Blog Coordinator

« Greetings to all of the new contributors to Flickers! | Main | A Question for Contributors »



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

Hi Carolina,

I enjoyed reading and thinking about your interesting post!

As for your question, I think both strategies could be successful. However, the success of the indirect strategy (the one you discuss here) depends on the success of the first strategy. Without a positive causal theory to directly explain why Ryder's steering Dobbin in that direction doesn't cause (or causally result in) the harm to the Romans, then people can argue that we should reject the principle of derivative responsibility since it is less obvious than premise (1). It seems that the most vulnerable link in the reductio reasoning is the principle of derivative responsibility instead of (1). In fact, it should be no surprise that many people would use cases such as One Path and Many Paths as counterexamples to the principle of derivative responsibility. Therefore, to use these cases in an opposite direction requires an independent motivation to support your principle.

Does this make sense to you?

Hi Carolina: Really interesting post! I’m definitely persuaded by your indirect strategy argument. But maybe this is only because I understand the theory of causation that you’ve set out and I think it makes a lot of sense. So I think that Jiajun makes a good point. It’s in some sense open that someone could reasonably reject the Derivative Responsibility principle to save (1), the causal claim, and so it’s important to have a plausible causal theory on which the causal claim is false. But I wonder about Jiajun’s claim that many people would use One Path/Many Paths as a counterexample to the Derivative Responsibility principle. I don’t think I’ve come across anyone who actually rejects that principle. And it seems massively plausible. Still, maybe we’d be justified in rejecting it and retaining (1) if we couldn’t come up with any causal theory that would make sense of rejecting (1).


Thanks for another great post.

Minor cavil: when we replace "blameworthy" for "responsible" in Derivative Responsibility, the principle churns out some odd results (depending on what the relevant epistemic conditions are). Suppose Cuthbert is blameworthy for X, X causes some starving children to be fed, and all the relevant epistemic conditions are satisfied. Is Cuthbert to blame for some starving children being fed? According to Derivative Responsibility, he is. But that sounds odd to me. So perhaps there needs to be some clause about Y being bad. Or do you think that one of the epistemic conditions is that the agent be aware or believe that Y is bad, or something like that?

More serious question/objection: This is sort of along the lines of Jiajun's remarks, but what is it about Sharks and Many Paths that makes it that the agent doesn't cause the relevant outcome? The following answer, which, if I'm not mistaken, is at least in the neighborhood of the one van Inwagen suggests, would not be friendly to actual sequence views: if it's already settled that p prior to any relevant action or omission on the part of the agent in question, then the agent doesn't causes it to be the case that p.

Hi Carolina,

Thanks for another excellent post!

I agree with the causal intuitions expresses in (1). That is, I agree that Ryder causes the harm in One Path but not Many Paths. And I agree that standing on the shore (rather than jumping in) causes the child’s death in No-Sharks, but not Sharks.

But I have an alternative view of how moral responsibility depends on causal facts such as these that doesn’t rely on Derivative Responsibility. Here is a quick sketch:

Start with an action individuation scheme. I prefer the Davidsonian approach, in which ‘flicking the switch’, ‘turning on the light’, and ‘alerting the prowler’ all describe the same action, but under different descriptions. One might use Goldman’s fine-grained approach as well. Nothing deep hangs on which action individuation scheme is used—only the formulation of the argument that follows would be different.

Next, notice that the following principle seems to be true:
(A) causal facts are relevant to what true descriptions are available to characterize what a person does.

For example, in No-Sharks, my ‘standing on the shore (rather than jumping in)’ and my ‘failing to save the child’ are both true descriptions of what I do. In Sharks, only the former is a true description of what I do. It cannot be correctly said that what I do counts as a ‘failing to save the child’.

Now add what I take to be an incredibly plausible principle:

(B) If what a person does cannot be correctly described as an instance of phi-ing, then she cannot be responsible for phi-ing (where the variable phi ranges over actions under a description).

A&B yield the same verdicts about MR in One Path/Many Paths and No-Sharks/Sharks as Derivative Responsibility. For example, in Sharks I am responsible for standing on the shore rather than jumping in but not for failing to save the child, because, among other things, the former but not the latter is a true description of what I do. Indeed, the two approaches, as far as I can tell, yield the same verdicts in other cases as well. They just arrive at these same verdicts in different ways.

