Blog Coordinator

« A Brief Introduction | Main | Beliefs in Free Will and Punishment: An fMRI Investigation »



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


Greetings from a fellow Tucsonan!

I think the concept of supervenience brings us closer to understanding the truth, but there may be a supplemental idea that’s needed. Supervenience is directly related to the theory of determinism, wherein lower-level properties of a system determine its higher-level properties. To complete the theory of supervenience and have it taken seriously, I’m thinking that we need to realize that new properties (i.e., life) may emerge at higher-level systems wherein those new properties aren’t determined 100% by lower-level properties. Yes, lower-level properties principally determine higher-level properties, so supervenience is generally true, but not *all* higher-level properties are determined by lower-level properties.

When we consider the claim that “freedom is exclusively a function of the actual sequence”, I think we need to include new emergent properties (life) as being part of the “actual sequence”. As soon as we do that, we’ll have an account of supervenience that’s taken seriously.

Hi Carolina,

Really looking forward to your posts and to your book! I just had a couple of questions.

Would it be enough to save Frankfurt's view from your worry that the relevant higher order desires get in the causal stream (with the typical caveats about non-deviant causal connections)? To do so, it needn't overdetermine the relevant outcome. Surely higher order attitudes play important roles in action production, and it doesn't strike me as obviously implausible or unmotivated to say that such desires are relevant to freedom and responsibility insofar as they play an appropriate causal role.

Second: is it the total causal history of X that matters on your view or just some relevant part of it?

Hi Carolina,

I think yours is a promising approach (for what that's worth, which is not much). A question about the criticism of the F&V view: I don't follow the claim that 'the mechanism itself (the sequence of events that is actually causally operative) could in principle be exactly the same in both cases,' Are we talking about duplicate agents in the two imagined cases? And if so, wouldn’t the mechanisms be the same (even dispositionally)?

Hi again Carolina--very thoughtful and interesting post.

First I wonder if some (Fischer maybe) would parse the supervenience of S in terms of moral responsibility rather than freedom. I’ll ignore that for now.

Second, I like your clear application of S to Frankfurt’s classic case. I see your point about the sufficiency of the addictive actual sequence in each case despite differing higher-order desires--it appears to undercut any real role for those desires in judging one addict freely “willing” and another “unwilling”. But one real difference in those desires is a value-difference. We judge willing addicts more negatively than unwilling ones (probably because of the poignancy of a losing battle with oneself as opposed to cavalier self-destruction). Given that difference, we want the difference to be a real and efficacious one.

So, in short, S may be about value-differences in distinct kinds of actual-sequence causal mechanisms of consciousness that supervene upon other causally efficacious mechanisms that finally drive overall behavior. There’s something to be said for this. “Freedom” and “moral responsibility” are concepts not devoid of value content in most contexts; I’d say more strongly they are essentially value-driven concepts.

But even if this is the case, though we can use differing values to say that there are very distinct actual sequences of willing versus unwilling addictive behavior, since these conscious valuation-mechanisms overlay underlying driving mechanisms that can be sufficient for the same behavior, then such conscious mechanisms appear to be causally sterile or epiphenomenal.

I think that’s why Frankfurt was motivated to create at least the appearance of asymmetry between higher-order willing and unwilling desires in terms of overdetermination. If the differing values of willingness and unwillingness don’t have real efficacy, then they can’t uphold claims of responsibility. However, if the basis of overdetermination (at least in addiction cases) is actually at a lower level upon which addicts’ conscious value-systems operate, then it’s hard to see how those values translate into being causally efficacious in terms of behavior. But maybe their efficacy is derivative, and not metaphysical but epistemic, driving our value judgments of willing and unwilling processes as manifesting axiologically distinct actual sequences of overdetermination.


I think this is a brilliant post. As I understand it, you are pointing out that actual-sequence theorists, like Frankfurt and Fischer, might not be as consistent in their allegiance to the actual-sequence, as we otherwise thought.

