steve’s account what’s fundamental is monadic properties about representative authority, then one wonders what explanatory work, if any, is done by relationality as accountability (nevermind relationality as directedness, which is in Steve’s account much less fundamental than in Jay’s). Facts about moral obligation period are fixed by facts about such transcendental presuppositions, the notional notion of a community of individuals. But as Jay rightly noted above, I think, this leaves behind the Strawsonian idea of a participant stance and leaves us with a detached, impartial perspective. That might be fine as an account of obligation period. But it’s not clear that actual mutual recognition or accountability is much needed to get an account of rationalism. The Leibnizian would be pleased again: the account is driven by the monadic property of having representative authority.

    To close this already too long contribution, the issue of relationality can also be put in yet another way: the relationship you both have to contractualism. As you both acknowledge, you have a strong affinity to Scanlon’s contractualism. And you both acknowledge that Scanlon’s contractualism, at least as Scanlon developed, did not appear to need either of your notions of relationality to try to support what Steve called ‘rationalism’ here. And so, you both try to offer a ‘relational foundation’ for contractualism. But one benefit of thinking about the fundamentality of relations in the way I proposed is that if, ultimately, relational properties are not fundamental, it seems like the story for you both runs the other way: it’s the non-relational contractualism that provides a foundation for your versions of contractualism. So, we are left wondering again: what exactly is relational about relational morality?

  21. Thanks so much, Jay, for your detailed and clarifying response. It’s helped me to nail down one important way in which we’re thinking differently about things.

    As I understand it, Jay accepts a two-stage model of practical reasoning. The first stage is contractualist-style reasoning that takes as inputs the objections that individuals can raise to principles permitting vs forbidding an action, and yields as output directed obligations – facts about what, if anything, I owe to whom.

    The second stage comes when one asks the distinct question of what one ought to do simpliciter or all things considered. Here Jay conceives of directed obligations as serving as inputs to this question – just conclusive inputs that function “as a presumptive constraint, determining what is to be done simpliciter.” But there are other possible inputs – e.g. considerations based in impersonal value – which can also affect what one ought to do simpliciter.

    (One thing I didn’t realize before was that Jay thinks that when directed obligations conflict with these other kinds of considerations, they aren’t overridden – instead, one faces a rational dilemma: “One ought all things considered to do X, insofar as it would be morally wrong not to; but it is also the case that one ought all things considered to do not-X, in virtue of considerations of a different kind.” I must admit that I have trouble making sense of the idea of a rational dilemma. Suppose I ask the Ethics Oracle, “Should I do X?” They reply, “No!” I say, “Should I do Y?” They reply, “No!” “Should I do something other than X or Y?” “No!” “Okay, should I do nothing?” “No!” This doesn’t make sense to me. I’m fine with it being the case that, no matter what one does, one will be accountable or criticizable in some way. But it seems to me like there must be some correct answer still to the deliberative question of what to do. But all this is in parentheses because I think it’s ultimately tangential).

    All this means that when we ask whether something serves as a normative input or output, we have to specify which stage it serves as an input to or output from. There are the inputs to contractualist reasoning, which Jay takes to be interests, which determine what my directed obligations are; and then there are the inputs to all-things-considered practical deliberation, which include my directed obligations.

    I think part of what may have led Jay to feel puzzled by my comments is that I was implicitly assuming a one-stage model of practical reasoning. On my view, there is only one set of normative inputs: your normative reasons for action (all of them, whatever they may be: personal and impersonal, relational and nonrelational, moral and nonmoral). But these normative inputs ground / generate / determine multiple different output normative statuses: they determine what you ought to do ATC *and* what you are obligated to other individuals to do (and, if you follow Steve in thinking this is a distinct question, what you are morally obligated to do). The difference between what you ought to do simpliciter and what you’re obligated to others to do isn’t that these are distinct stages of practical reasoning with different inputs, but that these are different relations your action can bear to your normative reasons for action. This is why I think the action lies in figuring out (a) what distinguishes the normative reasons that ground my obligations to you from other normative reasons and (b) what relation an action must bear to these reasons in order to make it something I owe to you.

    One reason I prefer this one-stage model is because I think that any normative reason that bears on what I ought to do simpliciter necessarily also bears on what I owe to you. You can motivate this idea within the contractualist procedure: as Jay and Scanlon say, we figure out whether I would wrong you by phi-ing by comparing the objections you can raise to a principle that permits me to phi with the objections that others, including me, can raise to a principle that forbids my phi-ing. I owe it to you not to phi just in case your objections to my phi-ing are stronger than anyone else’s countervailing objections.

    Now suppose there is some normative reason that counts strongly in favor of my phi-ing all things considered. Maybe it’s a reason of impersonal value – that phi-ing will protect natural beauty or promote aggregate well-being or something like that. Won’t that reason bear on the question of whether it’s reasonable for you to reject a principle that permits me to phi?

    Maybe Jay will reply that it doesn’t, because it doesn’t ground an objection that any individual can raise on their own behalf. One response might be to say that *anyone* could object to a principle forbidding my phi-ing on the basis that there is strong impersonal reason for me to phi. A more stubborn response, which I’m inclined to, is to say that if the contractualist picture can’t allow that any strong normative reason for me to phi is an objection to a principle forbidding me from phi-ing, then we should reject the contractualist picture. To ask whether I would wrong you by phi-ing, I think we have to ask whether I have sufficient reason, all things considered, to phi – and thus take all of my normative reasons for action into account.

    The better route for the contractualist to take here might be simply to deny that impersonal value considerations are normative reasons for action at all, and to insist that any normative reason for action must be tied to some individual’s interests. If they take that route, though, they are agreeing with me that the reasons that bear on what I owe to you are identical to the reasons that bear on what I ought to do all things considered. There is just a distinction to be made within those reasons: some of my normative reasons can generate obligations I owe to you, while others can only defeat such obligations. (The contractualist can say: the former are reasons based in your interests/objections, the latter are reasons based in others’ interests/objections).

    How does all of this bear on the overridden promise case? Well, if we accept Jay’s two-stage model, then the way a promise bears on what I ought to do when it isn’t overridden is really quite different from the way it bears on what I ought to do when it is overridden. A non-overridden promise results in a full-throated directed obligation, which then serves as a presumptive constraint on all-things-considered obligation and (necessarily?) settles the question of what I ought to do, determining that I ought to keep the promise (even if I also ought to do something else for impersonal reasons). While an overridden promise doesn’t have this output. Instead, when a promise is overridden, we have to look at the first stage, asking “what principles could the promisee not reasonably reject governing how I behave when I could save a life by breaking my promise?” In answering this question, we can’t appeal to even a pro tanto directed obligation based in the promise, because whether there is such a thing is what is in question. Instead, we look to the promisee’s interests, compare them with others’s interests (such as the drowning person), and arrive at a set of principles governing how to act in this situation, which says roughly “save the drowning person, but explain and maybe compensate the promisee for having broken the promise.” And these principles take a specificationist form: there seems to be no lingering requirement, even pro tanto, to keep the promise, just a requirement to explain and compensate. This helps me to better understand what struck me as a weird discontinuity: on Jay’s picture, it’s a principled difference.

    While on the one-stage model I prefer, the normative reason to keep your promise is identical in overridden and non-overridden cases (assuming an exception clause hasn’t been triggered); all that has changed is that other reasons have entered the balance. When the drowning person shows up, it’s not that my reason to keep my promise is replaced with a reason to explain and compensate for breaking it. The reason to keep my promise remains and has the very same weight. My reasons to explain and compensate don’t replace the reason to keep the promise, but instead share a common cause: the fact that I promised gives me reason to keep it, and also gives me reason to explain and compensate if I break it.

    (Let me clarify here about the concession I made: I want to concede that there can be implicit exception clauses for promises, and that *in principle* those implicit clauses could include exceptions for high-stakes emergencies. But that is a concession for the sake of argument: the point is that whatever we say about high-stakes emergencies, the specificationist view requires that there is an implicit exception clause *every time* I have sufficient reason to break a promise, and I deny this. As a matter of fact, I don’t think there’s an exception clause in emergencies either. But that’s harder to see clearly, because it’s difficult to intuitively distinguish existent-but-massively-outweighed reasons from nonexistent reasons).

    As I elaborate on these models, though, the distinction between them is becoming blurrier in my mind. Couldn’t Jay agree that there is a remaining reason to keep my promise, because the promisee still has an interest in my keeping it?

    I’ll end this somewhat wandering reflection with two questions for Jay:

    (1) Do you think that there are reasons that bear on what I ought to do simpliciter that don’t bear on what I owe to you, or vice versa? Elaborating each direction a bit: if I have strong normative reason to phi ATC, does that always count as an objection to a principle requiring me to phi? And if you have reason to reject a principle permitting me to phi, does that count directly as a reason against my phi-ing all things considered? If they aren’t identical, how are the inputs to directed obligation and the ATC ought related to each other?

