A sociologist friend told me a few months ago that she had finally read my book Justice, and that it was the first time that she had encountered Judith Jarvis Thomson’s violinist case (which she thought was pretty neat). The full text of the article is here; the wikipedia entry is here. The discussion reminded me that I keep intending to post something about thought experiments in general, and Thomson’s article in particular, partly to defend thought experiments against JQ’s scepticism. I haven’t read a great deal of the secondary literature about the violinist, but I have taught it numerous times, and discussed it extensively with colleagues and students; what follows is just my take, in the light of those discussions.
First of all, it should be obvious that the violinist case does not establish the permissibility of abortion even in the case of rape. In fact, I would say that the focus on the permissibility of abortion (which Thomson encourages, not least by her title) is a bit misleading. Every semester a very small number of my students say that they do not think it is permissible to unplug oneself from the violinist. Not a single sentence in the article speaks to them: they can get off before even getting on (interestingly, every now and then a student (usually they turn out to be some sort of lefty) who thinks there is a right to abortion thinks it is impermissible to unplug oneself, because one has extensive and stringent duties to aid stranger in need). So what does the example establish?
It seems to me that it establishes two things. First consider a combination of views that is widespread among American conservatives (and that Thomson must have encountered a thousand times): that there is no right to abortion, and that we do not have extensive and demanding enforceable obligations to strangers. People who hold that combination of views ought to be deeply discomforted by the violinist example (and by the people seeds example that follows). Neither example PROVES that this combination is inconsistent, but both examples press the person who adheres to it to give us more explanation of the differences between the cases. The second thing that the example establishes is that for many opponents of abortion (all those who say it is ok to unplug oneself from the violinist) the right to life of the fetus plays only a small background role in their justification of abortion’s impermissibility; of course, if the fetus did not have a right to life then abortion would be permissible, but these people believe that it is permissible to kill innocent beings which have a right to life in circumstances very like the circumstances of the fetus growing inside the pregnant mother. They need to offer an explanation of the difference.
One thing students come up with to deal with both problems is that the fetus is genetically descended from the mother: but it is far from clear why this makes a difference. Imagine a world in which fetuses are not genetically descended from the people in which they are lodged. Would this really make a difference to the permissibility of abortion?
Suppose that the reader decides that the combination of views is inconsistent. As I say, the paper says nothing to them about which of the views they should reject. Thomson clearly thinks we should reject the idea that we have extensive and demanding enforceable obligations to strangers, but she provides no argument to that effect, except to show that, in general, people do not believe that we have such obligations.
So what does this tell us about what thought experiments do in moral philosophy? They never tell us how to act, even in the highly stylised and unrealistic cases that they describe. Nor do they tell us whether to be consequentialists, deontologists, or whatever (I think that distinction is trivially exhaustive because deontologist means something like “not consequentialist”). What they do is to trigger our awareness of conflicts between judgments that we previously held in combination. They alert us, in other words, to inconsistency, and prompt further argument and search for principles that do more substantive work. My sense of JQ’s scepticism about thought experiments is that he sees them as intended to be conversation stoppers. Far from it: they are intended to (or should be intended to) open up new conversations.
JQ makes one other objection, which I’ve puzzled about for the several months since I showed him the draft of this: that these examples tend to be unnecessarily blood-thirsty. I think there are two reasons for this. The first may not be so good: I think it is thought that in cases where the stakes are high it is easier for respondents to see that there is a real need to make a judgment and judgment is, somehow, sharpened. As I say, this may not be so good a reason: it might, as I think critics of the method suspect, entrench intuitive (in the sense of gut, not in the standard philosophical sense) reactions. The second, though, is that historically, thought experiments in ethics were designed to scrutinize principles in ethics that were designed precisely to justify killing of various kinds; so the bloodthirstiness was really a property of the principles of interest.