I don't have much further insight to add. But I think that you might like an old post I made at The Garden of Forking Paths, where I highlighted similar problems for compatibilists who are concerns with counter-factuals:

Hi Carolina,

Terrific post! I am really looking forward to the discussion that follows.

Even though I am in agreement with you that the correct theory of moral responsibility should be an actual sequence theory, I disagree with S. Here is why. I am a deep self theorist. My version of the view says that a person is morally responsible for her action A if A expresses her deep self. Expression, as I see it, is in part a *semantic* relation: A person’s action A expresses her attitude B if A aligns with the *content* of B (this is rough, but hopefully the intuitive idea is clear. I develop the account in detail elsewhere). A deep self view nicely captures what distinguishes the willing and unwilling addict. The willing addict deeply loves the pleasures of being high above all else, and thus his drug-directed actions express his self and he is MR for them. The unwilling addict deeply hates his addiction, and thus his drug-directed actions do not express his self and he is not MR for them. (I know Frankfurt’s higher-order desire view might itself be thought of as a deep self view. His view suffers from enough well known problems that I thought I’d present my version of a deep self theory directly.)

I am not trying to convince you that a deep self theory is correct (though I have been known to proselytize a bit!). Rather, I want to try to convince you that a deep self theory should be considered an actual sequence view. This is because on the deep self view, everything that matters for moral responsibility is present right there in the sequence that actually unfolds. In particular, the question of whether the person’s action expresses or fails to express her self is settled by looking at her actual action and her actual deep attitudes, and assessing whether the former express the latter. We do not look at whether the agent ‘could have done otherwise’ or ‘had access to alternative possibilities’ or whether the mechanism that issued in her action would have issued other actions in other situations.

Yet the deep self view, as you correctly point out, violates principle S. As you nicely point out, the willing and unwilling addicts illustrate this: What distinguishes the addicts is not a *causal* difference, but rather a difference in the *semantic* aspect of the expression relation: The actions of the willing addict align with the content of his self, while for the unwilling addict, they do not. In light of this, I think the right course of action is not to declare that the deep self view is not, after all, an actual sequence view. Rather we should amend principle S to allow both for causal properties as well as semantic properties of the type implicated in the expression relation:

(S’) An agent’s freedom with respect to X supervenes on the properties (causal or semantic) that obtain in the sequence of events that eventuates in X.

What do you think of this amendment, and to my argument that the deep self view should be thought of as an actual sequence view? Thanks ahead of time for your thoughts on this.

Thanks to all for your thoughtful comments and questions! I’m going to try to give a sort of collective response to the reactions so far, since it seems to me that in some cases the points are interconnected. (I apologize if in doing this I fail to respond to some of your points!)

First, there is a reason I formulated the supervenience claim for freedom instead of responsibility: I don’t believe it holds for responsibility. Responsibility is typically taken to include a metaphysical component and an epistemic component. Given its epistemic component, I don’t believe that the arguments that support the claim that freedom supervenes on actual sequences extend to responsibility. A quite common example that I think shows that is: imagine someone who performs a certain act that he knows will have a bad consequence but doesn’t perform it because he wants the bad consequence to occur (he performs it for other reasons). He is blameworthy for the bad consequence, simply because of the belief (or knowledge) that he had that performing that act would result in such bad consequences. But that belief (knowledge) is not part of the causal chain that led to his act. In fact, the causal chain is relevantly the same as that of someone who performs the act not knowing that it would issue in the bad consequence (i.e. someone who is not responsible for the consequence). So there can be a difference in responsibility without a difference in the causal chain. Supervenience fails for responsibility. I think this is right. The belief that you are acting wrongly can make you blameworthy, in the right circumstances, simply by being there. But, when we are dealing with the non-epistemic, metaphysical component, the story is a different one.

