Many thanks to all for your thoughtful comments on the first post! It was encouraging to see so much interest in these topics, and the discussion has been very helpful to me as I continue to develop my view. As I anticipated in my first post, in this second post I’ll discuss some potential challenges to the supervenience claim S, the claim that I take to be central to actual-sequence views.
As most readers of this blog will recall, van Inwagen offered two kinds of examples in support of the alternative-possibilities view—examples that involve omissions and consequences instead of actions. His example involving omissions has the same structure as the Sharks scenario we discussed in the first post, so I’ll just use the Sharks example instead. Van Inwagen’s claim is that examples like Sharks suggest that responsibility is grounded in alternative possibilities. When there are no sharks, and thus the agent is able to save the child, his responsibility for failing to save him is grounded in that ability. This is suggested by the fact that, if there had been sharks, he wouldn’t have been responsible for his failure to save the child. (Call the two cases Sharks and No-Sharks.)
Van Inwagen’s second example, the one involving consequences, is this:
Ryder loses control of his horse, Dobbin, when he approaches a crossroads. He can’t stop Dobbin, but he can steer him in different directions. He has reason to believe that only one path leads to Rome. Since he despises Romans and wants them to get hurt, he steers the horse in that direction (knowing that they’ll be harmed by the runaway horse).
Imagine that Ryder was right and only that path led to Rome; then he’s responsible for the Romans being harmed. Again, van Inwagen claims that Ryder’s responsibility is grounded in his ability to bring about a different consequence. For imagine that, unbeknownst to Ryder, all paths led to Rome and thus the harm to the Romans was inevitable. In that case, van Inwagen claims, Ryder isn’t responsible for the harm. (Call the two cases One Path and Many Paths.)
Besides providing some support for the alternative-possibilities view, van Inwagen’s examples work as potential counterexamples to the supervenience claim S—the claim that freedom supervenes on actual sequences. (This is not a coincidence. Rather, this is because van Inwagen’s examples suggest that factors external to the actual sequence, those in virtue of which the agent lacks the ability to do otherwise, can get the agent off the hook.) Consider, first, Sharks and No-Sharks: the agent’s freedom and responsibility for failing to save the child is different in each case, but it is very natural to think that the actual sequence is the same, since the sharks never intervene when they are present. Similarly for One Path and Many Paths: Ryder is free and responsible for the harm in One Path but not in Many Paths. However, since he steers Dobbin in the same direction in both cases, and everything is exactly the same from that point on, it is very natural to regard the actual sequence as one and the same in both cases.
As I explained in the comments thread of my first post, my strategy to rescue supervenience from these apparent counterexamples is to argue that, despite initial appearances, the causal histories are not the same in these cases. My failure to jump into the water doesn’t result in my failure to save the child (or in the child’s death) in Sharks, but it does in No-Sharks. Similarly, I suggest that Ryder’s steering the horse in the relevant direction doesn’t causally result in the harm in Many Paths, but it does in One Path.
Now, there are two different strategies one could use to support these claims. First, one could use a direct strategy: one could directly appeal to some causal intuitions, or to some theory of causation, or to some specific features of the causal relation, in support of those claims. (This is the type of strategy I used in my “Actuality and Responsibility” paper.) Second, one could use an indirect strategy. This is the strategy I’d like to explore here.
The indirect strategy seeks to establish the same causal results by appeal to certain claims about the connection between causation and responsibility, instead of by appeal to causal considerations per se. Here’s how I implement this strategy:
Note, first, that in all these scenarios we can safely assume that the agent is responsible for something (in Sharks: the failure to try to save the child, or the failure to jump into the water; in Many Paths: the attempt to cause harm, or the steering of the horse in the relevant direction). Next, note that the following is a well-established principle about the relation between causation and responsibility (one that offers sufficient conditions for “derivative” or “inherited” responsibility):
Derivative Responsibility: If an agent is responsible for X, X causes Y, and all the relevant epistemic conditions obtain, then the agent is also responsible for Y.
(Perhaps a non-deviance condition should be added, or perhaps the non-deviance requirement should be regarded as included in the epistemic conditions, as the claim that the agent could foresee that Y would come about in roughly the way it did.) In what follows, I will work with blameworthiness, since this is the kind of responsibility at issue in these cases.
A defense of supervenience can be built on the basis of this principle. Consider, for example, Many Paths. Imagine, for reductio, the following:
(1) Ryder’s steering Dobbin in the relevant direction caused the harm to the Romans.
Now, given that:
(2) Ryder is blameworthy for steering the horse in that direction.
And given that the relevant epistemic conditions obtain, it would follow that:
(3) Ryder is blameworthy for the harm to the Romans.
Which, we are assuming, is false. Hence the reductio is complete: (1) is false. In contrast, Ryder’s steering the horse in One Path is clearly a cause of the harm. Thus there is a difference in the causal history, and supervenience is safe.
In other words: our concept of causation (or, if there is more than one, the one that is relevant in these debates) must be such that Ryder doesn’t cause the harm in Many Paths. Otherwise, given that we think that he’s responsible for attempting to cause the harm, we would also think that he’s responsible for the harm.
My first set of questions for you, readers (if anyone got this far!), is: Does this strike you as a good strategy? Is it more or less successful than the direct strategy (or are they both equally successful/unsuccessful)? Why?
(And for those of you who want to keep reading…) Some further comments:
There are other cases where our intuitions about responsibility are (presumably) less clear. Imagine, for example, that the sharks are not yet in the water but, had I jumped in to save the child, an evil man would have put them there. Or imagine that only one path was open when Ryder steered Dobbin in that direction, but another evil man would have opened the other paths if Ryder had steered Dobbin toward them. I imagine some people would want to say that it’s not so obvious that the agent is not responsible in those cases. (What do people think of these examples? Can you think of other examples of this kind?) In any event, assuming this, I would argue that the underlying causal intuitions are unclear to a similar extent, and that’s why our responsibility intuitions are unclear. What’s clear (drawing, again, on the principle of derivative responsibility) is that, if the relevant causal relation obtains in those cases, the agent is responsible for the relevant outcomes, and otherwise he is not. But this means that these cases are never a threat to supervenience.
A more general lesson can presumably be drawn from all this: There is no good argument against supervenience, or against actual-sequence views and in support of alternative-possibilities views, that appeals to cases where the agent’s responsibility, if it exists, is derivative in this way. For the principle of derivative responsibility guarantees that supervenience is met in all those cases.