I prefer A&B over Derivative Responsibility because I feel the former is somehow more basic and uncontroversial than the latter. (On a side note, I am writing a paper on the epistemic criterion for moral responsibility, and principles similar to A and B that appeal to actions under a description appear throughout my account, suggesting that A&B are part of a more general approach.) I’d be curious to hear your thoughts about the A&B approach. Let me make clear though that both approaches (A&B and Derivative Responsibility) respect actual sequence supervenience, so I am definitely not challenging that.

Hi again, Jiajun and Derk, and thanks a lot for your thoughts.

Jiajun raises some important issues. In order for the indirect strategy to work, it must be the case that the principle of derivative responsibility is more plausible than, in particular, claim (1). Indeed, my view is that this is the case: the principle of derivative responsibility strikes me as more plausible than the claim that Ryder causes the harm when all paths lead to Rome. Part of this is that I, like Derk, find the principle itself very plausible. Another part of it is, of course, that I have serious doubts about the causal claim.

But notice that it doesn’t follow from this that one needs to have a theory of causation that explains why (1) is false. I have a theoretical explanation of why it’s false, but it’s not even a theory of causation. Notice that, if I were to say: “Here’s my theory, it entails that (1) is false”, that would probably lead to complaints that my theory is not the right theory, or that I have not adequately shown that it is true, or preferable to other theories. But you’re right that having some sort of explanation of why I think that (1) fails helps. So perhaps the best strategy is a combination of the direct and indirect strategies (which is, in fact, what I’m trying to do in my book!).

Which leads me to a related thought: One thing I wonder about is if the “direct” strategy is really as direct as I was describing it to be. When I implement the direct strategy, I appeal to certain causal intuitions (such as the intuition that the steering of the horse is not a cause in Many Paths) but I note that certain theories of causation entail otherwise. Instead of then trying to refute every other theory out there, at that point I say that I want to leave open the possibility that there might be different concepts of causation, and that those other theories capture a different concept. All I want to claim, in other words, is that the causal claims I make are true with respect to the *relevant* concept. Now, what is the relevant concept of causation? Clearly, one that is related to responsibility. And, one could add, the principle of derivative responsibility expresses certain important features of that concept! (As in: the relevant concept is one that can transmit responsibility, or can let responsibility be inherited from other things, in the right kinds of circumstances.) Again, then, perhaps this is a reason to think that the two strategies are in fact intertwined, or that the best strategy is one that combines the two in this kind of way.

Does this make sense?

Thanks for the interesting points, Justin.

Indeed, van Inwagen’s view on causation seems to be one that is (perhaps unsurprisingly) not at all friendly to actual-sequence views. In fact, he seems committed to the idea that causation requires counterfactual dependence (C causes E only if, if C hadn’t happened, then E wouldn’t have happened). (See p. 177 of *An Essay on Free Will*. He’s one of 2, or perhaps 3, people I know who believes this!) The view that causation requires counterfactual dependence entails that the agent is not a cause in Sharks or Many Paths. It also entails that the agent is not a cause in a Frankfurt scenario, which goes against what I take to be a main claim of actual-sequence views.

Is this a reason for actual-sequence theorists to be worried? I don’t think so. The claim that there is no causation is scenarios of causal preemption is so initially counterintuitive that the burden is on someone who puts forth that claim to defend it, and not on actual-sequence theorists to attack it. Notice, in particular, that van Inwagen doesn’t appeal to the causal claim he believes is true of Frankfurt cases in order to defend his views on responsibility. The only reason he happens to mention what his views on causation are seems to be that a referee pointed out to him that the premises of his argument for responsibility had those consequences about causation… (Kudos to the referee for having noted this!)

On your starving the children example: It’s an interesting example. Why are you thinking that Cuthbert (where did you get the name?) is blameworthy for X, even if the epistemic conditions are met (e.g., he knows that it will result in some starving children being fed)? Perhaps you’re thinking something like this: X will result in 10 children being fed, but not-X will result in 100 children being fed? In that case, Cuthbert could still be blameworthy for X, even if Y is a good outcome that Cuthbert could foresee. I haven’t really thought carefully about what we should take the epistemic conditions to be. But this example makes me think that, as you suggest, we should probably take the epistemic conditions to include something like the ability to foresee that X would result in some bad consequences of the kind exemplified by Y.

Also, perhaps we could avoid these tricky issues altogether by reformulating the principle in something like the following way:

If S is blameworthy for X because he was able to foresee that X would causally result in Y, and X causally results in Y in roughly the way S could foresee it would, then S is blameworthy for Y.

This formulation of the principle makes it seem even more plausible, and it can still do the work I need it to do.

What do you think?

Chandra: Thanks for your response, and for bringing up the A&B proposal.

Frankfurt’s own reaction to the Sharks case (I think it was in “An Alleged Asymmetry between Actions and Omissions”) was along the same lines, if I am not misremembering. Basically, Frankfurt pointed out that *there is no omission* on the part of the agent when the sharks would have prevented him from saving the child. Frankfurt didn’t put it in causal terms, though, as you do, and thus for him there’s still a question about why there is no genuine omission in that case but there is a genuine omission in other cases, whereas you have a clear answer to that question in terms of causal facts. By the way, note that, when you put it in Frankfurt’s terms, the equivalent of your principle B becomes not just “incredibly plausible” but also trivial. For, when applied to Sharks, it’s the claim that if there is no omission O, then the agent cannot be responsible for O. This is an instance of the more general and trivially true principle: if X doesn’t obtain, I cannot be responsible for X.

A potential limitation that I see with that kind of view is that, as it stands, it doesn’t obviously allow us to extend responsibility to outcomes in the world. Imagine that, when the child dies, his family is devastated. The principle of derivative responsibility allows us to say that the agent in No-Sharks is responsible for that too. How would you deal with outcomes? Or perhaps you don’t want to extend responsibility to outcomes, or you think that responsibility for outcomes doesn’t matter?

One thing you could always say is: There is something the agent does, in No-Sharks (or, in your terms: there is an accurate description of what he does), which concerns the family’s sadness. Namely: he brings about the family’s sadness (perhaps, even, he makes them sad). Thus, by explicitly appealing to causal connections of “bringing about” or related notions, you can incorporate outcomes into the account. But then I start to feel like the two proposals are not essentially different. For, clearly, you can only be responsible for bringing about a family’s sadness when you meet certain epistemic conditions, the ones that the principle of derivative responsibility talks about. And so really what you’re saying is that you’re responsible for the family’s sadness when what you do or fail to do causally results in their sadness, and you meet certain epistemic conditions, which is just what the principle of derivative responsibility says.

Upon reflection, I guess I’d say the same thing about the omissions case too. Even if you and Frankfurt are right and there’s no genuine omission by the agent in the Sharks scenario, it is true that the child dies, and it is true that you don’t save him. So there’s still an issue of whether you can be responsible for that, even if your not saving him cannot be adequately called an “omission” of yours.


Yes, something like the revised principle sounds better.

Hmmm, is van Inwagen (or those who hold positions in the same neighborhood as his) committed to the claim that there's no causation in preemption cases? Perhaps there's something he could say here. Consider, e.g., Ravizza's Erosion case. Betty detonates explosives at t1 thereby causing an avalanche that crushes the enemy base at t3. By doing so, she preempted a natural process that would have resulted in an avalanche at t2 that would have crushed the enemy base at t3. Assuming the natural process antecedently guaranteed that there would be an avalanche that crushes the enemy base at t3, I think van Inwagen would say that Betty didn't cause it to be the case that there's an avalanche that crushes the enemy camp at t3. What she did cause to be the case, however, is that there's an avalanche started by her that crushes the enemy camp at t3 (or something like that). It's a sort of flicker strategy but applied to causation.

This is all about causation of facts. What about causation of events? Well, if I'm not mistaken van Inwagen holds that events are individuated in part by their causes. If that principle is true (big "if," to be sure), then there's no problem with preemption cases, as the preempted sequence would have resulted in a distinct token event than the event that resulted from the preempting sequence.

So it looks like van Inwagen has things to say about all this stuff that would enable him to handle preemption cases. Whether what he says is ultimately plausible is, of course, another story.

Thanks for the response, Carolina.