So, assuming you buy this, a question arises about what’s the best way to deal with the “deep self” view. One possibility is to say that higher-order desires can be like beliefs in that they can be relevant to responsibility without doing any causal work. I’m not sure what to think about this. One problem is that it doesn’t help with the freedom claim: it doesn’t seem like someone like Frankfurt wants to say that the willing and unwilling addicts differ with respect to only their responsibility; they also differ with respect to how free they are. Another possibility is to weaken the supervenience claim for freedom itself, in the kind of way that Chandra was imagining. I actually consider that possibility in the relevant chapter of my book. I imagine a distinction between two different kinds of facts that can be relevant to freedom: facts that can be relevant by playing a certain causal role, in actual *or* counterfactual sequences (call them “causally relevant” facts), and facts that are of a quite different sort, and can be relevant in other ways. Then an actual-sequence view would be one that says that freedom supervenes on the *actually* causally relevant facts, plus the other kinds of facts.

What one should try to avoid, however, is to weaken supervenience in a way that would deeply go against the initial motivation provided by Frankfurt cases. This would be the case if one were to say, for example, that freedom supervenes on *all the actual facts* (be them causally efficacious or not). For the actual facts in a Frankfurt case include, for example, the presence of the neuroscientist. And the kind of thinking motivated by Frankfurt cases is that the presence of the neuroscientist is just not relevant, we could pretend he’s not there and that wouldn’t affect the agent’s responsibility. The same goes for the view that wants to add purely dispositional or counterfactual properties of the actual mechanism (Josh: what I mean by this is: properties that are not *in fact* causally efficacious, but would only be efficacious in other, counterfactual scenarios). In other words, one needs to be careful about how to weaken the supervenience claim, but I’m not in principle opposed to the idea that supervenience could be weakened in certain ways, if needed. (Sorry if this is too sketchy!)

About whether one could argue that the higher-order desires could be seen as being part of the causal stream (as Justin was imagining), in some cases, perhaps. But not in all the cases that I think Frankfurt is imagining. I’m reading Frankfurt as claiming that, even when the desire to take the drug is so irresistible that it by itself is sufficient to result in the agent’s taking the drug, the willing addict can be free. There’s no reason to see the higher-order desire as doing causal work in all cases of this kind. It sounds like the higher-order desire could be acting as basically a backup mechanism of some kind, which is ready to step in if needed, but which doesn’t feel the need to, in the actual case. (In this way, the higher-order desire would be like the neuroscientist!)

(By the way, Justin, on your other question: you’re totally right, I believe freedom in fact supervenes on the relevant elements of the actual sequence—those that ground the agent’s freedom. Obviously, it also supervenes on the whole causal history. But you’re right that the stronger claim is the more useful one; I was just using the other formulation to keep things as simple as possible.)


Thanks for the thoughtful post.

There are various different things one could mean by "responsibility", in "responsibility depends only on the actual sequence" or "responsibility supervenes on the actual sequence". One might mean that the "content" or "degree" of responsibility so depends; I do not hold these claims. Rather, I hold that whether someone is morally responsible at all supervenes on the actual sequence. And I do not think the Sharks Cases or any other case (of which I am aware) is a counterexample to this claim. I develop this view, in, among other places, my APA Presidential Address, "The Path of Life".

Hi John,

Thanks a lot for commenting! I’ve been thinking about your interesting proposal in “The Path of Life.” Here are some reactions I had when I read it. You suggest something like the following reformulation of the supervenience principle:

(S*) An agent’s responsibility (whether he is responsible “at all”, in a certain context, not the specific degree or content of his responsibility, in particular, not what he is responsible *for*) supervenes on the actual causal history of his action.

You’re thinking that, even in the Sharks case, the agent is responsible for something, or to some degree, since he is responsible for at least not trying to save the child. Thus Sharks doesn’t pose a problem for S*.

I worry that the new principle S* is too weak, given what we actual-theorists want, and, in particular, I worry that it doesn’t adequately capture the insight provided by Frankfurt cases to the extent that we want. I take the insight provided by Frankfurt cases to be the thought that in determining an agent’s responsibility *for something* all we need to look at is the causal history *of that thing.* For example, if Frank is the agent in a Frankfurt case and the question is whether he’s responsible for a certain choice C, all we need to look at is the causal history of C. Roughly, then, the reasoning is this: given that the neuroscientist is not part of the causal history of C, we can evaluate Frank's responsibility for C as if the neuroscientist had never been there. In other words, the supervenience claim allows us to reason by moving from the scenario where the neuroscientist is not present to the scenario where he is present. Given that Frank is responsible for C in one scenario, he’s also responsible for C in the other scenario (again, because the causal history *of C* is the same in both cases).

Notice that, if we embrace S* but not S, this move is blocked. In fact, for all S* says, Frank (in the Frankfurt scenario) is less responsible, or responsible for fewer things, than Frank in the neuroscientist-free scenario. This is because S* is compatible with the claim that factors that are irrelevant to the causal history of something can have an influence on the agent’s responsibility for that thing. For this reason, I don’t think S* captures what we want, or everything we want. What do you think?

Hi Carolina, thanks for starting off the discussion with such an interesting post. I'm just trying to understand principle S a bit better. Perhaps because of my current love affair with interventionism, I feel like analyzing the "actual causal history of X" necessarily includes some counterfactual analysis of relevant features of that history, specifically the causal powers or capacities of those features (and perhaps also the opportunity of those powers to be exercised, though this seems to be precisely what you're trying to avoid). Will you be suggesting an analysis of the relevant causal powers that is 'actual sequence' in the sense that it does not use counterfactual analyses at all?

Just to put a bit of flesh on the bones of this question, suppose we're looking at the willing addict and trying to figure out if his second-order desire to use the drug (or his 'deep self' or his identifying with his addiction) played an actual causal role in his action of using. Won't we need to consider some counterfactuals (sorry if this sounds like modeling)--e.g., holding fixed the causal influence of his first-order addictive desires, what would happen to his decision to use if we varied the existence (or strength) of his second-order desire (or identification)? Or if we are trying to discern if an agent's action was caused by reasons-responsive mechanisms, don't we have to test the responsiveness of the mechanisms by considering counterfactuals?


Thanks for your response. That seems right.

Your remarks about the supervenience claim not working for responsibility because responsibility has an epistemic component got me thinking. Freedom seemingly has an epistemic dimension too. In many cases of unintentional action, for example, we often do something without having any idea that we did it. In such cases, it seems clear that we did not perform the relevant action freely, and this seems to be partly due to our impoverished epistemic situation. So if the supervenience claim doesn't work in the case of responsibility because of the epistemic dimension of responsibility, won't similar problems arise in the case of freedom?

Eddy and Justin,

Great questions!

I take the view that freedom supervenes on actual causal histories to be consistent with the claim that certain counterfactual considerations can be relevant to freedom. To the extent that certain counterfactuals can be relevant to what the causal history *is*, they can be relevant to freedom, and in a way that is purely consistent with the supervenience claim.

Consider, as an example, the Sharks case again. There, I argue, the agent’s failure to jump into the water to attempt a rescue doesn’t causally result in the child’s death. Thus the causal history of the death doesn’t include the agent’s failure to jump in. That’s the difference, I think, between the Sharks scenario and the scenario without the sharks: simply that the causal history of the death includes the agent’s failure to jump in in one case but not in the other. Now, the fact that, had the agent jumped in, the sharks would have attacked him (and thus the child would still have died) certainly plays an important role in accounting for the agent’s lack of responsibility in the Sharks case. It does this, however, by means of playing an important role in determining what the causal history is (or is not) in that case. The truth of that counterfactual is at least part of the reason why the agent’s behavior doesn’t result in the child’s death in that scenario. In this way, the role played by the counterfactual fact is consistent with the supervenience claim.

These are tricky issues! Consider the example discussed by F&R that Mele picks up on in his paper “Moral responsibility for actions: epistemic and freedom conditions.” A man backs out of his garage and unintentionally drives over a kitten, killing it. Does he freely kill the kitten? As Mele points out, it sounds a bit odd to say that he did. Compare this case to the case of someone who did it knowingly. Again, for these scenarios to constitute a potential counterexample to supervenience, we should imagine that he did it *knowing* that he’d be killing the kitten, but not *moved by* an intention to kill him (otherwise there would be a difference in the causal chain between the two scenarios). Assuming there is a difference in freedom between the two scenarios, then, supervenience seems to fail for freedom too.

What should an actual-sequence theorist say about this? I’m really not sure. One possibility is to say that, as far as the purely metaphysical component of responsibility goes, the agent does have all it takes in both cases. So he freely kills the kitten, in the relevant sense, in both cases. Of course, he’s not responsible in both cases, due to responsibility’s epistemic component. Perhaps this is the right thing to say (despite the fact that it seems a bit odd to say that the agent freely kills the kitten in the case where he does it unintentionally). Another possibility that I consider in the chapter where I discuss this is to distinguish an epistemic from a non-epistemic component within freedom itself, and to claim that supervenience holds for the non-epistemic component only. This would be the right thing to say, in particular, if one were moved by Mele’s considerations when he argues that the distinction between the epistemic and freedom conditions of responsibility is much murkier than people think. If there were no clear distinction between the two kinds of conditions, then I think that, indeed, this would be the right move to make.

Hi Carolina,

This is a really interesting post. It is relevant to what I am working. Thanks for it.

First of all, I have concern about the claim that Frankfurt-style cases motivate an actual-sequence approach to free will or responsibility. But for sake of argument, let's assume it to be so.

My question is whether your response to Eddy will undermine the asymmetry between non-Frankfurt-style cases such as Sharks and Frankfurt-style cases. By saying that "the causal history of the death doesn’t include the agent’s failure to jump in" due to the presence of sharks, may I also conclude that the causal history of the death of Smith doesn't include Jones's shooting due to the presense of Black, therefore, Jones is not responsible for the murder. It seems an absurd analogy at first sight because Jones's shooting is obviously a part of the causal history of Smith's death. But to say so seems to assume a strict (what scientists use) sense of causation which is not adopted by you for Sharks. The *obvious* cause of the drowning child's death is the drowning, the struggling, and the suffocation. *Obviously*, the agent's failure to jump in is not part of the causal history. But you claim that the failure to jump in is not part of the causal history, *not obviously but due to the presence of sharks* seems to suggest that you adopt a loose sense of causation (it is more similar to ordinary explanation in contrast to strict scientific explanation. Also note that some early formulations of actual-sequence does not refer to causation but only to explanation). But if it is the case, at least, it is less absurd to think that Jones's shooting isn't part of the causal hisotry of Smith's death as well, due to the presense of Black. What explains Smith's death is the fact that Black left Jones no option of not shooting. Such an explanation would leave the actual shooting out of the picture.

Of course, I still think that Jones's shooting is part of the causal history of Smith's death. But I find it easier to grasp this fact in strict sense of causation. It is more difficult to grasp it in your terms. I think that this difficulty should call for a more detailed formulation of causal theory to defend your intuition that failure to jump in is not part of the causal history.

How do you think about it?

Hi Jiajun,

Thanks for your thoughts! You’re right, I am assuming that there is a concept of causation according to which omissions can, in principle, be causes (so the reason my failure to jump in is not a cause of the child’s death in Sharks is not simply that it’s an omission; it’s rather something having to do with the presence of the sharks). Any actual-sequence view needs to assume something like this (at the very least, it needs to assume that there is some concept of causation, or of something that behaves very much like causation, which applies to omissions). Otherwise we wouldn’t be able to account for the responsibility of agents for/by their omissions.

Also, as you point out, I am relying on the claim that, whereas there is no causation of the relevant kind in Sharks, there is causation of the relevant kind in Frankfurt cases. In fact, this is the general kind of claim that I’m pushing with respect to actual-sequence views: the claim that actual-sequence views rely heavily on certain assumptions about causation and certain features of the causal relation.

