The following seems prima facie quite plausible:
- You intend that p in doing A if and only if it is a part of your reason for doing A that p might result from your doing A.
Until just about now, I was thinking that there were decisive counterexamples to this principle. Consider for instance the case of the strategic bomber who does not intend there to be civilian casualties, but chooses the time of her bombing raid in such a way as to maximize the availability of hospital space for the civilian wounded. Then that her action might result in civilian casualties is a part of her reason for performing the action at the time at which she performs it. But the time of an action is a part of what identifies the action. So that her action might result in civilian casualties is a part of her reason for performing the action she does. Yet she plainly does not intend the civilian casualties.
If this example does not convince simply because one thinks that strategic bombers really doe intend civilian casualties, consider a different example. I choose to undergo a simple operation at an excellent hospital, rather than a low-end hospital, because the operation might result in complications that are better treated at the excellent hospital. It seems to be a part of the my reasons for acting as I do—having the operation at that hospital—that the operation may result in complications.
Or so I thought. But now I think I was confused, and these counterexamples to (1) don't work. This doesn't show that (1) holds. But it would be really nice if (1) did hold—we would then have a neat reduction of intentions to reasons.
Here's why I was confused. Take the hospital example, where things are clearer. Let A be the action of asking the surgeon at the excellent hospital to perform the operation. Then the objection to (1) goes as follows: it is a part of my reason for performing A (specifically, the "at the excellent hospital" part of A) that complications may result from A. But I think not.
The problem is related to a mistake that is sometimes made in explanation, which is to make it insufficiently general. Here is a case of that. "Why did Smith eat chocolate ice cream?" "Because the choice was between chocolate and mint, and he slightly preferred chocolate." While we speak in this way, that is in fact not the right way to speak. As Wes Salmon said, irrelevance is harmless in arguments but fatal in explanations. The "he slightly preferred chocolate" implicitly contains an irrelevant claim. That he slightly preferred chocolate is equivalent to the conjunction of two claims: "he at least slightly preferred chocolate" and "he at most slightly preferred chocolate". Of these two claims, the second does nothing for the explanation of why Smith ate chocolate ice cream. It is quite irrelevant. It is only the first claim that enters into the explanation. Thus the correct explanation is that the choice was between chocolate and mint, and he at least slightly preferred chocolate. This explanation leaves out the irrelevant fact that he at most slightly preferred chocolate, a fact that makes the explanandum more puzzling if anything (explanation removes puzzlement).
We can now formulate my reason for choosing to have the operation in the better hospital more precisely. Let's say the less good hospital is the Sextus Hospital and the better is the Harvey Hospital. Then my reason for opting for the Harvey Hospital should be put as follows:
- The chance of complications is no greater at Harvey than at Sextus, and the chance of poorly treated complications is greater at Sextus than at Harvey.
Now go back to the strategic bomber case. The correct way to render the bomber's reason for choosing the time of the raid is:
- The number of civilian casualties at this time is no greater than at the other time, but the number of casualties receiving inadequate hospital treatment is smaller than at the other time.
Now, since in fact it is unlikely that the bomber actually thinks the complex thought (3), we shouldn't identify reasons with actual thinkings of the listed contents.
In an earlier post I said that intentions supervene on reasons, and that an important task is to give an account of how reasons determine intentions. I currently have no counterexample to the claim that (1) is such an account. But I am also very suspicious in general of attempts at necessary and sufficient conditions, so I am sceptical that I've succeeded in (1).
It occurs to me that Kamm's triple effect cases might be counterexamples to (1), depending on how one understands reasons for action. I read Kamm's cases as cases of defeater-defeaters. If we don't count a defeater to a defeater as a reason-in-favor, then (1) survives Kamm's cases.