Pamela Hieronymi interviewed by Richard Marshall.

[Photo: Rose Lenehan '11]

Pamela Hieronymiis always brooding on forgiveness, on the two errors about it, on why blame is tough, about its core, its justification, on why vice can't exempt, on the minimalist attraction of the contractualist moral theory of Tim Scanlon, on two kinds of agency and 'the right kind of reason', on believing at will, on intending, Kavka's Toxin Puzzle, on trust as a test case and why we shouldn't confuse technology with college teaching. Yup, this is an original hair-trigger. Jiving.

3:AM:What made you become a philosopher and so far has it been worth while?

PH:I’ve been interested in philosophical questions as long as I can remember. When I was in the fifth or sixth grade, my father gave me a copy of Descartes’ Meditations, and my nerdy friends and my nerdy self sat in the back of the bus trying to figure out if we were being deceived by an evil demon. In deciding my undergraduate major, I asked myself where I could take the required number of courses without taking courses I wasn’t interested in taking. The answer was obvious. Still, I didn’t go directly to graduate school after my BA. I spent two years working at a non-profit in Washington DC. It wasn’t until that time that I decided to go the academic route. It has worked out for me. I have been very fortunate.

3:AM:You like to untangle confusions in big but everyday ideas. One of the first of these is the ethics of forgiveness. In the literature and in everyday parlance, forgivenesshas to be uncompromising – we’re not to do it begrudgingly. That suggests we all think that if we compromise our forgiveness it isn’t forgiveness. Don’t we risk losing the contrast Schiller makes between doing the right thing gracefully or not in thinking that? Why can’t I just forgive badly or awkwardly or clumsily?

PH:I guess I do like to untangle confusions in big but everyday ideas—I hadn’t thought of it that way, but that’s true.

You can definitely forgive badly, awkwardly, or clumsily. In my first publication I claimed that forgiveness must be “uncompromising,” by which I meant that you cannot genuinely forgive unless you continue to believe three things: that the wrong was a serious one, that you deserve not to be wronged in that way, and that the wrongdoer is someone who can be expected not to behave as he or she did. If you eliminate your resentment by giving up on one of these three claims, you are doing something other than forgiving. To give up on the first or the second claim is to excuse, not to forgive. To give up on the third is either to excuse or to do something like dismiss the wrongdoer or hold her in contempt—also not forgiveness. So, forgiveness must be uncompromising, to be forgiveness at all. But genuine, uncompromising forgiveness can certainly be done badly or awkwardly or (though this is more interesting) ungracefully.

3:AM:So you say everyone discussing forgiveness makes the same two errors. What are these and how does your approach avoid them?

PH:The first concerns how we relate to our own emotions. In thinking about how to “overcome” resentment (and so achieve forgiveness), people tend to think of resentment as a kind of force or affect to be managed—to be “dispelled,” “let go of,” or “conquered.” Of course, we can manage our own emotions (some of us are better at it than others). That is one way we relate to them. But emotions like resentment are not just forces at work within us, to be managed—like a disease or a headache—they also embody, or constitute, our take on some aspect of the world, our sense of what is important or threatening or wonderful or wrong. And so we relate to them, not just as forces to manage, but also as the stance on or posture towards the world that we, ourselves, inhabit and sustain. And that means we might alter them, not merely by “dispelling” or “conquering” them, but also by changing our mind about what is important or threatening or wonderful or wrong.

We do sometimes revise our resentment in response to new information in this second way—in something like the way we revise our beliefs in response to evidence. For example, we revise our resentment when we learn it was just an accident. In this example, though, we are excusing, rather than forgiving.

I believe an adequate philosophical account of forgiveness must do more than tell us what good reasons we have for managing ourselves out of resentment—it should instead provide the reasons for which a forgiver revises her take on the wrongdoer and the wrong, in something like the way we revise our resentment when we learn it was just an accident. It should be, as I put it, an “articulate” account. But it is hard to see how an account of forgiveness can be both articulate and uncompromising—it is hard to find reasons for revising resentment that do not amount to excuse or contempt, rather than forgiveness. This challenge provided the title for my paper: “Articulating an Uncompromising Forgiveness.”

Back to your question: if you make this first common mistake—if you think of resentment simply as an affect to manage—you will not be looking for an articulate account. You might agree that not just any “overcoming” of resentment will amount to forgiveness (taking a pill that eliminates resentment, e.g., would not count as forgiveness). But you will not be looking for the internal logic, so to speak, of resentment, and so you will not be trying to understand the way in which a forgiver revises her take on the world, the wrongdoer, and the wrong.

The second mistake is related to the first. In trying to think about how to “overcome” resentment and so forgive, many people appeal to love, compassion, or empathy. They think that, by achieving compassion for the wrongdoer, you will eliminate your resentment in a way that counts as forgiveness. But this is to misunderstand the relation between compassion and resentment—or between love and anger. Love and anger are not incompatible, and neither are compassion and resentment. Suppose we are close friends, but, in a business transaction, you have seriously wronged me and, though you are uncomfortable about it, you are also unrepentant. Moreover, those around us do not take your wrong very seriously (“bankers will be bankers,” they might say). If we are close and I am kind, I may feel compassion for you—for the way in which you are distorting your life and your relations in order to avoid acknowledging your wrong and seeking forgiveness. Your discomfort need not please me—it may add to mine—and I need not desire your suffering. But I also need not, for all my compassion, be any less angry about or resentful of the wrong you did me. Humans are emotionally sophisticated creatures, able to feel many things at once, and, while compassion and resentment may compete for our attention—it may be hard to attend to both in the same moment—they are not otherwise incompatible. So, to tell someone that she needs gain to in compassion or empathy in order to overcome her resentment may be to wrongly accuse her of lacking something she already has. In counseling compassion, we may be confusing kindness, or what we might call a readiness to forgive, with forgiveness itself.

3:AM:Blame is another idea you examine. We tend to think that blaming requires more justification than describing or grading. But you think this is confused and that once we understand the link between blame and judgment the idea of needing extra justificatory stuff goes. Can you explain your thinking here?

PH:Blame is tough—it’s a hard topic, for many reasons. High on the list is that we don’t all mean the same thing by “blame.” And, in philosophy, we often aren’t using it in the way the word is used in ordinary life. So, I tried to write my paperon what philosophers call “blame” without using the word “blame.” That didn’t work.

It’s relatively common, in the literature, for philosophers to say that blame is unlike mere grading or negative evaluation, in that it is somehow more “forceful” or “deep.” Suppose I point out that your feet are flat, or that your singing was sharp, or that the move you made in your chess game was ill-considered. I am criticizing you. I am evaluating either some aspect of you or something that you did against a standard, and I am claiming it falls short. But I am not blaming you. Blaming, the thought goes, is somehow more than mere grading or negative description. The question is, what more is there to blame?

There are (at least) two ways in which people think blame goes beyond negative evaluation. First, people tend to think that blame is a negative evaluation with an extra tag line. The extra tag line reads, “and if you had paid attention and tried harder, you could have done better!” In cases in which the tag line is false, they think, it is obviously wrong to blame.

But the tag line isn’t enough to capture the entire difference between blame and mere description. It may have been that, if you had paid attention and tried harder, you could have done something about your chess move or your singing. Even so, a negative evaluation of those won’t be blame (or, at least, so say the philosophers—I am less convinced about the ordinary use). The criticism in these cases is thought to lack what Jay Wallace calls the “quality of opprobrium.” So, what more do we need to add, to get to blame?

Blame is often thought (not only to come with the tag line but also) to involve a kind of sanction. The sanction is sometimes thought to come in the form of a negative emotion, like resentment or indignation. I call this the “penalty-conception” of blame. I think we come by it naturally: Start with a case in which we would naturally use the word “blame”: “The Dean is to blame for the delay.” In such a case, we tend to think the Dean should be the one to make things right, if that is at all possible—if he is to blame for the delay, he should put in the work to remedy its consequences. Sometimes it isn’t possible to make things right. And, in some of those cases, we think that the Dean, or the one who caused the traffic accident, should nonetheless be burdened with some penalty. If the mess can’t be cleaned up, perhaps the mess-maker should be made to clean up something else, or to pay a fine, or at least to sit in the corner for a while. And, I think it is easy to think that, if we can’t make even that happen (perhaps because we aren’t in the right kind of institutional context), then at least the mess-maker can be made to suffer as the target of our negative affect. He can endure our opprobrium.

I think this penalty-conception of blame is at least part of what underlies and animates much of the philosophical literature. I find it natural, but also unattractive. It confuses what I would like to call “blame” (if I am going to use the word at all) with something closer to what I would call “guilt-tripping”—trying make someone feel bad about what he or she has done.

So, again, others have tried to answer our question—what more is there, to blame?—by looking at either what extra thing blame claims (“you could have done better if you tried harder!”) or what extra thing it involves (some sanction or negative affect).

I think all of this is unnecessary. I think we can identify the extra “force” of blame just by looking at what blame is about—just by considering the subject matter, so to speak, of this particular negative evaluation. (In saying this, I am following Tim Scanlon.)

In criticizing your feet, your singing, or your chess move, I am negatively evaluating each of these against the standards that apply to such things. Likewise, in the case of moral criticism, I am saying that you have failed against the “moral” standard. But what is that standard? It’s remarkable how unclear we are about this thing—morality—that many of us think very important. But I think it at least relatively safe to say that, at least in most cases, moral standards are those that require respect for persons. If so, then, in violating the moral standards, you have shown others disrespect, and, in saying that you have failed against the moral standard, I am saying that you have shown disrespect either to me or to others like me.

But that “negative evaluation”—that you have shown disrespect to me or to others like me—is already, in and of itself, more, or at least differently, forceful than claiming that your feet are flat, your singing off, or your chess move ill-considered. It is not more forceful because being respectful is somehow more important than being in tune or clever. Rather, this negative evaluation is differently forceful because it is about us—the one making the evaluation and the one being evaluated. The subject matter of this evaluation concerns, or carries implications for, the relation in which we stand. And so this criticism gets personal—to make this negative evaluation is already to be in an altered relation to you. (Our relation is altered, just by my making the evaluation, privately. I need not confront you or otherwise express my evaluation to you, in words or deed. We need not even meet. In fact, our relation has been altered even if my evaluation is false.)

I think that’s the core of blame: a negative evaluation against expectations of respect or regard. This is an extraordinarily minimal conception of blame. It leaves out many interesting, important things (like the tag line, and the sanction, and most everything that Freud and Nietzsche wanted to talk about). But I do think that the judgement that someone has been disrespectful or shown disregard will, all by itself, give blame its own distinctive force and so distinguish it from “mere grading.”

Now, you asked about justification. I’ll be quick about this. If we think of blame as a kind of sanction or penalty, as on the penalty conception, then blame would have to be justified in the way sanctions or penalties are justified. Sanctions and penalties face high justificatory burdens. They can be shown unjustified in lots of ways, including by appeal to their negative consequences and by appeal to the fact that the one being penalized did not have an adequate opportunity to avoid those consequences. But if, instead, blame is simply the judgment that someone has been disrespectful (not even the expression of that judgment, just the judgment itself), then it is justified just in case that judgment is justified. The fact that the judgment has bad consequences, or that it’s target did not have an opportunity to avoid those consequences, is beside the point (assuming these do not show the judgment inaccurate). This very minimal conception of blame faces very minimal justificatory burdens.

3:AM:Your view involves the possibility that someone can be morally accountable for something they can’t possibly achieve. This seems wild: how can I be held accountable for something I haven’t a chance of avoiding? Surely if something is beyond my rational capacity it’s not my fault and I shouldn’t be blamed?

PH:“Ought implies can” is, I think, one of the most misunderstood, misused, misapplied slogans in philosophy. There is some truth to it—but it sows far more confusion than clarity.

Despite its origins, the slogan is now typically understood, not to herald ability (“look, you ought, so of course you can!”), but rather to limit obligation (“look, you can’t, so it can’t be the case that you ought”). That is to say, the slogan is now employed in its contra-positive form. But inability doesn’t always cancel obligation.

Now, I need to grant, right away, that obligations are often sensitive to abilities. If I can’t swim, then I am not obligated to jump into the water and save the child. Perhaps, in a different culture (not ours, but one that requires regular travel over water or prone to constant flooding), I might nonetheless do wrong by the child, because, in that context, everyone is obligated to learn to swim by the time he or she has reached my age, and I have failed to do so. But, even in that context, if I could not have learned to swim—if, say, I do not have use of my limbs—I cannot be obligated to do so. So, we need to grant that a lack of ability can shape what obligations people fall under.

But a sensible moral theory must be able to put this fact together with another commonplace: humans are born into the world in need of personal development and moral education, and such development and education is a messy and hazardous affair. The ability to satisfy moral demands must be learned, and the psychological resources needed for it must be gained. Often enough, that process goes wrong. Tragically, many people reach adulthood too insensitive, too touchy, too competitive, or too self-absorbed to be able to show other people the respect and regard that is (nonetheless) owed to them. So, these unfortunate souls are destined to do wrong—they lack the psychological resources required to consistently show others respect. I like to refer to this commonplace grandly (and sloppily) as “the fact of original sin.” It has always been hard to square original sin (the fact that one is destined to do wrong) with the doctrine that ought implies can.

I doubt that Kant, to whom we owe the slogan “ought implies can,” was much impressed by (what I am calling) “original sin.” His position, insofar as I understand it, was that we know that we are under obligation, and, because we know this, we can also know (what is not an empirical fact) that we are free, and so we can also know that we can do what we are obligated to do (despite appearances to the contrary).

But, if you are less sanguine about human ability to rise above empirically-given psychological limitations, and if you are impressed by the fact that obligation is sensitive to ability, and, especially, if you think blame includes the tag line mentioned earlier (”and, if you had paid attention and tried harder, then you could have done better!”), then you might, instead, be tempted to think that obligation should retreat in the face of such things as entrenched insensitivity, competitiveness, frailness of ego, or incorrigible jack-assedness, in just the way it seems to retreat in the face of an inability to swim. And this will lead you to the conclusion that incorrigibility is its own excuse (or, in fewer words, that vice exempts).

But this, I think, can’t be right. Vice cannot exempt. If it did, moral demands could not do their job of adjudicating the competing interests of people trying to share a world peacefully. And so I am left to deny the slogan “ought implies can,” when it is used to suggest that, say, the incorrigible chauvinist violates no obligations when he makes a suggestive remark in a professional context or dismisses a candidate on the basis of gender. (I deny “ought implies can” rather than insist that this character somehow does, under it all, have the ability to do what he is obligated to do. The latter seems to me overly hopeful, unhelpful, and, in the end, too severe.)

My thinking, here, is driven by my sense (a broadly contractualist sense, which I owe to Tim Scanlon) of what moral demands, or moral obligations, are and need to do. It seems to me that moral demands are placed upon us by our need to share a world with others “equally real” (to appropriate Thomas Nagel’s evocative phrase). Thus, moral obligations must adjudicate the competing interests of each, treating each symmetrically. And they will not be doing this job if they exempt the incorrigibly insensitive.

They can, however, do the job while acknowledging other kinds of limitations. The principles that would adjudicate the competing interests of each, treating each symmetrically, are, roughly, the terms we would agree to live by, if we needed to reach agreement on such terms, and no one had any advantage to exploit in our negotiation of those terms. Now, consider whether I am obligated to jump in the water to save the small child. Whether I am obligated will depend on the costs, to me (or, to someone in my position), of jumping (and the additional costs of being obligated to jump), as compared to the costs to others of being without my obliged help. In the usual case, the cost to me is minimal (a bit of effort, ruined clothes, running late), while the cost to the child is catastrophic. And so the terms we would agree to, it seems, would require me to help. I could not reasonably object to being so obligated, given the costs to the child. But a difference in abilities changes things considerably. If I cannot swim, not only are the costs to me prohibitively high, but my efforts would serve no purpose. And so the terms we would agree to will not obligate me, if I cannot swim.

What about the costs incurred by those unable to show respect? It is true that it would be costly for them to go through the psychological overhaul required to show respect. (In fact, the costs may be so steep that we can predict with certainty that the vicious will choose to avoid them. Worse, certain vices, such as insensitivity, guard against their own elimination.) It is also true that it will be costly for them to be under obligations they consistently fail to meet. At a minimum, they will suffer from substandard interpersonal relationships. (Though, they will suffer these in any case.) And yet it seems to me that, in adjudicating the competing interests of people equally real, we need to say that other peoples’ interests in having their interests respected outweigh these costs to the incorrigible. Even given the costs to the vicious, they are still subject to obligation—even if they will never satisfy it.

And so this broadly contractualist way of thinking about moral demands can put together the fact of original sin with a limited sensitivity of obligation to ability.

Philosophers often think this limitation is outrageous: they insist that we cannot be obligated to do things that we cannot do. But, in fact, it seems to me that a great many, maybe most, demands in ordinary life are like this. The demands of teaching, or parenting, or of being president, do not bend to accommodate the abilities and limitations of the particular people who fill the role. We hope the people who find themselves in those positions will rise to meet the demand. Sometimes they do not. And if someone is too insensitive, or too disorganized, or too frail of ego, to competently fill the role, the demands of the role will not adjust to fit—they cannot, because students, children, nations, and organizations have certain legitimate needs. So, the person in the role may be condemned to be a bad parent or a bad president. That may be tragic. But it is neither outrageous nor confused. The source of the demands—the needs and interests of others—are simply not the sort of thing that bend to those limitations.

The pedagogical demands placed on students, or on children, provide an exception. These can and should be custom fit to the abilities of the student or the child (set, in fact, just a tad beyond them). If you think of moral demands (not as the demands put upon us by our need to share a world peacefully with others equally real, but rather) as that which is asked of us by a divine parent, or by those (perhaps our entire community) coaching us to become better exemplars of humanity (and penalizing us when we fail to pay attention and do our best), then it might make sense to think that moral demands should bend to the abilities of each. But that is not how I understand moral demands.

One might try to take a middle position. One might grant that the incorrigible jackass is not exempted from moral demands; he is under obligation. He fails to meet those obligations all the time, predictably. But, one might insist, we can’t blame him: because of his inability, he is excused for violating his obligations. This is a common position. When people say, “ought implies can,” they very often mean, instead, “blame implies could have done otherwise.”

But now we are back to blame. If you think blame must always carry the additional tag line, “and, if you had paid attention and tried harder, you could have done better!”, then, of course, blame implies could have done otherwise. But it is not the case that all forms of (inevitably negative) reaction to disrespect include this tag line. And, as noted above, I think that the judgment that someone has shown disrespect—and, even, the judgment that someone is incapable of showing respect, because she is so insensitive—is, by itself, a kind of blame. It already carries blame’s distinctive force. It has immediate implications for the quality of your interpersonal relationships. That is enough.

3:AM:Is your view that we have to blame if we have any moral commitments at all? And does it follow from your position that if we don’t have moral commitments we can still blame, or is the connection one where without moral commitments blame goes too?

PH:If blame is just a judgement that someone has shown disrespect, then, if you have enough moral commitments to think that there are standards of respect, and you think they are sometimes violated, then you will already be in the business of blaming.

3:AM:Both blame and forgiveness are central to ideas about agency and having the right reasons and attitudes for doing something. But there’s a philosophical problem involved in having the right reasons, labeled ‘the wrong kind of reason problem.’ It hinges on ambiguity between two different kinds of considerations that count as reasons for certain attitudes, and only one of them is actually relevant. Can you explain what the ambiguity involves and how you attempt to sort it out?

PH:There are lots of things people mean by “the right kind of reason.” The particular issue I am interested in, under this heading, has already come up in our discussion of resentment.

Let’s start, though, with a currently standard account of what a reason is. A reason is widely taken to be a consideration that counts in favor of an action or attitude. So, the fact that it is getting late is a reason to go home—it is a consideration that counts in favor of going home. And, his desire for revenge and his ready access to the home are reasons to believe the butler did it—they are considerations that count in favor of that belief. Notice that, on this view, a reason is a consideration standing in a relation—namely, the counting-in-favor-of relation—to an action or to an attitude.

There is no problem, as far as I can tell, with thinking that reasons for acting are considerations that count in favor of acting. But to understand reasons for attitudes as considerations that count in favor of those attitudes is to ask for trouble. The issue is easiest to understand in the case of belief.

So, suppose that you would feel much better if only you could believe that everything will be okay. The fact that it would make you feel better counts in favor of believing everything will be okay—in just they way that it counts in favor of getting more sleep or taking an aspirin. So, if reasons are considerations that stand in the counting-in-favor-of relation to actions or attitudes, it seems that the fact that it would make you feel better should be a reason to believe. But, there is something wrong with this reason—it is the wrong kind of reason to believe. In contrast, the fact that your test results were negative or that you were just offered the job also counts in favor of believing everything will be okay—but these seem to do so in the right sort of way; they seem like the right kind of reason for that belief.

This same distinction seems to appear for other attitudes. So, in one fanciful example from the literature, the fact that an evil demon has threatened to harm you unless you admire him counts in favor of admiring him, but, you might think, it does so in the wrong way (you will think it’s the wrong way unless you think such ruthlessness is admirable). The demonic threat would also seem like the wrong kind of reason for resenting your co-worker or being envious of your neighbor. But such a threat counts in favor of admiring, resenting, or being envious, in just the same way it would count in favor of giving the demon your money, had he demanded that instead.

In the case of belief, it is relatively easy to say which are the right kind of reasons: they are the ones that bear on the truth of the belief. Other reasons, which merely show the belief good to have, are the wrong kind. But philosophers have had trouble giving a principled account of what seems to be the same distinction, as it appears across a range of different attitudes.

I think the trouble starts with the standard account of what a reason is and that it can be resolved by replacing that account. So, consider an alternative:

A reason, I would insist, is not a consideration standing in relation to an action or attitude—standing in relation to an event or psychological state. That seems to me a kind of unkosher blending of the rational and the empirical (to put things grandly). A reason is, rather, an item in a piece of (actual or possible) reasoning. Reasoning, in turn, is (actual or possible) thought directed at some question or conclusion. Thus, reasons relate, in the first instance, not to psychological states or events (not to items in the world), but, rather, to questions or conclusions (to objects of thought). A reason, I would say, is a consideration that bears (or is taken to bear) on a question or conclusion.

If we use this as our basic account of what a reason is, then, if we want to relate a reasons to actions or attitudes—to items in the world—we must do so via some question or conclusion. We have to consider the relation between questions and attitudes. By requiring this extra step, we bring into view the use of reasons by thinking subjects—we bring rational agency into view. Doing so helps with many different philosophical problems, one of which is this wrong kind of reasons problem.

So, consider how attitudes relate to questions. One way is this: we form certain attitudes by answering certain questions or coming to certain conclusions. So, when you answer for yourself the question of whether everything will be okay, or whether the butler did it, you form a belief. If you answer the question positively, you form the belief that the butler did it. Likewise, if you answer positively the question of whether to go home now, you therein intend to go home now. More generally, in concluding that p, you believe p. In deciding to x, you intend to x. Certain attitudes happen, one might say, as thinking subjects answer questions or draw conclusions.

But there is another, very different, way in which attitudes relate to questions: attitudes sometimes appear in the question being asked. You might ask whether it is good to believe p, whether you ought to intend to x, why she resents him, or how it came about that he intends to go home now. The reasons that bear on some of these questions explain the attitude mentioned—they explain how that attitude came about. Reasons that bear on other questions will, instead, show something good, right, valuable, or appropriate about that attitude.

We can now see why the standard account asks for trouble: considerations can “count in favor of” an attitude in two very different ways. They can bear on a question such as whether p or whether to x—a question the answering of which amounts to forming the attitude. These considerations “count in favor” of the attitude in the way that your negative test result counts in favor of believing everything will be okay. But considerations can also count in favor of an attitude in another, entirely straightforward, way: they can bear positively on the question of whether the attitude is in some way good to have. So, the fact that it would make you feel better counts in favor of believing everything will be okay by bearing on whether that belief is good to have.

Once we understand the ambiguity, we can solve the wrong kind of reasons problem: when considerations count in favor of the attitude in the first way, they are reasons of the right kind. When they do so only in the second way (without also doing so in the first way), they are reasons of the wrong kind.

Now that we have the distinction on hand, we can see that it was at work in our earlier discussion of forgiveness: when I was earlier insisting that we should provide an account of forgiveness that allows us to see, not just how to manage our emotions, but how to revise them according to what I then called their “internal logic,” I was asking for an account that appeals to the right kind of reasons for resentment. (Some people doubt that you could arrive at emotions like resentment by settling questions. But I believe that, insofar as it makes sense to ask you why you resent, that is, insofar as it makes sense to look for your reasons for your resentment, there will be some questions those reasons bear on. Unearthing them will require difficult philosophical excavation. I expect that work to unearth things of value.)

We can also see this distinction at work in the earlier discussion of blame. If we think of blame in the minimal way I suggested—as the belief that someone has been disrespectful—then blame cannot be shown unjustified because it it has bad consequences for the one blamed. Consequences are the wrong kind of reason to criticize a belief.

Before leaving the topic, I want to draw attention to one other thing: even bad reasons can be of the right kind. The good/bad distinction runs orthogonally to the right kind/wrong kind distinction. Suppose you think everything will be okay because you are a Leo and the moon was in thus-and-such position. That’s a very bad reason. But it is, at least, the right kind of reason—it is something you take to show that everything will be okay, rather than something you take only to show that belief would be good to have. It is a bad reason of the right kind. There can also be bad reasons of the wrong kind: believing everything will be okay might upset your partner and so satisfy your spiteful desire to cause him or her suffering. That seems to be a bad reason of the wrong kind.

I draw attention to the fact that the two distinctions cut across one another, because I find that philosophers often think that, by solving the wrong kind of reason problem, we will also solve problems that could be solved only by locating the good reasons for an attitude. I think examples like the ones I just gave should be enough to show that this is not so. But I also have a view about why it is not so: the distinction between the right and the wrong kind of reason concerns, not the justification of our attitudes, but our relation to them, as agents—and (controversially) I do not think that facts about rational agency will tell us which are the good or bad reasons.

3:AM:Agency is linked with intentions and there’s a lot about intentions going on at the moment. Anscombe is getting fashionable isn’t she? You argue that you can’t intend at willin just the same way as you can’t believe at will. This seems a little strange to me at first sight because perhaps I’m thinking that we wrap up intentions with choice in a way that belief isn’t. How does your thinking about this make good your claim?

PH:Let’s first try to get clear on the sense in which you can’t believe at will. It is much harder to specify than people typically realize.

There are two different, robust, ways that you can successfully exercise your agency with respect to your beliefs, neither of which counts as believing at will. (These correspond to the two ways you might relate to your resentment.) One is this: you can think about what is true. You can come to conclusions, or change your mind, about what is the case. In doing so, you will form or revise your beliefs. I call this exercising “evaluative control” with respect to your beliefs. You exercise evaluative control by finding the right kind of reasons (which I also call “constitutive” reasons) convincing.

But suppose you find convincing only the wrong kind of reason for believing (which I also call “extrinsic” reasons). Perhaps it would make you feel much better if you could believe that everything will be okay, or if you knew that your children arrived home safely, or if you thought you would win the match. Because you do not take these reasons to bear on the truth of the relevant belief, you cannot, simply by finding them convincing, come to believe—they bear on the wrong question. (By finding convincing reasons that you take to show it good to believe p, you will arrive at a second-order belief: the belief that it is good to believe p. You will arrive at the first-order belief, the belief that p, by settling the question of whether p. But you do not take the reasons at hand to bear on that question, and you can’t settle a question for a reason that you, yourself, do not take to bear on it.) So, you cannot believe simply by finding convincing the reasons you only take to show a belief good to have. You cannot exercise evaluative control employing extrinsic reasons.

But you are not simply helpless. Our beliefs are not only our answers to certain questions, not only our take on the world, they are also facts about us, facts in the world, of a more-or-less ordinary sort. They interact in more-or-less predictable ways with their environment. And, if something interacts in more-or-less predictable ways with its environment, we can think about how to take action and change it. We can and do act to change other peoples’ beliefs, all the time. Often enough, it would be good if someone believed this or that (that we are home, that we won’t be able to make the party, that our product is the best on the market), and we have very effective strategies for bringing it about that other people believe this or that—ranging from simple telling, to presenting evidence, to elaborate schemes of deception. We can, likewise, take action to affect our own beliefs. We have less need to do so—since we need not tell ourselves what we already know—and we are more constrained in doing so—since deceiving yourself is trickier than deceiving someone else. But we can and sometimes do, nonetheless, take action to change our beliefs. If you want to believe your children are home safe, you could give them a call to see if they have arrived. If you want to believe everything will be okay, you could take some anti-anxiety medication (or perhaps induce amnesia about the relevant piece of history). I call taking action to affect your beliefs exercising “manipulative” or “managerial control” over your beliefs.

So, there are two very robust ways in which we can exercise our agency with respect to our beliefs: evaluative control, on the one hand, and managerial or manipulative control, on the other. You can change your beliefs by changing your mind about what is the case, or you can change your beliefs by taking action aimed at changing your mind. But neither of these are what people have in mind, when they think about believing at will. So, what would it be to believe at will?

It would be to be able to believe, for reasons that you take only to show the belief good to have, without having to take action to bring your belief about—to believe, for extrinsic reasons, without exercising managerial or manipulative control. Put differently, it would be to believe in the way you act—you act for reasons that you take to show the action good to do, without needing to first take another action so as to bring it about that you act.

I argue that you cannot believe in the way you act—nothing done “directly” for reasons extrinsic to a given belief could be that belief. The best you could do would be to act so as to bring about the belief very, very quickly—to bring about the belief by what philosophers call a “basic action.” But, I argue, that would just be exercising managerial control very efficiently. It would not be believing at will.

I won’t go through the argument to this conclusion. I’ll only try to show what the analogous claim would be, in the case of intention.

To intend at will, in the sense at issue, would be to intend, for reasons you only take to show the intention good to have (reasons you do not take to show the action to be done), without first taking action to bring it about that you intend.

We need an example, but it is harder to find examples, in the case of intention, because it is harder to find reasons that show a intention good to have without showing that the relevant action should be done. It is harder to find such reasons, because the fact that an intention would be good to have can, itself, be a reason to perform the action. (In contrast, the fact that a belief is good to have only very rarely shows that belief to be true.) Suppose you would be upset if I did not at least intend to attend your party. You are, in fact, less concerned that I actually attend than that I at least intend to attend. Now, I might take myself to have reason to have the intention—I do not want to hurt your feelings. And that reason can, in turn, be my reason for attending—I can take the fact that you are hurt by my lack of intention to be a reason to come to your party.

But this is not always so. Kavka’s “Toxin Puzzle”provides one kind of example. But the most intuitive example, here, is the romantic one. Suppose you have no intention of marrying your partner, and he or she is very unhappy about that fact. You like to please your partner, and you would be very happy to house the intention, so long as you do not actually have to go through the wedding ceremony or enter the legal arrangement. In this case, you have extrinsic reasons for the intention to marry—reasons you take only to show the intention good to have, without showing the action to be done. And, in this case, it seems you are in a position analogous to the one you are in, when you would feel much better if only you could believe that everything would be okay. And, I argue, just as you cannot, in order to make yourself feel better, simply believe—you will not believe unless you are convinced that everything will be okay, and you do not find the reasons at hand sufficient for that conviction—so, too, you cannot, in order to make your partner feel better, simply intend—you will not intend unless you decide to act, and you do not find the reasons at hand adequate for that decision. The best you can do, in either case, is to take action to bring it about that you have the attitude in question. But that is not believing or intending “at will,” in the relevant sense.

3:AM:Trust is a big concern, and you have an interesting take on it. Why should we trust other people? we ask and you argue that to whatever degree we have reasons for trusting someone we fail to trust them to that degree. Is that right? So are you saying that trust is covered by counterprivate logic like modesty is (I can’t know I’m modest because that’s immodest, but another person can) or that its sui generis, and there are no reasons?

PH:I am interested in trust mostly as a kind of test-case: it is a simpler case in which to think about an argument I’d like to present in the more complicated case of virtuous (and vicious) action. I’ll try to explain that argument, since it will help to see the point about trust (and, I think it’s interesting in its own right).

I assume that an action is kind (rather than, say, conscientious or spiteful) if it was performed for certain reasons and not for other reasons. If you decide to help your colleague because she is exhausted, then your helping is (typically) kind. If instead you help her in order to ensure the job is done competently, your helping is conscientious—or perhaps meddling. Which adjective describes your action depends on the reasons for which you acted. The reasons that would qualify your action as kind I would call the reasons constitutive of kindness.

But, the reasons constitutive of kindness do not exhaust the considerations that count in favor of performing a kind action. You might, e.g., want to impress your colleagues by acting kindly, in order to advance your own self-serving agenda. This prudential reason counts in favor of performing a kind action, but it is a reason of the wrong kind. It is an extrinsic reason for performing a kind helping action.

Though it is not entirely straightforward, I believe I can argue that, just as you cannot believe at will, so you cannot perform a kind (or conscientious or spiteful) action for reasons extrinsic to it. Insofar as you act on the extrinsic reasons, to that extent you will not be doing what the reason recommends. You will be, at best, engaging in self-management—acting so as to bring it about that you perform a kind action.

In “The Reasons of Trust” I present a roughly analogous argument, for the case of trusting someone to do something. Full-fledged trust, I argue, requires what I call a “trusting belief” that the person will do what you are trusting her to do. To illustrate: suppose it would hurt your feelings if I didn’t trust you to get the job done, and I take this to show that it is important for me to trust you (or, I might be your parent, and so think it is important to your development, or…). But suppose that, in addition, I do not believe you will get the job done. I might, in such a case, decide to “trust” you anyway—that is, I might act on the assumption that you will get the job done, entrusting certain goods to you and giving you the benefit of the doubt. But I will not fully trust you, insofar as—to the degree that—I do not believe that you will get the job done.

From these materials, I argue to a conclusion I find at once surprising and intuitive: to the extent that my reasons for doing what I do concern the importance of having a trusting response, to that extent I do something other than trust (I act so as to avoid hurt feelings, or encourage or promote trust, or to encourage your development, or to discharge duties of trust, etc.). Thus it seems, the degree to which one trusts varies inversely with the degree to which one must rely on the importance of having a trusting response.

In the course of making this argument, I come across an interesting puzzle: Not just any belief that the person in question will do the thing in question will count as a trusting belief. I might believe the person in question will do the thing in question just because I have good evidence that she is reliable, without trusting her. So, what are the reasons that support a trusting belief? I put forward a conjecture. I start with the thought that, when I decide to do something, I believe I will do it on the basis of the (practical) reasons I have for doing it. To illustrate: if I decide go to the noon lecture because the speaker is a friend, I believe I’ll be in the lecture hall at noon, and my reasons for believing this include the fact that the speaker is a friend. I conjecture that trusting beliefs are formed in the same way: when I trust you, I believe that you will do what I am trusting you to do on the strength of the (practical) reasons you have to do it—including the reason given to you by the fact that I am relying on you. So, if I (fully) trust you to drive carefully while borrowing my car, I believe you will drive carefully while borrowing my car, and my reason for believing this includes the fact that I am relying on you to drive carefully while driving my car (in the same way that my reason for believing that I will be at the lecture hall at noon includes the fact that the speaker is friend). This conjecture seems to me to capture the distinctive vulnerability of trust.

3:AM:In ‘Two Kinds of Agency’ you talk about ‘agency over our minds.’ Doesn’t that imply that there must be something else other than our minds (but mind like) controlling our minds? If there is, wouldn’t that need something having agency over it? Isn’t there an infinite regress threatened here?

PH:Your question highlights exactly the problem I am hoping to avoid, by introducing these two kinds of agency. It’s a problem I address, at length, in a book I am working on tentatively titled Minds That Matter.

So consider, first, the idea of control. It seems to me that we arrive at our ordinary notion of control by thinking about our control over our own actions, like going to the kitchen or writing our name, and over ordinary objects, like our plate or our pen. In these cases, we control something—our location, our plate—by bringing it to be as we would have it to be. We control it to the extent that we can make it conform to our thoughts about it. I sometimes refer to this as the “represent-and-cause” notion of control—roughly speaking, you represent some change and then you bring about the change you represent. (When we exercise managerial or manipulative control over our own beliefs or resentments, we are exercising this form of control.)

But, if this ordinary sort of control were the only kind of control, we would have a problem, as your question points out. We would have a problem because, as we exercise this ordinary sort of control over some action or object, as we represent and cause this or that, we are engaged in an activity—namely, the representing and causing—that is not, itself, the object of that exercise of control. It is the controlling itself, not the thing controlled. And so, if the only way we could be in control of anything were by bringing it to be as we would have it to be, then, in order to be in control of activities by which we control some object, we would have to have higher-order thoughts about those activities. We would have to represent and cause our own representing and causing. We would have to make higher-order decisions about which decisions to make. But then the problem would reappear, at the third order. A regress looms, and we would have to conclude that we are not in control of the activities by which we control other things. (Or, we would have to posit some kind of reflexive relation that avoids the regress and explains our control. I think this will not work, and I argue against it in my current work.)

But, it can’t be the case that the activities by which we exercise control are themselves out of our control. We need another notion of control, to capture our relation to them. And I believe we have one. While we control both ordinary objects and our actions by thinking about them—by thinking about what we are doing (at least at some level)—we can control our intentions, beliefs, and certain other states of mind, not by thinking about them, but by thinking about what they are about—we control our intentions by thinking about what to do, we control our beliefs by thinking about what is true, we control our resentments by thinking about the wrongs done (or not done) to us. We can control our own states of mind (not only by exercises of managerial or manipulative control, but also) by exercising what I have called evaluative control with respect to them. And thus we avoid the regress.

I have found that people resist the claim that evaluative control is a “genuine” form of control. I admit it is a strange one. The things over which you exercise evaluative control are exactly those things that you cannot do “at will” in the sense above—they are things, like beliefs, intentions, trusts, resentments, and kindness, for which you can have the wrong kind of reason, and which you cannot do for those reasons. Thus, the things over which you exercise evaluative control are not “voluntary,” in one sense of that word—you cannot do them for any reason you take to show them worth doing. Moreover, you can exercise control with respect to these aspects of yourself without having any intention to do so, and, in fact, even if you are unaware of them.

So, evaluative control lacks many features we associate with control. Nonetheless, I think it far better to expand our notion of control to include this form than to accept that we are do not enjoy any form of control with respect to the activities by which we control ordinary objects and our own actions.

3:AM:You like Tim Scanlon'sidea that contractualism is a moral theory? The parade case of such a theory is John Rawls’s ‘A Theory of Justice’. You argue that many philosophers have missed this dimension and end up diminishing the point of such a theory. Can you say something about this and are you then a contractualist in the moral mode?

PH:I find contractualism, and in particular Scanlon’scontractualism, very attractive, as a moral theory, largely because it is a minimal theory. As I understand it, contractualism as a moral theory takes morality to be given to us by our need to find a way to share a world peacefully with others equally real (much as contractualism, as a political theory, arose from the same need). It thus excludes much of what one might have included, in morality—it is not about how to live a good human life, or how to be a good person, or how to exemplify rational nature, or how to be free, or how to bring about what is good. It is, instead, about the expectations and burdens we could place on each other, in trying to find a way to share our world, without exploiting differences in power. And so it can allow each of us to have differing, and even contrasting, conceptions of what an ideal human life would be and what goods are to be brought about. Insofar as we each have an interest in being left to work out, for ourselves, the best way to live our lives, we need to find a way to share our world that allows each freedom to pursue his or her own conception of the good life, insofar as that pursuit does not interfere with others doing the same. And so a contractualist morality should preserve a significant degree of freedom of conscience.

Many are off put by this minimalism. They think that morality is more, and should do more, than this. It should hold out some higher ideal for human life, or it should be more deeply inescapable, perhaps grounded firmly in our capacities for rational action. But what they find a cost I find a benefit—because what you gain in higher ideals or greater necessity you lose in freedom of conscience.

Sometimes contractualist theories claim “neutrality” between conceptions of the good. This is overblown. A contractualist theory will allow you to pursue your conception of the good only insofar as your pursuit does not interfere with other people pursuing their conception (insofar as their pursuit does not interfere with yours). But certain conceptions of the good will not be consistent with this restriction. To put the point a different way: contractualism requires us to think that the most important thing is to find a way to get along with others who also think that the most important things is to find a way to get along. It thus requires each of us to put finding a way to get along with those with whom we disagree above our other commitments—above the things about which we disagree. And so it requires steep compromise. This is not consistent with every conception of the good.

So, what if we disagree about whether the most important thing is to find a way to get along? If there is disagreement on that point, further disagreements won’t be resolved by compromise and toleration. Some people seem to think the thing to do, in the face of that kind of disagreement, would be to work for an even harder prize—to try to persuade others to come over to our side. We might try to convince them that what we find most important really is most important. Or, we might make arguments appealing to the necessity or inescapability of a more robust form of morality. But it seems to me unlikely that people who are not ready to compromise and tolerate the disagreement will either convince the opposition or be convinced by the opposition. It is more likely, I think, that we will be left to exploit differences in power—to negotiate or to fight.

3:AM:You recently made some sharp and timely comments about the purpose of education in your article ‘Don’t Confuse Technology with College Teaching.’ Here in the Uk there is a crisis in education because policy makers and government administrations seem less and less committed to the ideal of education as a public good for all. What’s your take on all this, and how should we be arguing back?

PH:We should be arguing back in every way available to us. Sometimes people argue by appeal to the future earning power of graduates or by appeal to the effect of education on GDP. Education does have these effects, and I am not against using them in arguments.

But I do think that these effects on income and the economy stand to education as the effects on sexual attractiveness stand to athletic fitness: an inevitable by-product, but not the primary human good at issue.

Education can be usefully compared to health care and to national defense. No one thinks that we should keep people healthy primarily because healthy people earn more money (though I expect they do). And no one thinks that the primary purpose of national defense is to boost the economy (though defense spending can have this effect). Rather, public spending on defense is meant to protect a free society. Spending on health care is meant to preserve human dignity. Education takes its place alongside these—it is also about freedom and dignity.

Richard Rorty had a picture of education I find attractive: Education must first stamp out citizens. This “primary” education inculcates the skills and knowledge students need to take their station in society as we have made it. Further education—“higher” education—must then undo this. It should provides students with the experiences, skills, knowledge, and perspective needed to step back from the ambient culture, to gain critical distance from the forces that have shaped them, and to consider, for themselves, what to embrace, what to reject, and what to improve.

So, the goal of higher education is a free citizen, rather than a skilled laborer. It aims to create graduates who can tolerate ambiguity and uncertainty; who are made curious, rather than defensive, by difference or foreignness; who can contribute thoughtfully and intelligently to civil discussion on contentious topics. When it goes well, it creates idealists who are not extremists, people who can creatively transform the world, for the better.

I was in China a couple of years ago, and I noticed that there was an office in the Philosophy department for the Party Secretary. We have lost sight of what oppressive regimes around the world seem to recognize: ideas matter. They are, often enough, a threat that requires response. One response is to forcefully restrict their flow. Another is to train citizens to handle them well—to traffic in ideas, civilly and judiciously.

3:AM:Finally, for the philosophically curious here at 3:AM, are there five books you could recommend (excluding your own of course which we’ll be dashing away to read straight after this) that will take us further into your world?

PH:It is hard to narrow to five! One needs to be Harry Frankfurt’s The Importance of What We Care about—especially the title essay. I find it hard to resist recommending Thomas Nagel’s Mortal Questions, whenever I am asked for a recommendation. For those who want more, I’d add Elizabeth Anscombe’s Intention, Richard Moran’sAuthority and Estrangement, and T. M. Scanlon’s What We Owe to Each Other.

You didn’t ask about articles, but I’ll throw in three. The first is only a few pages long: Gregory Kavka’s “The Toxin Puzzle”(written during the arms race). The other two are by P. F. Strawson: his justly famous “Freedom and Resentment” and his hardly-ever-cited “Social Morality and Individual Ideal.”

Richard Marshallis still biding his time.