I agree with you that you don't need a positive theory of causation if the principle of derivative responsibility is already more plausible than the causal claim (1). However, I don't totally share your confidence in the principle and your suspicion of (1), even though I share them to a certain degree. Now, it depends on what aim you set for your project: to move the other side's opinion or to merely propose a reasonable position. But no matter what, treating the two strategies as combined effort seems to be a really good idea.

And what you said in the second part of your comments makes a lot sense to me. In fact, I take such a weak defense to be the best option for your purpose. The only worry is that you need to be careful of a potential circle: why is the principle of derivative responsibility more plausible than (1)>partly because (1) is false according to the relevant concept of causation>why is (1) false according to the relevant concept of causation>because the relevant concept of causation is the one mentioned in the principle of derivative responsibility and the indirect strategy shows that (1) should be rejected in light of the principle of derivative responsibility>how does the indirect strategy show that?>because the principle of derivative responsibility is more plausible than (1).

If I uncharitably understand your claim (of course, I believe this is not what you mean) that "the relevant concept is one that can transmit responsibility", it will strengthen the circular worry. The feature of the relevant concept of causation has to be more concrete and substantive than "being one that can transmit responsibility", which might be understood as a potentially circular claim that the relevant concept of causation is simply the one (or anyone) that makes the principle of derivative responsibility very plausible.

So if you stay with the indirect strategy of defending supervenience principle based on the claim that the principle of derivative responsibility is already more plausible than (1), you will be safe. But the significance of such a project will be limited as well. However, if you try to use the direct and indirect strategies in an intertwined way to support each other, it brings the danger of circularity. Again, I think it will help avoid the danger if you provide a positive theory of causation (not mean you need to prove it or disprove other theories, just show it is a plausible one). At least, you could provide us a pretty clear picture how the relevant causation looks like.

Sorry for the tedious response. I hope you find it helpful.

Justin: Yes, thanks, that is van Inwagen’s treatment of preemption cases as I understand it too. So he wouldn’t say simply that “there is no causation in preemption cases”, as I put it. That’s a misleading way to put it. But he does say that in such cases there is no causation of an event that would have occurred anyway, or an overdetermined event, or an inevitable event (in that sense, there is no causation in a case that is genuinely a preemption case, or one where the effect is genuinely overdetermined). More generally, he’s committed to the claim that nothing can cause an event of that kind, and this seems intuitively wrong.

On his view that events have their causes essentially: I side with Lewis here. This just seems wrong about events as we ordinarily understand them, and also as events as causal relata. What matters in this context, of course, is how we think about events as causal relata. And it seems clear that, given how we think of events as capable of standing in causal relations with other events, events don’t tend to have (all) their causes essentially. Do you agree?

Hi Carolina,

Jumping back to a point you made previously, I’m interested in your appeals to causation’s role for responsibility being only regarding the ‘relevant’ notion of causation. I had two thoughts on this score, though neither is particularly well-formed (thus, perfect for a blog setting).

First, once we have disposed with causation simpliciter, and instead have restricted our focus to causation in the relevant sense, I wonder whether we can reliably distinguish our intuitions about causation from those about responsibility. That is, could it be the responsibility horse pulling the causation cart, to butcher a metaphor? It may be that in cases like the Sharks and Paths pairs, we cannot separate out responsibility ascriptions and thoughts about causation in the sense relevant to responsibility. So I guess my first question is: can you say more about the relevant sense of causation that restricts it from causation simpliciter but doesn’t amount to ‘causation in the responsibility-relevant sense’?

A second, but related, thought concerns the peculiarity of omissions. Consider an analog to the Sharks pair of cases. Suppose there’s an inflatable raft with a motor (but driverless) headed toward the child, which would allow him grab on and not drown. In the sharks version of the scenario the sharks will attack the raft and pop it (but not so in no-sharks, obviously. Now suppose motor fails en route, but before the sharks. If this pair is to be treated analogously with the standard versions, should we think that the motor’s failure causes the boy to drown in the no-sharks version but not in the shark’s version? Or do the sharks not matter because the boat’s failure to reach the boy can’t cause anything?

One thought is that motorboats can perform omissions anything anymore than they can perform actions, but of course, in thinking merely about causation, non-agent objects act on other things all the time. But they don’t omit. So I wonder whether the causal claims here are only relevant because we’re considering what an agent might have caused, and that’s principally because we’re interested in what he might be responsible for. But then, as above, I wonder whether the causal judgments aren’t really just elements of responsibility judgments.

Jiajun: Yes, this is helpful, thanks. I guess what I’m thinking is that one could look at the two strategies as basically just one main strategy that combines elements of the two. That main strategy consists in arguing that a claim like (1) is false, and thus supervenience is safe, because the relevant concept of causation is such that (1) is false. What’s the relevant concept of causation? If there is more than one, it’s the one that’s relevant to moral responsibility debates. At least part of that concept, I think, involves (roughly) the idea that causation can transmit responsibility or, more generally, can make people responsible for things under the right kinds of circumstances. This isn’t necessarily to commit oneself to the idea that the principle of derivative responsibility is itself essential to that concept, but only to the thought that causation and responsibility are connected in important ways (presumably, some of those ways will reflect at least something in the neighborhood of the principle, though). And, as you point out, there would still be much more to the relevant concept than that. As far as I can see, this avoids any vicious circularities in the reasoning. But I definitely agree with you that it helps if one can, in addition, provide a plausible story about how/why (1) is false.


I tend to side with van Inwagen about causation of facts, but with Lewis and you about events. This position strikes me as unstable and is probably untenable in the end, but I'm a novice when it comes to causation and don't pretend to have all my ducks in a row there. So in the Ravizza case, I don't think Betty caused it to be the case that an avalanche crushed the enemy base at t3, and I don't think she's responsible for the fact that an avalanche crushes the enemy base then. What she did cause to be the case, and what she is (or may be) responsible for is some more precise fact. When we turn to events things get tricker. Take the destruction of the enemy base. Did Betty cause that event? I think so, and I think it could have been the same event even if Betty hadn't been in the causal stream leading to it. So I think I'm willing to say that Betty is responsible for an event should couldn't have prevented. That's consistent with an alternative possibilities view, though, for I'm inclined to say that a person is responsible for an event she caused only if she could have avoided causing it. In any event, I don't think any of this bears on your supervenience claim or your defense of it.

Hi Matt! These are interesting and important questions.

In claiming that what matters is the “relevant” concept of cause, I meant to claim that there is some *natural* concept of cause that is connected to responsibility in important ways, and that that’s the concept that matters for our purposes. You’re right that in some cases our intuitions about causation can be tainted or driven by our moral intuitions (in particular, our intuitions about responsibility), and this means that one needs to be particularly careful in dealing with those causal intuitions.

Your motorboat case is great, precisely for that reason. It can be used to illustrate the kind of concept that I have in mind. The relevant concept is, I think, one that entails that the motor’s failure *is* a cause of the child’s death in a No-Sharks scenario (but not in a scenario like Sharks). In other words, it is one according to which the agent’s failure and the motor’s failure are completely on a par (even if they are obviously not on a par with respect to responsibility). We can imagine a similar variation of the Paths examples, one where a gust of wind steers the horse in the relevant direction. I think there is a clear sense in which those things are causes of the ensuing outcomes when the circumstances are right (in No-Sharks and Many-Paths types of cases). So this suggests to me that the causal intuitions in the original No-Sharks and Many Paths cases, the scenarios where an agent is involved, are genuinely causal intuitions and are not tainted by our moral intuitions (and, of course, the same goes for the Sharks and One Path cases).

By the way, you note that it’s weird to speak of things “omitting to act” on other things, whereas it’s quite common to speak of things “acting” on other things. That seems true. (I wonder why that’s so… it would be interesting to find out.) But the relevant question is not whether things can omit doing things, is it? The relevant question is whether “negative” happenings involving things can bring about or causally result in other happenings. And it seems clear that we think this way, at least in some cases (consider: the alarm clock’s failure to go off resulted in my being late for my appointment, the absence of rain caused the drought, etc.).

I’m not sure that I’ve answered your questions. Does this make sense to you? Thanks for your comment!

Hi Carolina,

Thanks for your detailed response!! You are right to say that my Davidsonian/ Frankfurtian A&B approach (maybe this might be called an ‘action description’ approach) does suggest that in Sharks, the relevant omission does not really occur (of course I omit to jump in the water, but I don’t in fact omit to save the child). This is of course because, due in part to the relevant causal facts, the action description ‘my failing to save the child’ does not correctly characterize what I do. I wonder whether this feature might very well be an advantage of the view.