Now, you ask, how can there be such a causal difference between Sharks and Frankfurt cases, if in both cases there is a failure of counterfactual dependence? The short answer is: counterfactual dependence is one kind of consideration that (presumably) helps determine whether something is a cause, but it’s not the only one. I have some more specific story to tell about this, but it would take too long to explain here. Luckily, I don’t believe that a full account of this, or of what causation is in general, is needed to address this. For I take it that most people would find it quite intuitive that there is such a causal difference between the two scenarios. (In particular, the causal structure of Frankfurt cases is the structure of a “causal preemption” case, and there tends to be a lot of agreement that the “preemptor” is what actually does the causing in those kinds of cases, *despite* the failure of counterfactual dependence.)

Hey Carolina! Wonderful post!

Just to follow up on Eddy's comment and your reply, there is nothing about your actual sequence view per se that makes it a source view as opposed to a more classical, could-have-done-otherwise view. Is that right? I'm not fishing for your own views at the moment, just trying to understand the consequences of your view. Thus, you might think "Well, the 'alternatives' are pretty abstract, nothing to ground freedom" and decide, for this reason, to be a source theorist. But is there anything in principle against someone thinking otherwise and coming away with a classical theory?


Here's a suggestion I sometimes find appealing: there is an epistemic component to freedom, and it has to do with being sufficiently aware of what one is doing. (I'm using "sufficiently aware" as a placeholder for whatever the "right" epistemic condition for freedom may be). So in the kitten case (which is just the sort of case I had in mind) the agent does not satisfy this condition, since she isn't at all aware of the kitten's presence, and so doesn't freely kill the kitten. But there is also an epistemic component to responsibility that's distinct from the epistemic dimension of freedom, and it has more to do with being sufficiently aware (again, I use this phrase as a placeholder for whatever the actual condition(s) might be) of the moral status of the action at issue. Notice in the kittens case, the agent may well satisfy this latter condition, since she presumably knows that she has at least a prima facie obligation not to kill the kitten. Does something like this strike you as plausible?


Thanks for your responses.

I agree that actual-sequence view needs to assume a more inclusive notion of causation or *of something behaves very much like causation*. I tend to side with the latter one with no specific reason but only a general hunch: the kind of control associated with free will/responsibility is a pretty dense one which makes it harder to be scientifically investigated.

I don't have any least mature thoughts on this idea. Could just be a silly idea. But an imaginary questions concerning this hunch is:

If an alien visits Earth for the first time, could the alien correctly judge a human agent's responsibility for a particular action if very advanced technology is available to the alien to fully investigate the whole actual sequence prior to the action?

Or imagine an antique machine has been digged up by archaeologist. Is it possible for us to be puzzled by its function (and whether it is still a good and functional machine) even if we can have all the scientific information about it? Does a machine, as a control system which helps us deal with the environment, have some essential features beyond scientific or physical investigation? (maybe some social facts would be relevant and needed)

Like I said, it is just a very premature hunch. It only expresses such a concern: even if actual-sequence view is motivated by Frankfurt cases, there is still a gap between actual-sequence and actual causal history. I have less concern about terms like 'explanation' and 'control' although they are obviously less satisfactory in another respect. At least to my ear, causal history sounds too scientific and overly crystal clear. It seems to me that your notion of causation is different from a very strict sense of causation, which makes it possible to avoid my worry (not saying my worry is a legitimate one). I would be very interested in the more detailed formulation of such a notion, maybe in your new book.

Hi Joe,
Indeed, there is nothing in the supervenience claim itself that makes a view that respects that claim a source view. I think it helps to put it in terms of the grounding relation. I understand actual-sequence views as views according to which freedom is exclusively grounded in actual sequences (causal histories) and the grounds of causal histories (any facts that ground those facts about causal histories). So, if someone were to argue that some facts about alternative possibilities ground the relevant facts about causal histories, then their view would be an actual-sequence view *and* an alternative-possibilities view at the same time. This is an interesting result, because the two kinds of views are typically considered to be logical contraries, while I’m suggesting they are not.

Now, as you note, there is an important difference between claiming that facts about causal histories can be grounded in counterfactual facts and claiming that those facts about causal histories are grounded in (sufficiently robust) facts about “alternative possibilities” or the ability to do otherwise. That’s why, in most or all actual cases, actual-sequence views are not alternative-possibilities views.

Yes, your proposal does sound plausible to me. If one wanted to avoid the claim that the agent who killed the kitten unintentionally did it freely, that would be a nice way of doing just that. In that case restricting supervenience to the non-epistemic component of freedom would make a lot of sense. For, again, the mere awareness of what one is doing, or of the kind of act that one is performing, can presumably be relevant to one’s being in control of the situation just by obtaining, without doing any causal work. So thanks for the helpful suggestion!

One problem for epistemic conditions is OCD-type cases like theft. As I understand them, in many instances OCD-theft is compatible with the full acknowledgement by the perpetrator that what one is doing is wrong, but still can't fight the overpowering urge. That's one reason why many insanity laws included something like a test for ability to have done otherwise that was separate from knowing-right-from-wrong tests.

Thanks for the great set of questions Carolina. There's a lot of important and under-theorized issues you’re raising. I am excited to hear your and others' thoughts.

I assume that you are using ‘freedom’ and ‘control’ interchangeably. If this assumption is false, I would be interested to hear how and why you distinguish them.

Here is a reservation about S. First, S is stronger than anything Fischer needs. Fischer is concerned with showing that moral responsibility is compatible with determinism. A crucial step in his defense is to show that an important species of ability (namely ability in the all in sense) is not necessary for moral responsibility. This is important since the consequence argument makes a strong case for concluding that this species of ability is incompatible with determinism. Much of his work is aimed at showing that we can account for our being morally responsible without possessing this ability. However, he never claims nor need he claim that possessing this ability would not increase our control/freedom. His claim is that this ability is not necessary, but, if I understand S right, it entails that this ability is not only not necessary but wouldn’t even enhance our control/freedom. But why think this? Frankfurt cases certainly don’t establish that claim. Perhaps this is why Frankfurt makes his claims about the enhanced control the willing addict exercises.

What Chris Franklin said!

Also, Carolina, you wrote:

"Fischer explicitly endorses the supervenience claim, at least when X is an action (see his “Responsiveness and Moral Responsibility”). However, when he develops his view in terms of reasons-responsiveness, he suggests that the difference between an agent who is reasons-responsive (and thus free) and one who isn’t can be purely due to the dispositional properties of the mechanism that operates in each case. Now, this means that the mechanism itself (the sequence of events that is actually causally operative) could in principle be exactly the same in both cases. If so, S would fail."

Hmm. These are very delicate and tricky issues. But I wouldn't have thought to identify "the mechanism itself" with "the sequence of events that is actually causally operative", at least on some interpretations of this. In my view, mechanisms are defined modally or dispositionally, so two causal sequences that have [in one sense] the same set of events could be manifestations of--or involve--different mechanisms, insofar as mechanisms are defined dispositionally.

Stepping back, in philosophy sometimes we can lose the forest for the trees. And sometimes it can seem that there is a bigger disagreement than in fact there is. So correct me if I'm wrong, but I understand that we are in basic agreement about the following: moral responsibility does not require freedom to do otherwise (in the "all-in" sense of can, as dubbed by J.L. Austin), and in this sense what matters for moral responsibility is the actual sequence. Further, we take it that the Frankfurt Cases provide support this this sort of actual-sequence approach. Addiitonally, we both are compatibilists about causal determinism and moral responsibility.

Is this correct? Of course, there can be disagreements still--disagreements about how best to explain and defend the broad positions we both hold. But it is important not to lose sight of the agreements as well. And I should add that I'm delighted that you are helping to explicate, defend, and develop an actual-sequence approach to moral responsibility with great rigor and sophistication. I especially like the way that "grounding" plays an important role in your work--thus connecting Semicompatibilism to new developments in metaphysics.

I’m afraid I don’t follow. The epistemic condition would still only be a necessary condition, not a sufficient one. So the agent you’re describing could still fail to be free, even if he meets that condition.

Thanks for joining the conversation!
Hmm… I would have thought that Frankfurt cases do precisely that: motivate the idea that the ability to do otherwise is *completely* irrelevant to freedom (=control) or responsibility. What am I missing? Why shouldn’t one think this?

Here are a couple of quotes from Frankfurt, which I think suggest he thinks this too:

“Even though a person is subject to a coercive force that precludes his performing any action but one, he may nonetheless bear *full* moral responsibility for performing that action.” (“Alternate Possibilities and Moral Responsibility”, end of section III; my emphasis)

“In that case, it seems clear, Jones4 will bear *precisely the same* moral responsibility for what he does as he would have borne if Black had not been ready to take steps to ensure that he do it.” (Section IV; my emphasis)

And Fischer, too, oftentimes talks in terms of what matters to responsibility, or what’s important for responsibility, being just actual sequences. It seems to me that the most natural interpretation of these claims is one according to which alternative responsibilities are just irrelevant.

I'm sorry Carolina--I'm not disagreeing. My point was that an additional test of ability to do the right in effect argued that knowledge that one was in fact doing evil was not sufficient for holding someone OCD responsible. Just offering a real-life example of how things like this have been handled.


I do not think freedom to do otherwise is *completely irrelevant to responsibility"! Perhaps some of my statements were insufficiently careful. Maybe the "natural interpretation" of some of my statements suggests more than what I really meant, or, upon reflection, should mean!

But after all I DID introduce the "Sharks" case, and I do understand that the "content" of moral responsibility--what the agent is morally responsible for--may well depend on freedom to do otherwise. In omissions cases such as Sharks this is particularly clear. But it does not follow that moral responsibility requires freedom to do otherwise (in the all-in sense); that is, being morally responsible at all (as opposed to the content or degree or responsibility/blameworthiness) does not depend on freedom to do otherwise (so construed).

I cannot speak for Harry Frankfurt, but I can report that he did write to me saying that my Presidential Address "hit the nail on the head" (or something to that effect). [Of course, perhaps he did not mean to endorse every part of my paper there.]

Thanks for helping to clarify these nuanced issues, and for affording me the opportunity to be a bit clearer than I have been in the past.

Chris & Carolina: Chris alludes to the idea that even if the ability to do otherwise isn't required for freedom/control, it might enhance it. Carolina, you respond, in part, by quoting Frankfurt to the effect that the power to do otherwise is completely irrelevant to MR. Chris, did you mean your claim to apply to MR too? For notice that your statement was about freedom/control, whereas the quotations Carolina provides are about MR. This leaves open the possibility that you are both right: if the ability to do otherwise is not necessary for FW or MR, it might nevertheless enhance FW but not MR. That said, since I'm in print defending the idea that if (big if, by the way) the power to do otherwise isn't required for MR, it might still enhance MR, let me say something to try to motivate that idea. I've got a much longer story about all this, but the most simple version is this: agents who have regulative control over their actions in addition to exercising guidance control over them make a difference to the world in a way that agents who only have guidance control don't. This additional "difference making" then grounds enhanced responsibility. (Carolina, if there are different ways to make a difference, this last claim of mine is consistent with your claim [in your nice Phil. Review piece] to the effect that there is a kind of difference making that is compatible with determinism, right?)

John: I'm curious. Why say what you say about the sharks case, but not about regular FSCs? Please correct me if I'm wrong, but your view is that in the "Sharks" case the agent is just as responsible (e.g., just as blameworthy) as she would have been in the absence of the sharks; it's just that the agent is blameworthy for something different when the sharks are present (the content of her responsibility is different). When the sharks are there, the agent is responsible for not trying to rescue the child, but if they had been absent the agent would have been responsible for not trying to rescue the child and for failing to save the child. Is that right? If so, why not say the same thing about the FSCs? That is, why not say that the agent is just as responsible in a version of the story in which Black is present as he would have been in Black's absence, but that the "content" of his responsibility differs? In the FSC the agent is responsible for "deciding on his own," whereas in the normal case in which he could have done otherwise he is responsible for the decision itself.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan