There are analogies, and there are analogies. Some analogies feature prominently in analogical inference, while others merely function as illustrations of general principles. In these latter cases, the term “analogy” is used loosely, but the former type of case involves analogy in a stricter sense. In analogical inference, or arguing by analogy, analogies function instead as premises in arguments.
Judith Jarvis Thomson’s violinist example in “A Defense of Abortion”* is rightly famous, but its role in the argument(s) of the paper has been misunderstood. Philosophers sometimes present the example as the centerpiece of an argument by analogy. It is better understood as a counterexample to an overly-broad principle assumed by anti-abortion philosophers and their opponents alike. It is an analogy in the loose sense, but not the strict sense.
Let’s set the stage. Thomson argues against the following anti-abortion argument:
- A fetus is a person from the moment of conception.
- Every person has a right to life.
- So the fetus has a right to life.
- Killing an innocent person violates their right to life.
- Abortion kills an innocent person.
- So abortion violates the fetus’s right to life.
This could be cleaned up but will suffice for our purposes, since the argument’s premises entail the conclusion, and this presentation highlights the details relevant to us. Thomson points out, rightly, that arguments over abortion often focus on the first premise. Billboards in my part of Connecticut declare the personhood of fetuses by asserting their possession of organs, smiles, and so on; so-called heartbeat bills aim for the same result. While Thomson denies that fetuses are persons with rights, she says that defenders of abortion have focused too much on defending their rejection of fetal personhood. That aspect of the debate has generated more heat than light. For this reason, Thomson grants that fetuses are persons for the sake of argument. She also grants – because she believes – that every person does have a right to life, when she says that “the primary control we must place on the acceptability of an account of rights is that it should turn out in that account to be a truth that all persons have a right to life” (p. 56).
In a defense of abortion, then, we should not ask about fetal personhood or about whether people have a right to life. We should ask instead what that right to life guarantees. We should focus on the fourth premise in the argument above. The violinist thought experiment is meant to disprove the principle there by counterexample.
How would that work? Here’s the example (pp. 48-49):
You wake up in the morning and find yourself back to back in bed with an unconscious violinist. A famous unconscious violinist. He has been found to have a fatal kidney ailment, and the Society of Music Lovers has canvassed all the available medical records and found that you alone have the right blood type to help. They have therefore kidnapped you, and last night the violinist’s circulatory system was plugged into yours, so that your kidneys can be used to extract poisons from his blood as well as your own. The director of the hospital now tells you, “Look, we’re sorry the Society of Music Lovers did this to you – we would never have permitted it if we had known. But still, they did it, and the violinist now is plugged into you. To unplug you would be to kill him. But never mind, it’s only for nine months. By then he will have recovered from his ailment, and can safely be unplugged from you.” Is it morally incumbent on you to accede to this situation?
The answer seems to be “no” – you can unplug with a clean conscience. It is not wrong to unplug from and thereby kill the violinist. Some people do not agree with this judgment, holding that it is indeed wrong to unplug, which makes it inconvenient for Thomson to lean so heavily on intuition. Her argument only goes through on the assumption that you agree with her on this basic datum: it is not wrong to unplug from the violinist.**
Here, then, is the first, immediate point about the violinist: despite the fact that he is innocent and unambiguously a person with a right to life, unplugging from him and killing him is not wrong. But if killing him under these circumstances is not wrong, then his right to life must not have been violated. After all, if killing him did violate his right to life, we would expect there to be at least a hint of wrongness. Yet there isn’t any such hint. This implies that, whatever the right to life protects, it does not protect the violinist in these circumstances.
Where to go from here? If all the claims of the previous paragraph are right, then we have a good case for rejecting the view that killing an innocent person violates their right to life. The violinist is innocent, he has a right to life, and yet it is not violated. The fourth premise in the argument is false, so the argument as a whole fails. Job done – the argument fails because one of its major premises has a counterexample. Ultimately, then, Thomson’s argument is simple:
- If it always violates an innocent person’s right to life to kill them, then it violates the violinist’s right to life to unplug.
- If it violates the violinist’s right to life to unplug, then it is wrong to unplug.
- It is not wrong to unplug.
- So it does not violate the violinist’s right to life to unplug.
- So it does not always violate an innocent person’s right to life to kill them.
If Thomson’s argument is logically sound, then the anti-abortion argument above is logically valid, but not sound, since the fourth premise is false. For my money, Thomson’s right, and her argument succeeds, but here I only want to highlight the logical character of her argument. Intuitions about the violinist example function here as evidence for the third premise of her argument. Once you accept her premises, the conclusion follows deductively.
Some philosophers understand the dialectic differently. They interpret the violinist example as an argument by analogy. The argument would work like this:
The violinist example shows us a case where killing is permitted. There are a number of features of the violinist case which explain why the killing is permitted. All those features are present, in relevantly similar ways, in pregnancy, in that a pregnant person is hooked up to the fetus inside them in basically the same way as you’re hooked up to the violinist. So if you can unplug and kill the violinist under those circumstances, then in relevantly similar circumstances, people can terminate pregnancies by “unplugging” from the fetuses they carry.
This argument has a different structure than the one just given. Thomson gives an argument by counterexample, with the violinist providing the counterexample; if we understand the argument as an argument by analogy, the violinist plays a very different role.
To give an argument by analogy, we start with two things: a domain of objects which we understand reasonably well, and another, more obscure domain which we want to model by means of the first. So suppose I have two drugs, X and Y. We know a lot about X, but less about Y. We know that they’re similar in many ways – X and Y both have a sedative effect – so maybe they’re similar in ways we haven’t yet observed. If X has a narcotic effect, then maybe, knowing that X and Y are so similar in other ways, Y also has a narcotic effect.*** It’s a risky, but rational kind of inference.
Part of the risk here comes from the nature of the inference. The logical structure of an argument by counterexample is deductive: the argument is valid when the truth of the premises guarantees the truth of the conclusion. The structure of an argument by analogy is not deductive, since the inference could, though rational, lead you astray.
To put the point another way, consider how the arguments differ in their stress points, in how they can break down. A deductive argument stands or falls on whether (i) the premises logically entail the conclusion and (ii) the premises are actually true. An argument by analogy involves truth, but crucially requires a good enough model to support the analogical inference, and whether a model is “good enough” isn’t as straightforward as asking whether it’s true. (No analogy involves perfect one-to-one correspondence between the compared domains; if it did, the analogy would be pointless.) Modeling a domain poorly can badly throw off inquiry. For instance, if you assume that combustion involves the loss of something rather than a gain, you might end up accepting the phlogiston theory, which gets the direction of causation backwards. (A burning log oxidizes, gaining oxygen atoms, the same underlying chemical process as iron rusting.)
So right away, if you read Thomson as offering an argument by analogy, even if you’re a sympathetic reader, you will point directly at the violinist example’s failure to act as a proper model for pregnancy and abortion. This is, in fact, what nearly all of my students have done. They note instantly that not every pregnancy occurs under kidnapping-like circumstances. Some pregnancies are intended; some are unintended but are the result of consensual activity. And what if you were actually a volunteer from the Society for Music Lovers? And so on.
These are all fine questions to ask, but they are dialectically beside the point. If Thomson were making an argument by analogy, here is how it would look:
- The scenario described in the violinist example is similar, in all relevant ways, to pregnancy.
- If it is not wrong to unplug from the violinist, then it is not wrong to terminate a pregnancy.
- It is not wrong to unplug from the violinist.
- Therefore, it is not wrong to terminate a pregnancy.
Again, we can only support the third premise with an appeal to intuition – not great in my opinion, but not important here. To defend the first two premises of this argument, though, we would need to spell out the relevant details shared by the violinist example and pregnancy. It’s in this instance that we call upon the violinist example to be the centerpiece of an argument by analogy, a model that can show us the morally relevant details that explain how to go from the permissibility of unplugging to the permisssibility of abortion. And this is where the flaws in the analogy – if it is an analogy – start to suggest themselves. Since an argument by analogy involves proper analogical inference, Thomson would need the model to be good enough to support her inference from unplugging permissibly to terminating a pregnancy permissibly.
In some ways, Thomson does herself no favors here. At her best, she is a stylish writer, and “Defense” is very stylish, with quick, precise volleys between point and counterpoint. But the speed and natural conversational flow invite interpretive hiccups right when she finishes the violinist case. The very next paragraph begins, “In this case, of course, you were kidnapped, you didn’t volunteer for the operation that plugged the violinist into your kidneys” (p. 49). Already we seem to be off to the races, quibbling about the details of the case. But Thomson only brings this up to make trouble for an anti-abortion position which makes an exception for rape: “Surely the question of whether you have a right to life at all, or how much of it you have, shouldn’t turn on the question of whether or not you are a product of a rape” (ibid.).
This can render the nature of the argument unclear to the reader, but it does not make the argument an argument by analogy. It’s unfortunate writing, but not bad reasoning. (The worst of it is that she describes her examples as analogies in the seventh section of her paper (p. 64). This is an inapt description on her part. She’s using the word “analogy” loosely, not in the stricter sense outlined above.) The violinist example itself involves all sorts of incidental features – Nine months! Sharing blood! Kidnapped! – but once Thomson starts changing the details of the story, that’s a clue that it isn’t meant to be an analogy.
Case in point: later in the paper (pp. 59-60), she imagines that the violinist only needs to be hooked up to you for an hour to be completely cured. Your health won’t be impacted at all; you just lose an hour of your time. And she also asks us to consider what we would think about pregnancies if they only lasted an hour. Perhaps we might think it was indecent to refuse the violinist our body, just for an hour.
Even so, Thomson says, the fact that it would be indecent to refuse does not show that refusal would violate the violinist’s right to life. All kinds of things can be indecent or mean without being unfair or unjust. Consider another of Thomson’s examples. A boy receives a box of chocolates as a gift; they are his chocolates and his alone. But his little brother would really like one of the chocolates. It would be mean, maybe even wrong, but for the big brother not to share, even if the little brother isn’t owed anything. The little brother has no right to the chocolate, but it can still be wrong to refuse him a piece, the point being that a wrong is not necessarily a violation of a right.
This brings us back to Thomson altering the details of the violinist case, shortening the ordeal to one hour. Even with this change, nothing fundamentally alters the moral situation with respect to rights. Our judgment might change because less is being asked of you, and so it’s worse to refuse. But the specifics of the rights in question have not altered. (Again, we can appeal to the boy with the chocolates. Let’s say he was gifted twenty boxes, and his little brother wants nought but a nugget. It’s even worse to refuse in this case, but our answer to the basic question of justice – does the little brother have a real claim on a piece? – hasn’t changed.)
This should be a major clue that the violinist case does not function as an analogy, especially when it comes to its superficial similarities with pregnancy (nine months, shared blood, etc.). If those details were so important, why would Thomson later change them in the paper without any corresponding change in intuitions about the cases? If the length of the ordeal was a morally relevant similarity between the violinist case and pregnancy, then altering the length should alter our conclusions with respect to the rights under question. But our intuitions about the rights don’t change at all. And this is not surprising, because the violinist isn’t functioning as an analogy. It’s functioning as a counterexample to the principle that killing an innocent person violates their right to life. That is how most of Thomson’s examples in this piece function. They poke holes in the principles offered by anti-abortion philosophers in defense of their views. Many are not meant to be realistic analogies to pregnancy or pregnancy-adjacent phenomena – some are fanciful in the characteristic Thomson style – because they are only meant to be cases that break the principles on offer.
This is part of why I resist the interpretation of Thomson’s article as an argument by analogy. It weakens the argument, because then we’re left with haggling over how our intuitions change when we tweak the model just so, whether the model is right after all, whether the relationship between parent and fetus can be appropriately compared with that between grown adults conjoined by Dr. Frankenstein, etc. If the violinist is simply a counterexample, then you’re left with a few ways to respond. You can (a) accept the case as an example of permissible killing, (b) deny the datum by saying you should not unplug, or perhaps (c) accept that unplugging is permissible, but deny that it’s a killing (say, arguing that unplugging allows the violinist to die, but does not kill him). But none of these involve deciding whether the analogy is apt, or supports an analogical inference, or whatever. The argument needs no further commitments around intuiting which similarities are relevant in the analogy, because the case doesn’t have to do that job. To assume that it does is to assume a pointless argumentative burden.
In a way, the analogy reading also gives in to the tendency to misrepresent the article. The violinist is memorable, but far from the only interesting or significant example in the paper. The violinist has taken up all the air in the room, and as a result, “A Defense of Abortion” is largely remembered as playing purely negative defense, simply offering counterarguments to anti-abortion arguments. (Sometimes she’s read as arguing that bodily autonomy rights outweigh the right to life, but as far as I can tell, this isn’t supported by the text.) But Thomson also gives positive proposals on how much morality can demand of us, and whether those demands carry over to pregnancy. It’s a subtle series of arguments, and we owe it to future readers and students — and Thomson — to present them accurately.****
* Page numbers refer to the original publication, which was in the first ever (!) volume of Philosophy & Public Affairs.
** In my experience of teaching the article, the vast majority of students agree with Thomson here. This is only an interesting sociological detail, though, not evidence for or against Thomson’s datum.
*** This example comes from the Stanford Encyclopedia of Philosophy’s article on analogical reasoning. There are other great examples at the article.
**** Thanks to Hanna Gunn, Nathan Kellen, and Toby Napoletano for talking about “Defense” with me and reading an early draft of this.