Blog Coordinator

« A Question for Contributors | Main | Weakness and Strength of Will »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Great questions Carolina. You have a gift for taking what have been platitudes in our field and turning them into quandaries.

One asymmetry that might be relevant here is truth versus falsity or error. When we teach logic we stress right routes of thinking--validity, soundness, probabilistic credence, etc. But when we teach fallacies we stress while there maybe are (maybe few) ways to be right, there are also almost infinite ways to be wrong. "Non-x" as a category generally characteristic as contrary to some stipulated "x" is as rich as the set of non-red colors constituted by all others that are not red.

So I take it that you mean that right actions have narrow sets of right reasons though wrong actions range across huge (maybe open-ended) sets of wrong reasons, and that is a purely logical point. That seems correct to me. But blaming wrong actions does require that there is some kind of *access* to right reasons as well as that large set of wrong reasons, and that is a metaphysical point that (it seems to me) must involve counterfactuals or the like. And that would appear to involve claims about abilities of who is accused of wrongdoing, and might go beyond the actual sequence that is in question. Hope that's clear.

A short follow up (forgive me for going on). In terms of set-sizes of reasons to refrain from immoral actions versus moral ones, the logical asymmetry I alluded to above holds put in those terms, and I should have said something about that, since that was your point. Then are you saying that smaller sets of refraining reasons favor the actual sequence in moral instances of behavior than larger ones in immoral behavior? So moral behavior is more (simply) understandable than immoral behavior in terms of an actual act-type? That point would avoid the access-to-alternatives for immoral action at least that I suggested, though it would be (I think) a purely epistemological asymmetry.

I'm sorry Carolina--probably thinking out loud too much!

Hi, Carolina.

Another cool post. I’ll try not to harangue at you so much this round!

I was just wondering how you conceive of the class of reasons against some course of action? I do see why it’s tempting to think that some actions, like drinking poison will have more reasons against than others, like drinking water. But I wonder whether this is like comparing two infinite sets, where one set can still be larger than the other. Since I think reasons come cheap, it is really easy to get reasons for some course of action (whether doing or not doing something), so long as well allow that it could be a reason of trivial weight. (I think Mark Schroeder illustrates this nicely in the Negative Existential Reason Fallacy.)

Of course, if one has a more robust view of what it takes to have a reason, then maybe one doesn’t face this proliferation of reasons. But on a simple view of reasons, we get lots of reasons to do stuff very quickly. It’s just that, ordinarily, only a smaller subset of those reasons have sufficient weight to be taken seriously. So I wonder, I guess, whether we don’t need some weight restriction here, before we can compare sheer numbers.

For example, I would think it true that actions for which one is blameworthy for have weightier reasons against than those one is praiseworthy for. And, the latter have weightier reasons for performing than the former. And on some views, this weightiness may be supplied by moral reasons, which the view takes to be so weighty so as to make the number of non-moral reasons irrelevant for deliberation. So weight would be more significant than quantity.

Hi Alan,
Thanks for your comments!

My thought was this. Compare a paradigmatic example of a bad act that someone might be blameworthy for (say, killing someone in cold blood) with a paradigmatic example of a good act that someone might be praiseworthy for (say, helping someone in need). According to an RS view, the ranges of reasons we need to compare involve (sufficient) reasons to refrain from performing those acts. In the killing case, there are actual moral reasons of that kind that the agent is not responding to, but RS views claim that the agent can still be blameworthy to the extent that there is a sufficiently large range of counterfactual reasons that he is responsive to. Those will typically include several moral reasons (for example, it could be that the agent would have failed to shoot his victim if he had reason to believe that his victim had five children; and other examples like that—you get the idea). In the helping case, there is no actual (sufficient) moral reason to refrain from helping the person in need, but there could still be some counterfactual moral reasons that the agent is responsive to (say, if he had reason to believe that the person is a serial killer looking for his next victim, then he wouldn’t help him). My thought was simply, given that it’s much harder to come up with (sufficient) moral reasons to refrain from performing a paradigmatically good act than it is to come up with similar reasons to refrain from performing a paradigmatically bad act, the range of potentially sufficient moral reasons to refrain is much larger for blameworthy acts. Thus Flexibility would entail that the range of reasons the agent needs to be responsive to is larger for paradigmatically bad acts than for paradigmatically good acts (at least as far as moral reasons are concerned).

On your worry about how all of this is consistent with an actual-sequence approach: you’re right to be worried! I think it’s tricky to insert an RS approach into an actual-sequence view, and that one needs to be careful here, especially if one wants to preserve the claim that freedom supervenes in actual sequences. My main strategy (that I didn’t talk about in the post) is to suggest that those counterfactual reasons the agent is responsive to in fact enter the actual sequence in the form of causally efficacious absences. For example, the blameworthy killer is responsive to those counterfactual reasons to refrain from killing his victim because the absence of those reasons causally results in his act (e.g., his not having reason to believe that his victim had five children was one of the actual causes of his killing his victim). So I end up endorsing an actual-sequence based notion of responsiveness to reasons, not one that is essentially based in counterfactuals. (I know this may sound crazy, but that’s my view!)
Also, recall that I believe that epistemic states (such as moral beliefs) can be relevant to responsibility without entering in the causal sequence (all I want to claim is that the freedom component of responsibility supervenes on actual sequences; the epistemic component may not--this is something that came up in the first post).

Thanks, these are good questions.

On the issue of weights of reasons: RS views typically deal with *sufficient* reasons to do otherwise (or sufficient reasons that agents have to do otherwise). The idea is to focus on those reasons that would have (at least in normal circumstances) resulted in the agent’s doing otherwise, as a result of having recognized those reasons as sufficient reasons. (Sorry, I should have been clearer about this in my post.) This limits the ranges of reasons somewhat (in particular, by eliminating “trivial” reasons or reasons that don’t have much weight).

But, even then, the problem you note remains. I realize that there are some ways of counting reasons that would result in an infinite number of reasons on both sides, and this creates obvious problems. I guess all I was trying to suggest is that there seems to be some natural way of counting them according to which there’s more on one side than in the other. To the extent that we see this in the A1/A2 pair, we should also see it in the contrast between good and bad acts, and this would explain why we perceive a certain asymmetry between blameworthiness and praiseworthiness.

By the way, all this reminds me of Jonathan Bennett’s claim that the difference between doing and allowing is a difference in the number of ways one could have moved that would have resulted in a given outcome… He had a sophisticated way of trying to count ways of moving that, he thought, successfully deals with the infinity problem. The main strategy seemed promising. If I recall correctly, it was something like trying to lump together potentially infinite ways of moving into a finite number of kinds. I guess one could try to do something similar here. But I certainly don’t have a definite proposal to make.


I find myself agreeing with all your main points, but then perhaps we go in slightly different directions in working out the defenses and details. So: I agree that we can remain actual-sequence theorists and still try to acknowledge *some* asymmetry of the sort Wolf and Nelkin (and Gideon Yaffe and John Locke) advert to. I'm not sure however that in the end we'll want to analyze the asymmetry in terms of a purely *quantitative* notion, just as I'm skeptical of Bennett's similar suggestion (about acts/omissions). But then again I feel the pull.

I have never thought the asymmetry thesis in question is plausible, because (surprise, surprise) I'm convinced by the FSCs (and other considerations) that freedom to do otherwise is not required for moral responsibility. And these ideas apply symmetrically to good and bad actions (as Ravizza and I argue in an Ethics piece on Wolf and also our book). Of course, this leaves it open that there is some *other* reason (in the neighborhood) why praise and blame are interestingly different.

The following has struck me for a long time: sometimes one has a simple, basic idea (not necessarily original), such as the insight that moral responsibility is based on the actual sequence and doesn't require alternative possibilities. This idea goes back to Chrysippus and Locke, and goes through Nozick and Frankfurt to the present. Also, there is the rationalist strand, via Spinoza and presumably others. The idea is kind of a basic and intuitive and natural one. But when one tries to explain and defend it, it gets tricky. I appreciate your efforts to do so! I guess I think that even if certain attempts to explain/defend the idea are problematic, the idea is still compelling!

Thanks so much Carolina for your detailed reply--that helped me get your point of view on the OP immeasurably. Very interesting and ingenious!

Here's a question then. Are the causally efficacious absences always necessary conditions for the state of affairs to have been otherwise? (i) Simple case: a match was struck, but in a vacuum. So the lack of oxygen caused the match to fail to light, and had oxygen been available (all other things equal) the match would have lighted. So (i) is an actual sequence cause by absence. But are causally efficacious absences sufficient to produce effects? (ii) More complex case: the match is struck under all-equal circumstances but a water-hose posed above the match is not turned on. The match lights, but only because the water-hose is not turned on. The water from the hose would have been sufficient to extinguish the match. Is the water's absence a necessary or sufficient condition for the match's lighting? Given its actual presence one could argue both I think, but given that the all-things-equal holds in this case it is sufficient for the match's lighting. Your more complex case: the killer lacks a belief that the victim has 5 children and kills the victim and that lack of belief leads causally to the murder. (So I understand.) That seems to track (i) rather than (ii) because there is no actual belief poised to intervene as there is a faucet that can be turned on. So the question is whether the killer's actual beliefs are sufficient to murder (they are), and whether they could have been stopped by the 5-child belief as a sufficient condition for refraining. But that requires a specific psychological account of the murderer including the 5-child belief, and that seems counterfactual to the actual situation, where he lacks that belief (unless we can say he has access to that belief, but that seems to involve counterfactuals). So saying the non-belief is as sufficient a condition as as the non-turned-on faucet case is not warranted. And that does not favor the actual sequence case.

Does that make sense? Thanks so much for the challenging post.

Welcome back, and thanks for your comment!

I wonder if all RS views won’t have to deal with the “infinity threat” too, after all. There are ways of counting reasons according to which agents who we’d say are not free are responsive to an *infinite* number of refraining reasons. (Silly example: an addict might be disposed not to take the drug if we offer him 1 million dollars, or 2 million dollars, or 3 million dollars, or…) So we still need some way of dealing with this problem so as to able to claim, for example, that the addict is responsive to “fewer” reasons than the non-addict. I realize that the comparison is not only quantitative. But it does seem to be quantitative at least in part. Does this sound right to you?

On the blame/praise asymmetry: There are probably many interesting asymmetries between blame and praise. My (again, at this point purely speculative!) thought is that at least one of them has to do with the reasons associated with blameworthy and praiseworthy acts (reasons for performing them and for failing to perform them), one that results in praiseworthiness being less demanding than blameworthiness, and that this is what Wolf and Nelkin got right. On Flexibility, more generally, my thought is that it seems wrong to expect that all types of acts should require sensitivity to equally broad ranges of reasons. Why would that be? Since different acts are associated with different pools of reasons (to perform them and to fail to perform them), it seems quite natural to expect that certain important properties that acts will have that depend on the reasons associated with them will vary depending on the type of act in question.

Hi everyone! I said in my introductory post that I was going to post on some speculative issues I’m hoping to get feedback on, and (as you can probably tell) this is one of them. I really don’t have a fully worked out view on these issues yet. So I very much appreciate all your comments.

A quick clarification on Bennett’s view: I didn’t mean to suggest that Bennett’s view works as an account of what the distinction between doing and allowing harm amounts to. All I meant to say is that he’s probably right that there are plausible, quite natural ways of dealing with comparisons between two infinite quantities. (The best example of this is probably something like: there are more natural numbers than even numbers.) Bennett’s own example, I believe, was comparing pushing a car down a hill to its destruction with standing back, or failing to interpose a rock so that a car that’s already rolling down the hill won’t fall to its destruction. He claimed that, in some natural sense, there are more ways of standing back than of pushing the car, and that seems plausible to me.

I’m not totally sure I follow. My view on the killer case is that his responsiveness to reasons is reflected in the actual sequence because the absence of the belief that the victim had five children (whether or not he actually had five children) is one of the causes of his act (and the same is true of many other refraining reasons). He kills his victim not just because of certain reasons he has for killing him, but also because other reasons he would have for not killing him are absent.

I don’t believe that those absences must be *necessary* for the effect in the sense that they are only causally efficacious when the effect counterfactually depends on them. Imagine the similar killer in a Frankfurt scenario: the neuroscientist is disposed to intervene and force him to make the same choice if somehow he were to believe that his victim has 5 children. In that case too, I’d say that the actual sequence includes the absence of the belief that his victim had 5 children. That’s part of what *actually* accounts for his choice (together with other positive reasons for making the choice), although in the counterfactual scenario, of course, something else would have accounted for it (the neuroscientist’s intervention).

Again, I’m not sure I’m answering all your questions. I'm sorry if I’m not!

Carolina, I am finding your posts (and the discussions) very interesting -- tho' I'm not sure I can get my brain around all the issues. One thing that strikes me right now is what you mean by a "type of act". You wrote:

//Flexibility: One of the things that determine whether the relevant range of reasons is appropriately wide is the particular type of act in question.//

//These reasons are, in fact, absent but are such that, were they to be present, the non-addict would (in normal circumstances) respond to them.//

//A1: Ingesting the contents of a can of poison//
//A2: Drinking a glass of water//

I'm concerned about whether A1 and A2 really are different types of act. Rather it seems that there is an underlying type (ingesting an amount of some substance) and some of the reasons for not doing it are present in A1 but absent in A2. For example -- don't ingest something if it'd kill you to do so. If the water *would* kill you, then you wouldn't drink it. The non-belief that the water will kill you is as much a cause of your doing A2 as the killer's non-belief that the victim has five children is of his act.

And if you press the issue, you could argue that the drinking and the killing are both just instances of doing. "It would make Nelson Mandela sad" is a reason not to do pretty much anything (IMO -- but but it's defeasible, of course). That reason is absent for every action I have ever considered taking, but I'd be sensitive to it if it ever did arise!

I'm wondering whether you have or envision a principled way of delineating the types of action to allow your principle of flexibility to actually bite.

And also a mathematical note: the "quantity"/"number" of natural numbers and even numbers is the same -- Alef_0. The sense in which there are more natural numbers than (positive) even numbers is set inclusion. But if you include negative even numbers, you lose even that. I think you do want to concentrate on weight as a metaphor.

Thank you so much Carolina--your actual-sequence patience with me is certainly laudable! I think I get it, but still feel uncomfortable with absent reasons as part of an actual causal sequence, metaphysically and even as explanatory. I get your motivation for this, and it would go a long way toward justifying actual-sequence accounts against counterfactually-based criticism, and I'm favorably disposed toward the moral sufficiency of such accounts. Best now for me to shut up and listen for a while!

Hi Mark, and thanks for your comment!

I was thinking that, in the same way that two countable sets can be such that one is properly included in the other, one countable set of reasons could be properly included in another countable set of reasons, and in that case the proper inclusion criterion would naturally lead us to think that the more inclusive set is a “larger” range of reasons than the other. As an example, compare a non-addict who is sensitive to financial counterincentives of $1,000 or higher with one for whom the “magic number” is much lower, say, $10. There is a natural sense in which, I think, we’d say that the latter is sensitive to a larger range of reasons. (Right?)

On your question about what “act types” are the relevant ones: This is an important question. In fact, I tend to think that this is another underappreciated issue that arises for RS views. How about this as a rough answer? When we’re trying to determine whether an agent is sensitive to reasons in acting in a certain way, we must identify acts in a manner that preserves their significant normative features. Consider, again, the act of pushing a button that you know will prevent the torture of your infant child (one of those cases for which I said it’s at least plausible to believe that no sufficient reasons to refrain from performing them exist). What would happen if one were to describe the relevant act as simply “pushing a button”? Or “moving your hand (in a certain way)”? (The list can go on and on.) There are lots of potential sufficient reasons to refrain from performing those acts, when described in those ways. So should one say that the button pusher is sensitive to reasons because, when his act is described in those ways, there are many reasons whose absence he’s being sensitive to? This doesn’t seem right, because those are reasons to refrain from pushing some button, but not that particular button that one knows to have important properties, or reasons not to move your arm in some way, but not in that particular way that is especially significant in this context. (As a limiting case, as you point out, all acts can be described as “doing something”. That’s clearly not the type that RS-theorists want, otherwise an agent would be equally sensitive to reasons in performing what seem like different acts.)
So I guess the answer I’m tempted to give is that one must at least include in the description of the relevant “types” those features that seem normatively significant in the context (the button that (you suspect) will do good, the substance that (you predict) will kill you, etc.). But I’m thinking out loud here, I hope this makes sense!

//[W]e must identify acts in a manner that preserves their significant normative features.//

Sorry if I'm being slow, but I don't see how this can work. Let's go back to the (non-)addict. The addict and the non-adict both take the drug, but only the non-addict is responsible because only he is adequately sensitive to the reasons not to take the drug. But /per hypothesis/ the reasons for not taking the drug are absent for this particular drug-taking action. If we identify the normative features of *this particular drug-taking action* then it will be a different set of features than we would get for drug-taking actions where those features are present. Having a different set of normative features would imply being a different type of act, wouldn't it? If pushing the button when you know it'll save your child is different from pushing a button when you don't know that, then taking a drug when you know it's not going to do any harm to anyone should be different from taking a drug when you don't know that.

In order to get the same set of normative features, wouldn't we have to already have decided that it's the same type of action -- taking a drug simpliciter, say, or maybe just acting simpliciter -- so we can figure out which absent features are relevant? What other way is there?

My thought was simply that in order to identify the relevant type of act one would have to do this by using the relevant degree of specificity, where, arguably, it’s not going to be too general and also not too specific, but somewhere in between. For example, in the case of the addict and non-addict, it’s going to be something like taking “this type of drug”: a drug with certain salient properties (not just taking something, or taking any drug, which is too general, and also not taking the drug in a very specific way, or in a way that includes all the properties of the drug and the agent, which would be too specific). And I was thinking that we could then think of the addict and non-addict as performing the same act, in scenarios where certain reasons to refrain are absent and also in scenarios where they are present. (I realize now that when I used the example “the substance that (you predict) will kill you” this might have been misleading. There I wasn’t referring to the scenario where the addict takes the drug, perhaps predicting that it’ll kill him, but to the example of drinking poison.)
I was assuming (perhaps naively?) that there is some natural way of identifying the relevant type in each case. But these issues are always tricky, aren’t they? Do people have any thoughts on this?


I have been thinking more about the flexibility principle, and I think now that some (and possibly all) of what I said against it is wrong. I failed to properly consider the fact that it's an only-if principle, and not an if-and-only-if.

What that means is that there's no danger from taking a view of the action that's too high-level. The actor may be sufficiently RS when we take the view of actions in general, but that wouldn't mean that the actor *is* responsible, only that he *might be*. Thus my concern about A1 and A2 having a type in common was misplaced.

The danger remaining then would be from getting too specific. Here I still wonder about the how your principle will apply. Consider the button example again. Its type (T) is such that there is no reason not to push the button. But suppose that there is a snake in the room, and the actor has a paralyzing fear of snakes. Is the actor responsible for not pushing the button? Well, if you look at it as type T, then he might be. But if you look at it as "Doing T while there is a snake present" (T') then the principle says he is not responsible. Which is as it's generally expected to be, I suppose.

The question that raises, tho', is /Does your principle explain his non-responsibility, or have I misused it?/ There is no single type for his action -- that I have varied until I found one under which the answer comes out "not responsible". If that's a valid thing to do, then the principle doesn't seem to offer sufficient guidance for how wide the relevant range would be -- differing description could yield differing ranges. If it's not valid, then there's still the question of which of these descriptions is the proper one. But perhaps that's a problem that all RS theories have, and so not a valid reason to object to your principle in particular. That's something I'm afraid I can't evaluate on my own.

Hi Carolina: It seems right to say there are often more reasons to refrain from performing a morally bad act than there are reasons to refrain from performing a morally good act. And this all by itself may show that at least we can make sense of the asymmetry you propose between blameworthiness and praiseworthiness without committing ourselves to the ability-to-do-otherwise.

But can we conclude that in general, all else being equal, one needs to be sensitive to more reasons to be blameworthy than to be praiseworthy, and that it’s therefore in general easier to be praiseworthy than to be blameworthy? Maybe it’s often true for a certain type of action that with a fixed number of reasons to refrain, the strength of these reasons makes the difference between blameworthiness and praiseworthiness in specific cases. So it might be that Truman was praiseworthy for deciding in favor of military intervention in Korea and Johnson was blameworthy for deciding in favor of military intervention in Vietnam, while each case featured a roughly equal number of reasons to refrain from intervening, and the strengths and not the numbers of the reasons to refrain made the difference. In such military intervention cases the cost of intervention is always an important reason to refrain, but this reason can be stronger or weaker and can make the difference between praiseworthiness and blameworthiness. It would be important if cases that fit such a pattern were fairly widespread. In these kinds of cases it doesn’t seem that one needs to be sensitive to fewer reasons when the action is praiseworthy, and it doesn't appear easier to be praiseworthy than to be blameworthy.

Mark’s point that Flexibility is an ‘only if’ rather than ‘iff’ principle helps to make the idea a lot more plausible. Given that it seems plausible, I want to offer my own alternative take on *why* it seems plausible. The short statement of my view goes like this: Flexibility is best understood as an *evidential* thesis rather than a thesis about what is *criterial* for moral responsibility.

Here is the long statement. I mentioned before that I am a deep self theorist. This type of view says S is responsible for her action A if A reflects her deep self. Now, all deep self theories face a basic problem: We are not ordinarily privy to the goings on in another person’s deep self (we don’t walk around with cerebroscopes), so we need to rely on other sorts of tests to assess whether an action does or does not reflect the person’s deep self. One particularly useful test is to assess the reasons-responsivity of the mechanisms by which the person acts. Based on prior observations of the person in different contexts, I might be in a good position to assess whether she acts from a reasons-responsive mechanism or not. If she does act from a reasons-responsive mechanism, I have some evidence that her action does in fact reflect her deep self.

Now the inference from the reasons-responsivity of a mechanism to the contents of a person’s deep self can in some cases be dicey and it will surely be context-specific. One might naturally insist on different standards for different actions. Take a pretty serious or action like drinking poison. Before I believe that this action does in fact reflect the person’s self, I might insist on strong evidence. Thus, the test of responsivity I will need to use will be stringent: I want to be sure that there are a wide range of cases in which the person would refrain from this act when there is sufficient reason to do so. Only then is the inference from responsivity of mechanism to underlying deep attitude reasonably secure. For a trivial action like drinking water, I don’t need to insist on so much evidence—responsivity across a narrow range of cases suffices. So the deep self theorist endorses the basic idea behind Flexibility, but again, it emerges as an *evidential* principle rather than a principle that is part of what *makes it the case* that a person is morally responsible.

The reason I formulated the RS claim as an “only if” principle is that there might be other necessary conditions for freedom and this is a possibility I want to leave open. However, I do want the claim to capture the whole extent to which freedom is grounded in sensitivity to reasons. So I think you’re right to worry about this, and that’s why I said earlier that this is an important and underappreciated issue concerning RS views.

Consider the case of the addict. RS theorists want to say that he’s not free when he takes the drug because he’s not sensitive to reasons. But imagine that we took the following to be one of the relevant reasons to refrain from taking the drug:

R: The drug won’t make you feel high

If the addict were to come to believe that the substance he’s about to take won’t make him feel high, then he wouldn’t take it. Does this mean that he’s sensitive to reasons?

We don’t want to say he is. We want to say that this is *not* one of the relevant reasons we should consider. Why? I take it, because we’re interested in the reasons-sensitivity of the agent with respect to his act of taking a drug of the kind that makes you feel high (or of his act of taking the drug, as it falls under that description). (I say “we” because I believe that this applies to all RS views, not just mine.)

On your case of the button and the snake: I take it you’re imagining a scenario where there are no potential reasons to refrain from pushing the button (the button would prevent the torture of your infant child) but you fail to push it because there’s a snake present and you have a paralyzing fear of snakes. In that case an RS theorist would want to say that you are not responsible because you’re not sensitive to reasons. In this case, however, I don’t think that RS theorists would want to say that the presence of the snake is included in the identification of the act itself, though, but in the circumstances that one is holding fixed when examining the reasons-sensitivity of the agent. The thought is that, given your paralyzing fear of snakes, and given the snake’s presence, no sufficient reason to push the button is such that you would have been able to respond to it.
By the way, perhaps my own RS view has a slight advantage here over other views because I would put it just in terms of what actually caused the agent to fail to push the button. And it seems quite clear that no sufficient reason to push the button is such that its absence is among the *actual causes* of the agent’s failure to push the button in that case (given the presence of the snake and the agent’s fear of snakes).

Hi Derk!
Those cases are interesting because they’re of a certain kind such that one could “just as easily” be praiseworthy or blameworthy for them, depending heavily on the specific circumstances. Ordering a military intervention is something you could be praiseworthy or blameworthy for, depending on the circumstances. In that respect they’re different from, say, telling the truth, which tends to be right, or my example of ingesting poison, which tends to be irrational, with a few exceptions.

As a result, it’s quite tricky to find the right way of expressing what I want to express. In a certain sense, the difference between Truman and Johnson *is* a difference in the number of relevant reasons to refrain because the relevant reasons are supposed to be *sufficient* reasons to refrain, and assuming that Truman did the right thing and is praiseworthy, the cost of military intervention was *not* a sufficient reason to refrain in his case, whereas it was in Johnson’s case. So there is a difference between the two cases with respect to, at least, the actual reasons to refrain. Do you see what I mean?

Hmm… How about something like this: Flexibility tells us that the *type of act* is relevant in determining the required range of relevant reasons the agent needs to be sensitive to. In some cases, the type of act in question is such that there are many more right-making circumstances than wrong-making circumstances (with the limiting case being acts for which there are only right-making circumstances). Thus, in those cases, Flexibility motivates the idea that the range of reasons to refrain that one needs to be sensitive to in order to be praiseworthy is considerably narrower. But then there are other types of acts for which there’s roughly an equal number of right-making circumstances and of wrong-making circumstances. In those cases the range of reasons required to be praiseworthy will still be narrower, since there will be a difference in at least the actual reasons, but perhaps there won’t be a significant difference beyond that.
All of these results are motivated by Flexibility (and the focus on act-types) and perhaps this is the whole extent of the asymmetry between praise and blame, everything there is to be captured. How does this sound?

Thanks for your comment! I can totally see how it would make sense, from the perspective of a deep-self theorist who regards reasons-sensitivity as evidence of deep-self attitudes, to be flexible about ranges of reasons. So I’m glad to see that this discussion can be of interest to deep-self theorists like you too!

Of course, we disagree about what’s ultimately explanatorily relevant. You would say that a certain range of reasons-sensitivity (whatever range is appropriate for the relevant act-type) is only evidence of some other feature of the agent in virtue of which he’s free with respect to the relevant act, whereas an RS theorist would say that it’s what makes him free. An RS theorist (one who is not also a deep-self theorist) could in turn say that there is some notion of “deep self” that is relevant to freedom, but is only a derivative one (one that can be extracted from the patterns of responsiveness to reasons exhibited by the agent), so it’s not a truly deep deep-self feature, after all. I guess there’s also logical room for a third view, a hybrid view that takes both types of conditions to be equally basic in grounding freedom (the kind of view that I think Michael McKenna posted about some time ago, when he was a featured author).

Hi Carolina: Thanks for the reply. I think that your revised statement provides a good way for Flexibility to handle these kinds of cases.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan