The Epistemology of Holding Responsible?

Consider the following principle:

(HR) It is fair to hold an agent S morally responsible for φ only if you know that S is morally responsible for φ.

Although (HR) seems true to me (at the very least it's related to the very plausible principle which holds that you should assert p only if you know that p), I won't be able to offer a defense of it here.  But tentatively accepting it, here are some questions.

Most compatibilists and libertarians believe that it is sometimes appropriate for us to hold individual agents morally responsible for their actions.  If you're one of these compatibilists (as I am) or one of these libertarians who thinks that it's sometimes fair for us to hold individual agents responsible, then given the putative truth of (HR), it seems like we're committed to the claim that we sometimes know that individual agents are morally responsible.  But this seems like a problem because how could we know that any particular agent satisfied the conditions of moral responsibility at a particular time?

I suppose that if Frankfurt's requirement on freedom of the will is correct, then we could just ask the wrongdoer about whether or not her second-order desire issued in the first-order volition that produced her action.  But what if Fischer and Ravizza's account of moral responsibility is correct?  How could we ever know that S acted from her own moderately reasons responsive mechanism?  And it wouldn't help to study those neurological or psychological mechanisms that actually issue in our actions because it's unclear what that could tell us about the modal properties of such mechanisms that are relevant to moderate reasons responsiveness.  Worse still, what if moral responsibility requires that we posses agent causal powers?  How could we ever know that we possess such powers?  Do we simply know it through introspection?  Or as a result of divine revelation?  Are those of us who think that it is sometimes appropriate to hold individual agents responsible committed to one of these answers?  Or should we simply abandon (HR)?  Of course, if we give up on (HR), we'll need another principle to replace it because surely there is some connection between what we believe of another agent and whether or not it is appropriate for us to hold that agent responsible.

I suppose a final option is that we radically alter our practices, not because we're not morally responsible (as the hard incompatibilist supposes), but because none of us is in the right sort of epistemic position to hold others morally responsible.  This seems to me (at least initially) to be a bad result, but maybe it wouldn't be so bad; I don't know, but I'd appreciate any help in sorting out the flurry of questions that seem to have emerged from this way of thinking about these issues.

Comments

[tnadelhoffer]

Justin,

First, this seems to me to be a good result! Second, I think you ought to modify HR in the following way:

(HR*) It is fair to hold an agent S morally responsible for φ only if you believe beyond a reasonable doubt that S is morally responsible for φ.

By my lights, you don't want to set the hurdle so high that it can't, in principle, be cleared. Of course, being the skeptical type, I don't think we can ever satisfy HR* either. But at least formulating the principle this way comports with how we determine the appropriateness/fairness of holding people criminal responsibility.

FWIW: This is an issue Kant addresses in his writings on retributivism. In short, Kant was a retributivist who nevertheless thought that punishment ought not be motivated by retributive hatred. On his view--espoused, for instance, in Religion within the Limits of Reason Alone--only God is in the epistemological position to "know the heart" of a moral agent. Here is an extended passage that seems relevant:

"We call man evil, however, not because he performs actions that are evil (contrary to law) but because these actions are of such a nature that we may infer from them the presence in him of evil maxims. In and through experience we can observe actions contrary to law, and we can observe (at least in ourselves) that they are performed in the consciousness that they are unlawful; but a man's maxims, sometimes even his own, are not thus observable; consequently the judgment that the agent is an evil man cannot be made with certainty if grounded on experience."

Of course, this doesn't prevent Kant from nevertheless trying to motivate retributivism. As he says:

"Because no punishment, no matter from whom it comes, may be infliced out of hatred. Hence, men have a duty to cultivate a conciliatory spirit. But this must not be confused with placid toleration of injuries, renunciation of the rigorous means for preventing the recurrence of injuries by other men."

I, for one, see Kant falling back on forward-looking considerations here. But that is a story for another day. For now, I just wanted to point you in Kant's direction since he seemed to raise similar concerns and yet he nevertheless went on to insist upon the fairness of punishment.

[Eddy Nahmias]

Hi Justin, nice post. If we take MR to be a graded concept and people can deserve praise, blame, punishment, and reward to varying degrees depending on the severity of the actions, then we might want to slip some of the degree talk into the epistemic condition you are right to point out is important in these discussions. Maybe something like:

(HR**) It is fair to hold an agent S morally responsible for φ *to the degree that* you are justified in (or can obtain justification for?) believing that S is morally responsible for φ ... in (some complex!) combination with the degree to which holding S morally responsible *mistakenly* would have bad consequences.

So, it probably would be OK to praise and even reward people for certain generous acts even if we are not that certain they are MR for doing them. With children, in some cases, we might think the level of certainty need not be very high since we might be justified in believing that holding them responsible for good and bad acts that they aren't really MR for yet can actually *help* them develop the capacities to be MR. But with criminals we plan to punish severely we may need to invoke a "beyond a reasonable doubt" level of justification, just as we do (in theory if not practice).

I think it's Randy Clarke who has a nice argument somewhere suggesting that if one really thinks agent causation is necessary for MR, then it's unlikely we can ever be justified in holding people MR, since we simply cannot be justified in believing people are agent causes.

[Robert Bauer]

Justin,

If i understand your principle HR correctly, you imply that there is a difference between knowing that S is responsible and holding S responsible. I believe you think about a difference between knowing that S is responsible and inferring that some action A is mandatory following this knowledge (which might e.g. be punishment).

Of course, whether this implication is justified is a purely ethical question and can not be answered by epistemological reflections. What seems to me to be the epistemological question: Can the condition of "knowing that S is morally responsible" be fulfilled?

This is to me as a social scientist a common problem of many philosophical terms. Many conditions which need to be fulfilled in ethical norms of the kind "if condition C is met, Action A is mandatory" are not empirical variables like age or a simple answer to an item in a questionnaire. They are "indicators", meaning they are constructed out of several empirical variables following spelled out rules of construction. An usual example for an indicator is the gross national production (GNP). Because "moral responsibility" is an indicator, the question is: are the rules for constructing it out of empirical variables satisfying? This is to me not an epistemological question, but an ethical theoretical one.

What i read from your post is that you did not want such an answer. In a sense, you might have wanted to know: What is the epistemological status of indicators (here moral responsiblity)? And: If the status of indicators is different from empirical variables, does this have implications for the mandatory action part of the ethical norm?

My answer: I do not know whether indicators have a different epistemological status than empirical variables. Empirical variables themselves are constructed following certain rules. Albeit, these rules are different than those used for construction of indicators. But even if one would argue that only empirical variables can be accepted to make an action A mandatory, it still holds true that every indicator can be disassembled into its core empirical variables. Whether the method of construction of indicators is valid is therefore to me a theoretical (maybe ontological) question, while the construction of empirical variables is to me an epistemological problem. This makes the question - whether someone is morally responsible - to be solved from ethical theory (or maybe ontology) and not by epistemology.

[Justin Coates]

Thomas and Eddy,

Because it came up in each of your replies, let me say something in defense of (HR). First, I just think it's an instantiation of Williamson's "knowledge is the norm of assertion" principle which holds:

(KA) S should assert that p only if S knows that p.

Since I find (KA) plausible and I think we chiefly hold others responsible through acts of communication (ranging from mild rebukes to harsh punishment), holding an agent responsible for her actions involves asserting something about the agent in question. Hence, (HR).

That said, Thomas, I find (HR*) plausible, but as you point out, I think it's probably subject to the same sort of argument. And you're right, it's not obvious that it would be bad if we had to alter our practices. The picture of the world that Gary Watson paints in his concluding remarks to "Responsibility and the Limits of Evil" in which we prune the negative reactive attitudes (and our expressions of such attitudes) doesn't look so bad.

Eddy, I'll have to think about something like (HR**) more, but my first response is to think that the second clause--while almost certainly part of some norm that governs the way in which we hold one another responsible--is not an epistemic norm. I'm glad you included something about degrees of moral responsibility in your response though. Perhaps, the way in which we hold others responsible is the sort of thing that can take us from relatively low stakes (e.g. if you were to rebuke an unruly student) to very high stakes (e.g. we're executing the terrorist). As the context shifts, what's required to know would shift as well. So maybe being a contextualist would help with this problem.

You're also right, I think, that the epistemic bar for praise is a lot lower than the bar for negative sorts of reactions (e.g. feeling resentment, punching someone in the face, throwing someone in jail, etc.).

Robert,

I don't want to say that knowing that S is morally responsible requires that we treat S in any particular way. I only want to say that knowing that S is morally responsible is a necessary condition on holding S responsible.

Moreover, I think you're right when you say that an agent's being morally responsible is "to be solved from ethical theory (or maybe ontology) and not by epistemology." But I'm interested in when it's okay to hold an agent morally responsible (which may not be coextensive with when agents are morally responsible). And surely facts about our epistemic standing are relevant to when holding others responsible is appropriate.

[Grant Rozeboom]

Justin:

Perhaps another way to avoid a "practice-changing" conclusion is to consider the possibility of substantiating the justification component of HR** with something like this:

(1) I'm morally responsible for my actions iff I'm in a (mental?) _position_ to entertain and respond to competing reasons and purposes for my actions.

(2) Someone else could make a justified judgment of my being morally responsible iff they assume my perspective and decide that they, based on their knowledge of the relevant factors of my condition (am I insane? am I subject to coercion?), would find themselves in a position amenable to the entertaining of and responding to reasons.

This goes a step beyond mere empirical inference and suggests something like an application of the second formulation of Kant's categorical imperative. By first being aware of those factors of someone's condition that are relevant to their agential standing, and by then imagining oneself as subject to those same constraints, one can fairly conclude (albeit imperfectly) whether one's neighbor is properly able to be reason-responsive and responsible for what they are doing.

As Eddy mentioned, if you wish to cling to an agent-causal picture of responsibility, this proposal will not do much good. It also means, among other things, that individuals who are entirely unfamiliar with the ways of life of those whom they are judging to be responsible/not responsible are not justified in making these attributions, for they cannot properly know the agency-relevant factors necessary for assuming the perspectives of those on whom they are passing judgment.

[John Turri]

Justin,

What do you mean by "holding responsible"? Is it just believing the person is responsible? Asserting that the person is responsible? Something else?

[Justin Coates]

John,

What do you mean by "holding responsible"?

I'm not exactly sure yet (can I get back to you in a couple of years?), but in my mind, the paradigm cases of "holding responsible" are expressing reactive attitudes, rebuking, chastising, sanctioning, and punishing. Of course, Wallace, who's done more to develop a notion of "holding responsible" thinks that the paradigm case is unexpressed reactive attitudes. Scanlon thinks (or at least, used to think--I'm still working through his new book) that it amounts to judgments of faulty self-governance. If Wallace or Scanlon is correct, then this argument won't go through, but if "holding responsible" tracks a plurality of behaviors--many of which involve actions towards other agents--then I'm inclined to think that something like (HR) is true of those ways in which we hold one another responsible that go beyond mere judgments or unexpressed reactive attitudes.

Grant,

That's interesting (although I'm worried when two commentators point me towards Kant!), but I'm not sure how being able to assume the other's perspective will tell me whether or not the other person satisfies _insert favorite sufficient condition(s) on responsibility here_. So maybe you could say more.

[John Turri]

Justin,

Thanks. And sorry to put you on the spot -- I know it's a tough question.

In any case, I wanted to mention that there might be another option for you here.

Earlier in the thread you mentioned the knowledge account of assertion (KAA). Many people who find that plausible also find it plausible that belief is the inner analogue of assertion, and that knowledge is the norm of belief: (KAB) you should: believe Q only if you know Q. (See e.g. Knowledge and Its Limits p. 255 ff.)

Suppose that KAB is true. Now, even if "holding responsible" involves nothing other than unexpressed reactive attitudes, so long as these attitudes require believing (even dispositionally) that the person is responsible, then something like HR would follow.

I say "something like" because HR involves fairness, which is a moral notion. KAA and KAB involve purely epistemic senses of 'ought' or 'must'.

[John Turri]

I said 'ought' or 'must', but in my formulation of KAB here I said 'should'. Same difference.

[Grant Rozeboom]

Justin,

Perhaps I should make clear that, first of all, I concede that there is probably no such thing like a "direct" or "certain" knowledge of another agent's fitness (or lack thereof) to be held responsible.

What I was suggesting was that, given a certain familiarity with an individual's context and circumstances at a given time, one can put oneself in that individual's shoes and make a reasonable (or justified) judgment of whether or not that individual "satisfies" the prescribed (likely compatibilist) conditions for being held responsible. (As I said above, agent-causation theories would probably have to be excluded, as I don't see how a first-person perspective sheds light on whether or not one is an agent-cause).

Consider Agent P, whom I observe carrying a can of gasoline out to a prairie, carefully sprinkling the gasoline at strategic locations, and setting fire to the grass. Were I, based upon this brief observation, to transfer myself "into his shoes," as it were, I would find myself to be someone setting a prairie on fire in accordance with purposes that I've set before myself and -- speaking loosely -- therefore I could judge that Phil is fit to be held responsible. But if this were all I knew of Phil, this would probably not be a very justified attribution of responsibility. Perhaps, unbeknownst to me, there is a gunman hiding behind a tree who has threatened to kill Phil if Phil doesn't set the prairie on fire. Were I to take Phil's perspective with this added _relevant_ bit of knowledge, my judgment of his culpability likely would change. I, in Phil's shoes, would no longer think myself to be in a position of "reason-responsiveness" (or whatever other conditions you might wish to insert to indicate the appropriate level of control), and so I would, with more justification, decide that Phil is not fully responsible for what he is doing and thus that I should not hold him fully responsible for torching the prairie. Before I can "know" that Phil is responsible and so hold him responsible (assuming the relationship you posited previously), I must (1) make an attempt to take into account all of the factors related to Phil's standing as an agent at the time of the actions for which he is allegedly responsible and then (2) be able to imagine myself as an agent facing the same circumstances. If, in this agential "role-play," I find that I am in a position to be reason-responsive, in guidance control, etc., then I can reasonably assert that Phil is responsible, and act towards him accordingly.

I suppose this may not do much more than the mere extrapolation from observing a person's actions and guessing whether or not they signify responsibility-fitness that was discussed in an earlier post. But, as far as I can tell, it adds as conditions for knowledge-of-another's-responsibility 1) the awareness of the relevant contextual factors to their agency, thereby granting you ...
2) the ability to relate to that person such that you can see -- to a signficant degree of clarity -- whether they were/are in a position to be reason-responsive, in guidance control, etc. I suppose there is always the possibility that there are hidden coercive factors (e.g. an implanted chip in their brain that spontaneously modifies their behavior) at play that cannot be taken into account in this game of responsibility role-play, but I don't think that detracts from the formulation of reasonable or justified attributions of responsibility-fitness. I also think that, given HR*, this points towards, if not epistemically sound, at least _fair_ judgments of responsibility-fitness.

Phew. That was more verbose than I intended. Perhaps it helps illuminate my original point, though?

[Grant Rozeboom]

I should summarize the principle of fairness I have in mind here as follows:

If Agent A, having as full of an awareness as possible of the relevant contextual factors affecting Agent B's agency at time T1, assumes the perspective B had at Time T1 and decides that she -- as B at T1 -- is in a position to be ... (reason-responsive, etc.) such that she -- as B at T1 -- is responsible for what she is doing, then A is justified (or, it is fair for A...) to hold B responsible for what B is doing.

This is what I mean by saying that this account of judging someone to be responsible via role play points to "if not epistemically sound, at least fair judgments of responsibility-fitness."

[Robert Bauer]

Justin,

If i understand you correctly, you ask whether it is correct that "knowing that S is morally responsible for φ is a necessary, albeit not sufficient precondition for holding an agent S morally responsible" I think your right with this inference. Of course, you have to have a concept of MR, which can only be given by ethical theory. Let´s assume it is given.

The conflictous point seems then to be: when do we know that S is morally responsible for φ. I am currently in an intense discussion with one of my colleagues at my institute. I believe that this is not a question to be solved by epistemology alone, but that the ontology of moral responsibility has to be discussed in addition, while he thinks epistemology is sufficient. Let me shortly state my string of argumentation.

If you believe that a certain epistemology is not sufficient to tell us whether the condition C for moral responsiblity MR is met you have two options: change your epistemology or your concept of MR to fit each other or make ontological assumptions, like that behind the phenomena we perceive, there exist the conditions C we can not perceive but that fulfill the condition. One example for such an insufficiency of epistemology to give clear answers whether S fulfills C for MR is Kant.

Even if you believe that a certain epistemology is sufficient to tell you whether S meets condition C for MR, you have to make sure that this conditions are truly met. Epistemology can only assert truth if you make sure that behind the phenomena there is not something that undermines your knowledge that conditions C are met. Again you have to discuss the ontology of MR.

Of course, what i have stated here is not an answer to your question, but merely an attempt to understand the framework for answering your truly interesting question.

[tnadelhoffer]

Grant,

I am very leery of your "agential role playing" response to Justin's worry. Michael Moore has developed a similar line in "The Moral Worth of Retribution" whereby we imagine ourselves to be the perpetrator(s) in question in an effort to determine whether they deserve punishment. For a criticism of his approach, see Dudley Knowles "Unjustified Retribution."

For now, a quick example might help. Take, for instance, Gary Watson's famous use of Robert Harris as a means for both stoking and allaying the retributivist emotions. Imagine, for a moment, that you are Robert Harris--i.e., you were (a) born premature when your father kicked your mother repeatedly in the stomach, (b) regularly neglected and seriously and serially abused throughout childhood, (c) repeatedly sexually assaulted while in juvenile detention centers, (d) a torturer of animals as a young adult (a tell-tale sign of psychopathy), etc. Now, imagine that you (a) unnecessarily murder two innocent sixteen year old boys so you can steal their car to rob a bank, (b) you laugh out loud that one of the boy's brain matter ended up on the barrel of your gun, (c) you callously eat the victim's MacDonald's hamburgers 20 minutes later while making fun of the unspeakably horrendous actions you just performed.

How are you doing so far in slipping into Robert's shoes? If you're like me, at least, you're doing a piss poor job. I have no idea how to perform this kind of counter-factual shoe swapping. As such, I have no idea whether Robert was reasons' responsive in the right sort of way. And even if I did judge that he was morally responsible on some level, I have no clue how to figure out what degree of responsibility he has.

[Sofia Jeppsson]

Thomas, I'd definetily have a hard time imagining myself in Robert Harris' shoes. But that doesn't have to be an argument against the "agential role-playing" account of moral responsibility.

I might have done something that actually goes against my principles, or something that I on reflection think was bad, and then I regret it. In this I aim some kind of retributivist feeling towards myself, because I think it's proper that I should feel bad when thinking about my action. This might of course have the effect on me that I try to act differently in the future, but that doesn't mean that what I did when I was regretting my action was some kind of behaviourist training of myself. Neither does it necessarily mean that there exists some forward-looking justification for my regret, a justification which is somehow there independently of what I think when I regret.

Now suppose some other person who is similar enough to myself (of course I don't know exactly where to draw the line of "similar enough"), so that I can imagine myself in that person's shoes, does something that's bad in my eyes. I could have reactive attitudes towards this other person and blame her for doing a bad thing, just like I could myself. Even if she doesn't have any principles herself that she broke by acting as she did, I might think that she should have thought as far as I myself have done on the matter, she's smart enough, she could have realized that it was a bad thing to do and then shown some resolve and avoided it.

But the Robert Harris case? well, on REFLECTION (as opposed to completely spontaneous reactions one might have) I don't think one can be justified in holding retributive views when it comes to agents who are too different from oneself. I think (warning: empirical psychological speculation to follow) that when we spontaneously want psychopathic criminals to suffer for what they did, we somehow imagine that they will come to realize themselves that their suffering is a "payment" for what they did, that they will somehow internalize the retributivism, so it becomes a large-scale counterpart to what I may feel towards myself when I've done some more everyday kind of wrong. But if it can't work that way, because the person to be punished will never internalize, I think the punishment becomes misguided.
I think there are only two proper considerations when it comes to the wrong-doing of agents very different from ourselves: 1. How can we prevent that he commits further wrongs that hurt others? 2. Is there any way to make him into more of a moral agent, so that he himself starts to recognize that he did wrong? I suppose 2 is possible to fulfill (through empathy-promoting therapy, say)in some cases but not in others, and if 2 isn't possible only 1 remains.

[tnadelhoffer]

Sofia,

You say:

"[O]f course I don't know exactly where to draw the line of 'similar enough'."

But that is precisely why I take the problem raised by Justin to be a good one. Absent some way of knowing what is going to count as "similar enough," it is unclear how you could know in the relevant way that it would be fair to hold the person morally responsible.

And that is to say nothing of the fact that I am even skeptical of whether you have the adequate introspective access to know all of the relevant facts about yourself when it comes to your own culpability. Bargh's work on automaticity, Batson's work on situationism, and Nisbett's work on introspection come to mind here. Imagine, for instance, that Harris himself relies on introspection to try to figure out his own culpability. How reliable ought we expect his self-reflection to be here?

Consider, for instance, that battered women often blame themselves, that victims of kidnapping often come to sympathize with their captors, that abused children often come to believe what their parents tell them about how worthless they are, etc. How good are they going to be at figuring out there own culpability? And if they couldn't reasonably expected to be very good at it, why think you would be any better at it by imagining yourself in their shoes?

Now you might simply respond that these are the kinds of cases that you allow are ones such that we can't know/judge their culpability. But these are ofter precisely the sorts of cases juries are asked to examine in criminal trials.

[Grant Rozeboom]

Thomas-

It seems that your objection to the agential-role-play account of knowing that someone is responsible hinges on a skeptical argument of sorts. I suggested that, given a significant level of familiarity with an individual's circumstances, one can adequately imagine oneself in that individual's shoes such as to evaluate whether or not one would hold oneself responsible for one's actions. If yes, then it is justified/fair to hold the individual responsible; if no, then not. Your main objections seems to be something like this: How can we know what is a "reasonable" level of familiarity, especially since we find it hard to identify with persons such as psychopathic criminals, the culpability of whom we are supposed to evaluate? I think my first response is this: What prevents us from having the ability to perform the role-play to a degree sufficient for making a judgment about culpability? In the Harris example, for instance, why conclude that we cannot imagine ourselves in his shoes when, I think, the proper claim that follows is that we cannot imagine ourselves fully responsible for the horrendous acts he performed when assuming his perspective. Certainly, we are uncomfortable just letting him off and allowing him unlimited movement in society (as Sofia suggested). But does this necessarily entail a blow to our ability to evaluate his responsibility-fitness through agential role-play? I don't think so.

Secondly, with regards to your justified worries about introspection, I'm not sure I see the implications for role-play evaluations of responsibility. Why does an assault victim's inability to accurately reflect on the awful trauma dealt to them inhibit my ability, as a third party, to imagine myself in their shoes and conclude that they are not responsible for being assualted? This is what is it at stake, is it not? The same goes for Harris: Just because he, in a disturbed frame of mind, would not claim responsibility (or lack thereof) for his crimes does not mean that I cannot make a justified evaluation of his agential standing on the basis of my knowledge of the relevant factors influencing his criminal actions. Putting ourselves in someone's shoes in order to form a judgment about their culpability for a given set of actions does not necesssarily mean taking on their own current backwards-looking attitudes, feelings, and (possibly distorted) thoughts about these past actions, no? Faulty introspection is a curious problem of its own, but I don't think it entails problems for agential-role-play, at least the sort I was describing above.

I still admit that there are unfortunately hazy distinctions to be drawn between a sufficient and insufficient levels of familiarity/similarity, but I guess I don't see this as a skeptical sinker.

[Mark Smeltzer]

Thomas,

"I, for one, see Kant falling back on forward-looking considerations here."

No surprise there. I, for one, see Kant's deontology as being built upon forward-looking considerations. If it weren't it would be utterly dogmatic, but Kant really wants it to be completely rationale.

I see Kant's deontology as a sort of rule-utilitarianism combined with what are taken to be quintessential facts about human nature (hence the allure to assert maxims a priori), but that is a story for another day...

Justin,

I support the following principle, and have stated it in similar terms previously on this blog:

(MR) Agent S deserves to be treated by agent O as O ought to treat an φ-er if O has warranted belief(s) that φ at time T entails that S is an φ-er.

Translated to a form similar to HR, it would read:

(HR**) It is fair for observer O to hold agent S morally responsible for φ if O has warranted belief(s) that S is morally responsible for φ.

This stating of the principle is relatively similar to Thomas' (HR*). Moreover, in (MR) and (HR**), when O's warranted beliefs are true, they count as knowledge. (HR) is too strong as stated because it has an implicit infinite regress: it requires knowing that we know.

I don't find it surprising that this is more or less how our legal system is setup either. I also don't find it surprising that Kant reached more or less the same conclusion.

Lastly,

Others in this thread have commented about principles that involve "wearing S's shoes". I think the "relativity" being discussed here comes out in this part of (MR): as O ought to treat an φ-er. If O is a judge in the legal system, the treatment would be different than if O is S's mother, than if S and O are both φ-ers. (MR) nicely captures the objectivity of responsibility, while allowing for a degree of relativity (not quite the same thing as relativism) about the deserved treatment by specific persons.

[Mark Smeltzer]

Btw, labeling the principle in my previous post as (MR) may be misleading. It is a principle about warranted treatment, not about moral responsibility. It would have made more sense to label it as (WT)... in fact, it should have been stated this way (I am happy with (HR**) as is):

(WD) Agent O is warranted to treat agent S an φ-er if O has warranted belief(s) that φ at time T entails that S is an φ-er.

A corollary of that principle would be a principle about objective desert:

(D) Agent S deserves to be treated at time T according to S' character at time T.

The principle for moral responsibility that I support is:

(M.R.) Agent S is morally responsible for φ at T iff there could exist an agent O that would form true warranted beliefs about S' character in virtue of observing the relevant casual chain(s) that rendered φ true at T.

The difference between (HR**) and (M.R.) is equivalent to the difference between (HR) and (HR**)... it allows for the subjectivity of holding people morally responsible while maintaining the objectivity of people being morally responsible.

[Ben Vilhauer]

Justin,

Great post. I think Eddy is right that that this is a matter of gradation, and that Thomas is right that reasonable doubt is important. It's only just to blame somebody in a way that involves doing serious harm to him if it can't be reasonably doubted that he was morally responsible. But if we're talking about truly harmless praise, i.e. praise that can be given without any negative consequences at all, not even hurting the feelings of third parties excluded from praise, then I think we can legitimately praise somebody so long as it can be reasonably doubted that he was not morally responsible. I also think that anybody who takes the debate about whether we have free will to be truth-conducive has to acknowledge that there is room for reasonable doubt in both these cases. I argue for these claims in some recent papers, one which is forthcoming in APQ entitled "Free Will and Reasonable Doubt", and another, as yet unpublished, which is entitled "Free Will and the Asymmetrical Justifiability of Holding Morally Responsible". If you're interested, drafts can be found on my website (www.benvilhauer.com)