    (2) Suppose we keep your two-stage model; why can’t we still say that the inputs to the first, contractualist stage are relational? You say that reasons for rejecting principles “are just ordinary monadic reasons, grounded in personal interests of the parties (as I call them), without any inherent relational structure”. But there is a deep and important difference in contractualism between reasons *you* have to reject a principle and reasons *I* have to reject a principle. I think this isn’t merely the difference between reasons grounded in your interests and reasons grounded in my interests, even if the two coincide. Consequentialists can distinguish reasons based in your interests from reasons based in my interests. It’s important to contractualism not just that there are reasons based in your interests, but that those serve as a basis for *you* to object to that principle / action. So some of the reasons that bear on what I’m permitted to do in contractualist reasoning are distinctively connected to you, and ground objections *you* can make to *my* actions. Why doesn’t that count as an “inherent relational structure”?

    Finally, a quick reply to Steve:

    One kind of pro tantoness is the contribution a fact makes to what is one’s moral obligation, all things considered. Considering the fact that I promised in this light, it is just a pro tanto moral obligation: it’s a contributor to what I’m morally required to do. But another coherent kind of pro tantoness is the contribution a fact makes to what I owe to you, all things considered. And it’s a distinct claim to say that the fact that I promised you to phi is a pro tanto contributor to my being obligated to you to phi. Why doesn’t that count as a directed and pro tanto concept?

  22. Sorry, a quick typo correction: in question (1) to Jay, it should be “Elaborating each direction a bit: if I have strong normative reason to phi ATC, does that always count as an objection to a principle *forbidding* me to phi?” I got my negations mixed up.

  23. Ariel, many thanks indeed for your super interesting reflections about relationality as a foundational notion, which are illuminating. I have some meetings this morning, so I’ll have to keep this pretty short (anyone who might have read to this point in the exchange is probably now breathing a sigh of relief!).

    The distinction between essentially relational and non-relational properties is a helpful one, even if it is imported from early Modern metaphysical debates. Note that in these terms, the property of moral rightness, as I interpret it, is an essentially relational property, which makes ineliminable and not merely “notional” reference to an individual besides the agent. It is the property of being owed by the agent to a second party, qua moral persons. This is a real (not merely a logical or a notional) relational relation, and it also defines a relation between two real (and not merely notional) individuals. I think something similar might be said about Steve’s conception of moral rightness; this is a matter (as we have seen) of its being the case that the representative person has reason to address the corresponding demands to the agent, where address essentially involves an addresser and an addressee. (As noted earlier, I have some doubts about whether this is a normative relation between real individuals, insofar as the representative person is speaking for the abstraction that is the moral community. But the property of rightness does ineliminably involve the relational notion of addressing a demand to another party.)

    The point is that, in terms of Ariel’s own distinction between relational and non-relational properties, the most fundamental moral notion, that of rightness, is an essentially relational property, on the accounts that both Steve and I offer. Our approaches differ in this respect from the other main accounts in the philosophical tradition, such as consequentialism or virtue ethics, according to which rightness is a monadic property, which does not essentially involve two individuals. The right act is the one that produces the best consequences (impartially conceived), or that instantiates an ideal of virtue, and there is nothing in these conceptions that irreducibly relates the agent to another party.

    Ariel’s concern about my view, if I understand him, is that it is not relational enough, because the relational normative property of rightness is “grounded” in something non-relational, namely the interests of individuals. Directed moral obligation is thus not “fundamental”, insofar as it bottoms out in a normative property that is merely monadic rather than relational. In support of this charge, Ariel notes that contractualism makes use of the non-relational notion of individual reasons in its basic account of what it is for an act to be wrong, with the result that rightness becomes relational in a merely “epiphenomenal” sense.

    I find this misleading, for a number of reasons. For one thing, contractualism (on my relational interpretation of it, which may differ from Scanlon’s own current understanding) does not merely appeal to the the interests of individuals, considered monadically in isolation from each other. Rather, the fundamental question for moral deliberation, on the contractualist approach, is whether it would be reasonable for an individual to reject principles of prohibition or permission, and this is (as I noted above) essentially a comparative matter. It is a question of the force of the objection one individual has, given the effects of the principle on their personal interests, compared to the objections that other individuals would have to the alternatives. It is the inherently comparative character of moral reasoning, on the contractualist conception of it to which I am drawn, that makes it plausible as an account of how we determine what one individual relationally owes to another. And this aspect of the account is read out of the picture if one treats it as an attempt to reduce relational rightness to the monadic reasons and interests that are inputs into the essentially comparative procedure of reflection that it defines. There is nothing at all that is merely “epiphenomenal” about the relational conception of the moral right, on this approach.

    An issue that came up earlier in the exchanges with both Simon and Brendan is whether the inputs into reasoning about the moral right might themselves sometimes be relational normative notions, or whether they should instead be understood to involve ordinary monadic reasons and interests. I favor the latter interpretation over the former, because I can’t quite get a grip on what a “directed” reason might look like if it isn’t already understood to be a duty that one party owes to another. But I think this question is independent from the question of whether directed rightness is itself an irreducibly relational notion. As I have been at pains to emphasize, even if the inputs into contractualist reasoning are monadic individual interests, the procedure itself brings them relation to each other through the inherently comparative questions that it poses.

    But if you don’t like the contractualist story, you don’t need to accept it to defend the relational conception of morality in terms of obligations that are owed by individuals to other moral persons. As I note in The Moral Nexus, you could be an intuitionist about directed moral obligation, someone who holds that there is no perspicuous general representation of the reasoning that enables us to discern when it is the case that one individual owes a duty to another. I think there are differences between this view and the contractualist variant to which I am myself drawn, concerning issues in moral epistemology (as it were). But there is no significant difference between them when. it comes to the question of whether moral rightness is a fundamentally relational notion.

  24. Between meetings, here are a few thoughts about Brendan’s latest.

    About the one-stage vs two-stage model issue: yes, this is a helpful way to frame things, It seems to me, though, that if you really favor a one-stage model, then moral rightness disappears as a normatively-significant notion in itself. There is just the question of what the agent ought to do, all things considered, and reasons of various kinds might feed into that, as inputs, Some of these reasons might be “moral” in character, and others will be reasons of other kinds; but an act’s being morally right or wrong won’t be among the moral considerations. Moral rightness is thus transparent to practical deliberation about the basic issue of what is the thing to do. (I guess if you have this view, you might think that it is fetishistic to attach normative significance to moral rightness per se, which is just a heuristic or summary concept with no independent normative import.)

    In these terms, I think most of the main ethical theories in the modern tradition are two-stage accounts of one kind or another. They offer interpretations of the moral right, with a desideratum being that the interpretations should help us to make sense of the kind of normative significance that rightness seems to have in application to the question of what is to be done. Steve’s account, as I understand it, is a two-stage theory in this sense, as is mine. We’re both trying to offer characterizations of what it is for an action to be morally right that will make sense of the significance of this consideration for resolving questions of what is to be done. (This is the question of the reason-giving force of the moral, as it is sometimes put. I talk about it in terms of making sense of the standing of moral rightness as defining obligations, which function as presumptive constraints at the last stage of reasoning about what to do; and Steve talks about it in relation to the hoary issue of the authority of moral rightness.)

    Brendan makes the astute observation that if we think in two-stage terms, there is at least the theoretical possibility that considerations that enter as inputs into reasoning at the first stage, about what it would be right or permissible to do, could potentially enter a second time in reasoning at the second stage. The effects of a principle on someone’s personal interests might be relevant to the question of whether it would be reasonable for him to reject the principle as a basis for social life. But even if it is not reasonable for that individual to reject the principle, acting on it will still affect them, and that is potentially an impersonal reason that might be taken into account at the second stage of deliberation.

    I try to exploit this possibility in Chap. Six of The Moral Nexus, arguing that there is a kind of “extramural concern for moral persons” that could have a bearing on the question of what is to be done even after we have figured out what it would be right or permissible to do. Relational morality might not deliver the verdict that you have to save more rather than fewer persons in the classic, Anscombe/Taurek kinds of rescue cases (as Anscombe put it, “Who would be wronged?” if you saved one rather than five?). But the impersonal value of the lives that stand to be saved might be a kind of extramoral reason to save more rather than fewer if those are your options. In sufficiently dire situations, of the kind that motivate different versions of “threshold deontology”, the aggregate impersonal value of the great many lives you might be in a position to save could conceivably ground a kind of (non-directed) practical requirement, which conflicts with, but does not undermine, the directed moral requirement that you owe to specific individuals. (The result, however, would be a practical dilemma, a notion that Brendan is skeptical about, but that I follow many philosophers in taking to be intelligible and perhaps even unsurprising, given the plurality of considerations that can potential bear on questions about what to do.)

    Brendan asks, perfectly reasonably, why considerations of impersonal value, if they are available as reasons at the second stage of practical deliberation, might not also be appealed to directly at the first stage, when we are thinking about what it would be morally right to do. This is of course a theoretical possibility, and it is one that seems to be exploited in the version of contractualism that is part of Parfit’s Triple Theory. On this account, an act is right just in case it is required by principles for the general regulation of behavior that anyone could rationally will, where individuals can rationally will an outcome if it would be impersonally best (in the aggregate). It is no surprise that the theory that results from this approach is a version of rule consequentialism.(As I understand the view, being required by principles of this rule consequentialist kind is a high-level property that makes actions right, in the indefinable sense of to be done—‚an input, as it were, into reasoning at that last stage.)

    Whatever conception of rightness Parfit is trying to elucidate with this theory, it is not the directed notion of what we owe to each other as moral persons. If contractualism is to be plausible as an interpretation of the first-stage reasoning that determines what we owe to each other morally, then we will have to place artificial constraints on the normative reasons that figure as inputs into this kind deliberation. This is the rationale I favor for the notorious individualist restriction, which allows as inputs into contractualist deliberation about what principles it would be reasonable to reject only considerations having to do with effects of those principles on the individual’s own life—on what I call their personal interests. It is considerations of this kind that have the potential to give rise to relational claims—though as I have emphasized above, determining whether this is the case requires us to compare one individual’s personal objections to principles of permission with the personal objections that other individuals have to principles of prohibition.

    Brendan asks, about the overridden promise case that has now come up again and again, whether there is a residual reason to keep the promise, “because the promisee still has an interest in my keeping it”. I agree that there is an interest of the promisee’s that remains in place here, and it is potentially significant at both of the two stages I envisage. At the first stage, it is relevant, even after we have determined that it is permissible not to fulfill the terms of the promise, when it comes to the different question of whether it is permissible not to give timely warning and to provide compensation. These are things that it is generally wrong to do, because promisees have residual personal interests that make it reasonable for them to reject principles that permit the promisor to defect without even letting them know or taking steps to cover the losses they may have occurred.

    At the second stage, the same interests will still be in place, and they might well function as free-floating reasons of some kind that have ongoing significance for the agent. I don’t think they will make any difference when it comes to settling the second-stage question of what is to be done, since we have already determined that it is permissible not to fulfill the promise’s terms, and there are compelling considerations in favor of the promisor’s acting otherwise. But at the very least, these residual promisee interests will give the promisor reason to regret that they were not able to accommodate them in their planning, something that is given expression when they offer the kind of apology I have invoked a few times by now, saying that they are sorry they weren’t able to do what they originally promised to do.

    I think answers to Brendan’s two specific questions might well be implicit in what I have already written to this point (in this post and others). But to sum up:

    1. The basic two-stage picture is that there are inputs to reasoning about what it is morally okay to do, which on my approach is reasoning about principles that determine what we owe it to each other to do. This in turn functions, at the second stage, as an input into the question of what to do, all things considered (in the guise of a practical requirement or constraint). On this picture, the consideration of what to do, all things considered, is not yet available to function as an input into reasoning about what would be morally right or permissible, since it partly depends on the output of that reasoning. There are ways of incorporating impersonal reasons as direct inputs into reasoning at the first stage, about what it is morally right or permitted to do, and Parfit might be an example of someone who goes down this path. But his rule consequentialism isn’t plausible as an elucidation of the notion of directed moral obligation; to make sense of that notion in relational terms, I think we need to impose more constraints on the inputs that factor into first-stage deliberation about the principles that determine what we owe to each other, morally.

    2. For reasons I think I’ve already canvassed, I can’t quite make sense of the idea that there are directed reasons, which enter in their inherently relational guise as inputs into first stage deliberation about what it is morally right or permissible to do. The normative considerations that are allowed into this reasoning, on my relational interpretation of contractualism, are just ordinary monadic interests, of the kind that connect to our personal interests. But I agree with Brendan that the relevant reasoning nevertheless has an “inherently relational structure”, for reasons that I tried to articulate in my reply to Ariel’s post. The basic point is that the question of reasonable rejectability we are trying to resolve through contractualist reasoning is essentially a comparative question, about whether the effects of a principle of permission on A’s personal interests make it reasonable for A to reject the principle, given the personal-reason-based objections that B has to principles of prohibition.

  25. Jay –

    Real quick, to make this not even more exhausting for you than it must already have been:

    Thanks so much for your characteristically thoughtful and thorough reply. Given your position it makes perfect sense for you to speak of “compensation” rather than “amends” in relation to the sort of case I was referring to, so I should have been more careful in my own rewording of your position. I also appreciate how charitably you took up my challenge — what does your account have to offer as to the difference between a case of the above sort and a conditional promise? — by noting that that challenge is “especially telling” given that you yourself would describe the former as one calling for “compensation” rather than “amends.” Finally, I very much agree with how you are inclined to analyze our inclination to address the promisee in the language of apology even in the case of a conditional promise: as a matter of acknowledging the costs imposed on the other, even though these costs don’t arise out of a wrong against the promisee on our own part here.

    You write: “I do indeed think that promissory commitments are always implicitly conditional, in a way” and then go on to suggest that the difference between the two cases has to do with the latter’s not merely implicitly but explicitly conditional character. The way you describe the latter scenario strikes me as extremely sensible, but I still wonder whether there is not perhaps a deeper difference here than you are admitted. Perhaps one way to put the point is to note that in the former case I am entitled to TRUST you to meet me for lunch, whereas in the case of a conditional promise trust in the promise’s consequent viewed in isolation seems out of place. I can of course trust you as far as the conditional promise as a whole is concerned, and I can moreover engage in PREDICTING whether its antecedent will or will not obtain, but that last bit is precisely different from trust. So all that seems to suggest that the difference here runs deeper than you might be willing to concede. I’m not sure.

    Either way: thanks so much for the exchange! Much appreciated.

  26. Steve –

    Thanks so much for clarifying what you had in mind. What I myself was after was the thought that my sense of guilt in response to both resentment and blame alike are premised on my taking either of them to be warranted. In THAT sense — i.e., in the sense of recognizing its warrantedness — I thus have to be able to share the point of view internal to the reactive attitude in question in EITHER case.

    I take all this to be consistent with the idea that blame but not resentment is as it were affectively accessible to me, given that blame is available to all members of the moral community and so also to me qua fellow member of the moral community, whereas resentment is available only to my sufferer, who is separate from me.

    However, the sense in which blame is whereas resentment isn’t available to me seems to me to be orthogonal to the question of whether I “agree with” the attitude, which latter question strikes me as a question of either recognizing or else refusing to recognize the attitude as warranted, in respect of which I assume resentment and blame to be on a par with each other, meaning that agreement AND disagreement alike are available to me in respect of both resentment and blame.

    You on the other hand seemed to want to frame the difference between blame and resentment as having to do with my ability to agree with the former but not the latter, which again seem to me orthogonal to where their actual difference lies, namely in what I called “affective access,” or at least so it seemed to me when you wrote:

    “I can fully accept that you have reason, from your perspective, to resent something I did, and so I must listen to your complaint, without agreeing with your complaint and internalizing the attitude as guilt.  It is BLAME that is internalized as guilt, where both involve the implicit addressing of a demand from a position of representative authority.”

    So that’s why I was slightly puzzled by that remark: it seemed to me to run together the ability to recognize an attitude as warranted on the one hand and the ability to access the attitude affectively on the other. It is with respect to the latter that resentment and blame are asymmetrical, not the former. Or so it seems to me.

    I hope this makes better sense!

  27. Thank you very much to Steve and R. Jay (and everyone else!) for their insightful debate. I would like to ask R. Jay something about the “emergency claim” he raised on the moral infringement permissibility in extreme cases. Is there in your words some claim about the contextual value of moral norms? In case you concede this, what is the relationship between intersubjective and “environmental” context? Are they equally relevant for suspending moral demands? You stated this thesis in your first intervention:

    “The unexpected emergency doesn’t make it all right to infringe a claim that remains fully in place; rather it shows the limits of the claim, and it is permissible to fail to honor to the terms of the promise precisely because you do not literally owe it to the promisee to live up to them in these exceptional circumstances (though you might owe them back-up obligations of explanation and compensation)”.

    If the status of moral norms depend on their relational character, why nature (for instance) could say something relevant about what we owe to each other? I could concede this idea intuitively, but I would like to get an insight from the point of view of the moral theory you defended here.

  28. Thanks very much, Micha, for the kind words and the interesting further comment. I am indeed starting to run out of steam here, and this will possibly be my last post.

    I like what you say about trust, and the difference in this respect between the explicitly conditional promise and the ordinary promise that has an implicit escape clause. I agree that in the merely implicit case, I am entitled to trust you to meet me for lunch, whereas we wouldn’t want to say that about the conditional promise.

    One additional consideration that might be relevant to explaining this difference hasn’t yet been brought up. In the conditional promise case, I said the promisor is putting the contingency that would release them from the promise on the agenda of promisee, as something they should take into account in their planning for the future. They are hedging their commitments explicitly, on account of anticipating a contingency that might affect their willingness to show up to lunch. Note that the contingency they are anticipating in this way is not necessarily of the kind that would undermine the promissory commitment if it were not explicitly mentioned in advance by the promisee. If you just promise to meet me for lunch, maybe the surprising arrival in town of my old friend would not be sufficient to get me off the hook. Still, I might not be willing to undertake a commitment that would preclude my meeting up with my friend in the possible scenario that they show up, and I accommodate this consideration by explicitly hedging (as I put it) the promissory obligation.

    The implicit conditions, by contrast, function somewhat differently. They are generally considerations that are capable of undermining the promissory commitment regardless of whether the promisor wishes to hedge the commitment in advance. The paradigm example might be the family emergency, which gives the promisor an objection to a requirement of promissory fidelity that is more compelling than the promisee’s continuing objection to principles that permit the promisor to attend to it.

    Now it is in the nature of emergencies that they typically involve surprising developments, which couldn’t reasonably have been anticipated in advance. Maybe your mother has a heart attack and needs to be taken to the hospital, or there is someone drowning in the waves on your way home from work, and you are uniquely positioned to save them. The implicit escape clauses built into promissory (and many other) moral commitments are a way of accommodating directed obligations to the fact that life is fluid and chaotic and that circumstances are capable of evolving in totally surprising ways.

    Compare these scenarios with contingencies of a similar kind that could reasonably have been anticipated by the promisor. Maybe your mother needs to go to the clinic in a couple of days, and is already relying on you to get her there. Or perhaps you are the only good swimmer in town, and it is a regular occurrence that there are people in maritime distress who need rescuing on your way home from work. If you promise to do something in a couple of days that would preclude you from addressing these contingencies, I don’t think their obtaining would get you off the promissory hook. You should have anticipated them in advance, and if you wished to accommodate them you ought to have put them explicitly on the agenda of the promisee, as conditions that would be understood by everyone to set the terms of the promissory commitment.

    A different way to put the point is that promisors have a legitimate and weighty interest in structuring principles of promissory commitment to allow for the kind of unexpected emergencies that are part of human life. But they don’t have the same interest in being able to escape from their commitments to address contingencies that they readily could have anticipated at the time when the commitments were undertaken. A principle that allowed this would degrade the value of promissory assurance for the promisee. And promisors already have a way of hedging the commitment by explicitly incorporating the anticipated contingency into the content of the promise, in the form of an explicit condition. You could always just say: “I’ll meet you for lunch in two days, unless I’m unable to find somebody else to drive my mother to the hospital.”

    The case with the genuine implicit escape clause is interesting from the standpoint of trust. Micha says, and I agree, that the promise might reasonably enable the promisee to trust the promisor to fulfill the terms of the promise, despite the implicit conditions to which the moral commitment is subject. A corollary of this is that, in the scenario in which promisor’s mom has a heart attack and needs to be taken to the emergency room, the failure to show up to lunch is not well understood as a betrayal of promisee’s trust, precisely insofar as it is justifiable to the promisee. This brings out the fact that trust is not a matter of blind confidence that another person will behave in a certain way, but is sensitive to the moral considerations that guide and regulate the behavior of the individuals in whom trust is placed.

  29. Many thanks, Ignacio, for that further question. I said in response to Micha that I was running out of steam, and that is even more the case right now. So I’m not sure how coherent I’ll be in reply!

    About “the contextual value of moral norms”, as the explanation for the implicit escape clauses: I’m not sure I have a firm grip on the very idea of the value of moral norms, so I’m not sure I’d want to appeal to the contextual nature of the alleged value to make sense of the phenomenon we’ve been discussing. I think I’d explain it more in the terms of my last reply to Micha, by appeal to the interest agents have in being subject to principles that allow them to address important contingencies of life that, in the nature of the case, outrun their capacity to anticipate.

    You are right that the kinds of contingencies we have been discussing up to now involve the agent’s relation to other parties. There is the agent’s relation to their mother in the case in which she has a heart attack and suddenly needs to be taken to the emergency room, or your relation to the drowning person that you are suddenly in a unique position to save (albeit at great risk to yourself, so that it isn’t exactly something that you owe it to the drowning individual to do).

    But I think I can imagine cases in which unforeseeable environmental contingencies might sometimes similarly get one off the hook. This is especially apt to be the case in situations in which it is one of the agent’s important personal projects to advance environmentally-significant goals. Maybe I’m devoted to saving the coastal redwoods from degradation, and an exceptional and surprising opportunity comes along to make progress toward this goal (a senator on the environmental affairs committee calls me up out of the blue, and wants to talk to me about the importance of the cause etc.). I think this might function in the same way we have been discussing, as the kind of consideration that could undermine my obligation to meet the promisee for lunch—though I would still owe it to them to provide timely warning and compensation if they suffer harms as a result.

  30. Thanks so much for that further reply, Jay. You are definitely going above and beyond! I won’t reply properly, not least because the evening over here is advanced at this point to put it mildly, except to note that I very much like your additional consideration to do with the conditions under which a release clause is in order. It seems to me that there are two interacting considerations at work here, moral weightiness as compared to the promissory obligation on the one hand and foreseeability on the other, and their precise mode of interaction may well merit further investigation. Not tonight and so no longer here though. Thanks again! I thoroughly enjoyed this.

  31. I think we can all agree that Jay has done a truly heroic job responding at length to an enormous number of comments and deserves a virtual round of applause! Thanks, Jay, for taking all of our input so seriously and engaging with it so gamely. Hope you can get some rest tonight without being haunted by dreams of missing lunch dates to save drowning people!

  32. Thanks to Steve for the probing and thoughtful exchanges, which have really helped me to understand much better where we agree about things, and where (and why) our positions diverge. Thanks too for the fantastic comments and probing questions from all the discussants. I’m gratified to see that there is still robust interest in the kind of grand ethical theory that both Steve and I tend to go in for, and it has been tremendously rewarding to think about these issues together with you. Thanks finally to Victor Tadros, who had the idea to host this session on Pea Soup after the excellent conference in Freiburg last summer that all three of us took part in (on “Relational Morality and Criminal Law”).

  33. Dear Jay and Steve, a heartfelt thanks for your thoughtful responses and this wonderful exchange between the two of you and with all the other discussants! It’s been truly rewarding and I thoroughly enjoyed this.

  34. Thanks so much to everyone who has been involved in this terrific discussion, and especially to Steven and Jay for their wonderful posts and engaging replies.

    All the best


  35. Sorry to be responding so late to comments and questions:

    For Brendan:

    About our disagreement about whether analyzing moral obligation in terms of normative reasons for action can be made to work. My main worry is the following. In many situations the amount of burden the agent would have to bear to respond to a need will turn the situation from one in which they are protected by an agent-relative, permission, option, or prerogative (pick your term) to one in which the burden is one they can reasonably be expected to bear. Suppose we are at the point of “least noticeable difference”. In situation A, the agent costs are low enough, that helping is obligatory, but in situation B, where the costs are just slightly greater, it is optional. In such situations, the reasons absent whether they justify a moral obligation or not have roughly similar weight. Once, however, we determine that they support an obligation in B, but not in A, and add in the obligation in B, we see that now there is a big difference, made by the fact of obligation. (Raz would say that in the one case, indeed, there is an exclusionary reason, but not in the other.) So I don’t see how we can reduce the obligation to the obligation-making reasons.

    Maybe that’s a little clearer? Anyway, thanks, as always for pushing me on this, which you’ve been doing for over fifteen years!

  36. Oh, to Brendan again, I take your point about the directed pro tanto. What I really mean to say is that although we can have directed pro tanto within obligor/obligee pairs, it doesn’t make sense beyond that. Does that make sense?

  37. Finally, for Ariel: Thanks so much for your really interesting post. I’ll have to think more about Kant and Leibniz on relations (which is really above my pay grade!). Just one thought on whether my relationality as relating is eviscerated (if that’s the right term) when I bring in the transcendental idea of the moral community. What I am thinking is that when you blame me as a representative of the moral community, and so take up that standpoint (or “wear that hat”), it is still you, as one person among others, who are relating to me, as one person among others. There is no “spooky stuff,” as the kids say. Does that help?

  38. Super finally. Thanks so very much to Jay and everyone else. This has been really great for me, both to continue to hash things out with Jay (as we’ve been doing ever since that wonderful afternoon of April 6, 2007, in San Francisco, when we both took part in an Author Meets Critics session on SPS along with Chris Korsgaard and Gary Watson) and to get to engage with everyone else on these issues. To be continued . . .

Comments are closed.