Consider Fischer and Ravizza’s case of Sue who (falsely!) thinks she can end a terrible drought by doing a rain dance. Mean-spiritedly and deliberately, she does not do the rain dance. Is it correct to say that in doing what she does in fact do, she performs the following omission: ‘She fails to save the town’? I think the answer is no. The action description view explains why this is the case: Because of, inter alia, the relevant causal facts, ‘failing to save the town’ is not a correct description of what she does. Moreover, since if I don’t omit to do X, I can’t be morally responsible for omitting to do X, it follows Sue is not morally responsible for failing to save the town. Now if a view says that in Sharks, I do omit to save the child but I am not responsible for the omission (which I think you want to say and F&R’s view does say), then by parity of reasoning, it seems it should say that in the Sue case, Sue does omit to save the town. This seems wrong.

I realize that we might have gotten off onto a tangent here discussing the nature of omissions. I just want to reiterate that to your main challenge—are there cases in this vein that are counter-examples to actual sequence supervenience—again, I am in agreement with you that the answer appears to be no.


I’m not sure whether your position about events/facts is unstable or untenable. It’s kind of hard to say because people generally pick one category of things, not both, as the causal relata…

What you say about responsibility for events is interesting. Do you think that the best reconstruction of the alternative possibilities principle for events is:

“S is responsible for E (an event that he caused) only if S could have failed to cause E”

instead of:

“S is responsible for E only if S could have prevented E”? (van Inwagen’s own principle, PP1)

I think so too (I’ve argued for that in my “Causation and Freedom” paper), but I thought no one agreed with me! As you suggest, the avalanche case is a counterexample to the second principle but not to the first principle. Betty couldn’t have prevented the destruction of the enemy base, but she could have failed to cause it.


I love interesting tangents!

Indeed, F&R use a quite permissive concept of omission so they’re open to the idea that there is an omission in the Rain Dance case. But they’d argue that the agent is not responsible for his omission, for the same reasons he’s not responsible in Sharks. I think you’re right to suggest that it’s weird to say there’s an omission in that case (or, for that matter, in the Sharks case). “Failing to save the child/not saving the child”, and “failing to end the drought/not ending the drought” sound better.

Speaking on their behalf (and, bear in mind, I might be wrong about their motivations), I suspect that the reason they use such a permissive concept of omission is that they don’t want to get involved in a debate about what counts as an omission (I think they have a footnote somewhere where they say this, but I might be misremembering). They could add (although this might just be me thinking) that, even if there’s no omission, a question would remain about why the agent is not responsible for the outcome (the drought, or even the negative state of affairs of his not ending the drought), so we’re merely postponing the central question.

At any rate, that’s what I’m thinking the problem is with that kind of reply, that it doesn’t get to the heart of the matter. (Do you see what I mean?)

A final point: There’s a natural way of interpreting F&R’s own view of omissions according to which it follows from it that there is no genuine omission in those cases. F&R understand complex omissions as something like a bodily movement resulting in a narrowly specified consequence (such as “the child not being saved by me”, or “the drought not being ended by me”). But if we understand “resulting in” as causally resulting in, and if the causal facts are what you and I think they are, then it follows from their view that there is no omission in those cases!

Hi Carolina,

Thanks for the reply. Parallel judgments for Sharks and Motorboat certainly addresses the second part of my question. So a follow-up question would be: is the notion of causation at play in Motorboat the same one as Sharks? That is, are they *both* the causation that’s relevant to moral responsibility? Or does Motorboat only pertain to causation simpliciter?

Here is one way to frame a potential worry I have (though I'm not sure it applies to your view specifically, rather than as a general worry). Omissions have two contrast classes. We can contrast them with actions. But we can also contrast them with pure failures. If I jump in the water in No-Sharks and swim toward the child but can’t get there in time, I fail to save the child (but I didn’t omit to). Following from Motorboat, if I fail to save the child, I have caused the child’s death. According to the Derivative Responsibility principle, it seems I might now be responsible for the child’s death.

One might want to say that the epistemic conditions fail here, but that doesn’t sound promising to me. I am in just as good an epistemic state when I fail as when I succeed. Instead, you might want to say I’m not responsible for failing to get to the child in time. But it isn’t clear this is promising either. Of course, there may be some cases in which this is true – I get a muscle cramp, or the wind picks up and thwarts my efforts. But we can imagine simple enough cases in which I’m just not fast enough. It would seem I cause the failure just as much as I cause the success.

This worry can be perhaps compounded if we think it plausible that, if failings are causes, we have an explosion of causes. For, just as the motorboat’s failure causes the drowning, so does the failure of a nearby dolphin to rescue or a wave to pick up the child and deliver him to shore.

To connect this to the first thought from my previous post: The above seems true of at least the alarm clock case you mention. Its failure to go off causes me to be late in the same way, it seems to me, as my friend’s failure to call early this morning, or my neighbor’s failure to start his motorcycle as he usually does, or any other event that would have been sufficient for waking me up. It is true that, when we explain why we’re late, we’re apt to point to the alarm clock as the culprit, rather than a host of other failures that, but for there non-happening, I would have awakened on time. But this makes me wonder whether it isn’t some species of responsibility judgment doing the work. Since I’m not apt to blame my neighbor or friend or whatever for my oversleeping, the alarm clock gets my ire. (Of course, it isn’t proper blame in this case, but I wonder whether it might be a similar or fictional analogue.)

Ok, that was way longer than I thought it was going to be. Apologies!


I haven't read your "Causation and Freedom" yet. But, yes, I agree that “S is responsible for E (an event that he caused) only if S could have failed to cause E” is the more plausible principle. So looks like we're allies about that principle.

Hi Justin and Carolina,

Your discussion about van Inwagen's response to overdetermination (maybe Frankfurt as well) sounds really interesting. There are two issues I would like to discuss with both of you.

(1) About whether events have their causes and effects essentially, I want to keep skeptical here. I don't know enough about these metaphysical issues. But I think van Inwagen has something to say either way. If its cause is essential to an event's identity, his response will works as he proposed in Essay. If we focus more on *intrinsic* features of events and treat their causes as extrinsic, then as Justin suggests, the destruction of the base it could have been the same event even if Betty hadn't been in the causal stream leading to it. However, this position leads to the following difficulty: we cannot be responsible for a particular event (including a particular action).

Suppose X has choice about whether to kill Y, but no choice about when. And X kills Y at t1. In another possible world, the same person X has no choice about killing Y (maybe due to coercion), but has choice about when (between t0 and t2). And X kills Y at t1. Then, X is morally responsible for the death of Y at t1 in both scenarios. If these two events "the death of Y at t1" are individuated only by their intrinsic features without appealing to their different causes and other relational properties, how can we explicate the moral difference between these two instances of responsibility. Even if X could has different things (other than the event as a result, e.g., different intentions) to be responsible for, the sentence "X is responsible for the death of Y at t1" or "X is responsible for killing Y at t1" would become almost meaningless. Because we cannot tell if, by saying so, X is held responsible for the death or the time, which has a huge moral difference.

Maybe there is a way to avoid this difficulty, I am glad to have your thoughts. Perhaps we should simply replace a particular event with a particular aspect of the event as the object of responsibility. It may bring about some interesting implications.

(2) However, I am not trying to defend van Inwagen's response here. His effort to save his principle of possible prevention seems to be ineffective in face of simultaneous overdetermination cases (Fischer discusses this somewhere I think). If two persons simultaneously shoot someone, the death isn't preventable to either of them. But it is obviously wrong to say that no one is responsible for the death. I don't see any plausible incompatibilist response to this problem.

It could be a good idea to reformulate the original principle as you do here. But I wonder if the new principle could resist the attack from Frankfurt's cases.

By the way, I am always curious why incompatibilists (maybe someone did) don't simply propose a principle like this: S is responsible for something (actions, omissions, or consequences) only if he can avoid such a responsibility. This expresses a large part of incompatibilist intuition more directly, and be good enough to be used against compatibilism.

How do you think about this?

Thanks, Matt, these are good questions.

The way I see it, “my failing to jump in” in No-Sharks is only a placeholder. (Consider: What if I were to jump in and just stand still in the water, instead of swimming toward the child? In that case my failure to jump in per se would not be what’s relevant…) What I think we should say (everyone, not just actual-sequence theorists) is that the relevant omission is the one that would have, in the circumstances, resulted in my saving the child. So perhaps it’s something like my failure to get in the water and swim at a certain speed, or something like that. Once we’ve singled out the relevant omission, the question is whether the agent is responsible for *it*, and only in that case (the principle of derivative responsibility would say, assuming the epistemic conditions are met) the agent is also responsible for the failure to save the child.

If it helps, for our purposes here, you may think about the relevant concept of cause as counterfactual dependence (or at least as counterfactual dependence being sufficient for it). There are many things that the child’s death, or my saving the child, counterfactually depended on, in No-Sharks: my failure to jump in (and swim at a certain speed?), a certain failure by the motorboat, another “failure” by the dolphin, etc. Also, consider a paralyzed or naturally very slow person’s not swimming toward the child at a certain speed. The child’s death also counterfactually depended on those non-doings (although we obviously wouldn’t call them “omissions”). So all of them would be negative causes, according to this concept of cause. As I see it, this “explosion” of negative causes is not a problem. Instead, it’s a reflection of the fact that we’re dealing with a purely natural concept, and not with a normatively loaded one. What would be a problem, I think, is if we were to try to *make* distinctions between failures by agents, motorboats, and dolphins, because then it’s likely that our causal intuitions would end up being driven by normative considerations, possibly including considerations about responsibility.

Justin: Woo hoo! I'm glad we agree.

Jiajun: The claim that events are not individuated by (all) their actual causes is consistent with the claim that whether an agent is responsible for an event (action) on a given occasion can depend on what its actual causes are. (In fact, that's what actual-sequence views say: that freedom is a function of what the actual causes are, where this doesn't mean that the actual causes determine the identity of the event itself.)

The alternative formulation of the alternative-possibilities principle you suggest (in terms of the ability to avoid responsibility) has been proposed, in different versions, by Wyma, McKenna, and Otsuka. Fischer and Pereboom have both replied to this strategy.

Hi JiaJun,

I'm not sure I understand your worry. Depending on the details of the two cases you mention, I think I'd say that X is responsible for "the death of Y" (where this is an event) in both cases. Moreover, X may be responsible for the state of affairs "Y dies at t1" in both cases as well.

Also, you suggest that incompatibilists should propose a principle along the following lines: a person is responsible for something only if he/she could have avoided responsibility for it. Some incompatibilists have proposed principles along those lines. Michael Otsuka has defended a principle like that, and Chris Franklin defends a version of PAP that is similar to the principle you suggest. But note: one needn't be an incompatibilist to accept alternative possibilities principles. Many compatibilists used to, and some still do, accept such principles. They just think agents can satisfy the conditions specified the principles even if determinism is true.

Hey Justin,

Exactly! If you will say that X is responsible for the same particular event in both cases, it's like you will say that both Hitler and his secretary were responsible for the breakout of WWII (or Hitler and his secretary as a group should be responsible for it) since Hitler decided to start the war and his secretary helped pick the date (let's say starting the war that day was particularly annoying to people like peace, therefore, it was not a morally neutral picking). But obviously, their responsibilities with respect to the breakout of WWII were extremely different. And I think that the ascription of responsibility must express such a difference. However, if what Hitler and his secretary were responsible for is exactly the same thing, i.e., the breakout of WWII as a particular event, how can we differentiate their different responsibilities with respect to WWII?

In my view, this is a serious problem. And it seems to be an inevitable upshot if we take particular events (if they are individuated locally) to be the object of responsibility.

Thanks so much for the exchange, Carolina. I'm learning a lot! I see the point about the 'naturalness' of the notion of causation, and I agree that we wouldn't want a normatively loaded concept. And I’ll accept counterfactual dependence as a characterization of that natural notion.

But turning back to the principle of derivative responsibility, doesn't it suggest that the swimmer is responsible for the child's death in No-Sharks when he fails (e.g., because he's too slow) to save the child? If he’s not, it would have to be for an epistemic reason, but it seems one is in the same relevant epistemic position during a failure as during a success. And since it seems to me mistaken to think that the swimmer is responsible for the child’s death in such circumstances, I wonder whether something is amiss with the Principle of Derived Responsibility. (Or, as is *at least* equally plausible, I’m getting something wrong.)

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan