This example comes from Alonzo Fyfe’s book, A Better Place. A doctor has the opportunity to save the lives of 5 terminally ill people by killing a healthy person, harvesting his organs, and transplanting them into the sick people, thereby saving their lives. The doctor is willing to take either action, but wishes to know which one she should perform.
The only circumstance under which the healthy person should die in order to save the five ill people is if he voluntarily gives up his life, and all five ill people consent to his death.
Available actions: The only two courses of action presented to us in this problem are to murder the healthy person or to do nothing.
Relevant beings: The action utilities of the sick patients and the healthy person are all contingent on at least one available action, so these are the relevant people. We will assume that the doctor’s happiness is not significantly dependent on which action is chosen: she is therefore not a relevant being.
Before launching into the calculus, it behooves us to consider consent at this point. We do not need to consider the consent of the doctor, because she is not a relevant being. The healthy person, we can safely assume, would not consent to being murdered for his organs. The view of most terminally ill people would, I think, be similar: even those who are at death’s door would probably not agree to treatments that involve the murder of innocent people. This highlights the unrealistic nature of this problem.
If the five terminally ill people object to receiving organs from a person murdered for that purpose, then our problem is solved: all relevant people object to the murder, so the only remaining action is to do nothing.
What if one of the five ill people actually wanted the doctor to kill the healthy person? Then we would have an objection to the second action (doing nothing) too. In this case, the Second Rule of Consent requires us to try to persuade the relevant people that the action preferred by consensual utilitarianism is the right one to take. So, what is the action preferred by consensual utilitarianism? Let’s follow the calculus and see.
Relevant time periods: For the healthy person, the relevant time period is the remainder of his life under the second action (not murdering the healthy person). The relevant time period for each terminally ill person is the duration of her life under the first action (receiving a transplanted organ from the murdered healthy person). Assuming that the transplants are successful, we can conclude that the relevant time period for each relevant person is approximately the same: the remainder of a healthy life beginning at the present moment (I will assume that all the relevant people are of approximately the same age).
Healthy person: If the healthy person is murdered, his CI (contentment index) will drop to zero, and remain that way. His action utility (CI integrated over the relevant time period) will therefore be 0. If he is not murdered, his CI will follow whatever course can be expected for a person of average happiness, so it would have some positive average value. Integrated over the relevant time period, this will produce some positive action utility A.
Ill person (I will assume that the CIs of all five ill people are roughly the same): If an ill person does not receive a transplant, she will die, and her CI will drop to zero. Her action utility will therefore be 0. If she receives a transplant, and makes a full recovery, then she will have a similar CI to the healthy person when he is allowed to live. She will also live a similar length of time, and so her action utility will be approximately A.
To summarize: If the healthy person is murdered, we will have action utilities of 0, A, A, A, A, A, for the healthy person and the five ill people, respectively. If the healthy person is not murdered, we will have action utilities of A, 0, 0, 0, 0, 0.
Both actions have the same minimum action utility (0) and so, according to our calculus, we should rank the actions according to their total action utilities. If the healthy person is murdered, then the total action utility will be 5A. If the healthy person is not murdered, the total action utility will just be A. The preferred action, then, appears to be the murder of the healthy person.
This result seems counterintuitive, if not plainly immoral. How can we recommend that an innocent person be murdered? This is not, in fact, the recommendation that consensual utilitarianism makes. To see this, we need to look at what the Second Rule of Consent requires us to do when all actions raise objections among the relevant people. The Second Rule requires us to make a case to the relevant people for the preferred action. But this action can only be carried out if all relevant people agree to it. Therefore, the death of the healthy person is permitted only if he agrees to it and if all five patients agree to it. And importantly, if the healthy person agrees to be put to death, he is not volunteering to be murdered: murder, by definition, is an act that is performed in against the will of the victim. Instead, he is giving up his life voluntarily. I think most people would agree that if a person voluntarily gave up his life to save the lives of five terminally ill people, this would be a very generous (if tragic) sacrifice, not a deeply immoral act.
The Second Rule of Consent, therefore, simply requires us to make the case to the healthy person that sacrificing his life would produce a great deal of good, more so than if he did not sacrifice his life. Of course, even if the healthy person understood this argument, it’s unlikely that he’d change his mind and give up his life. So be it. Consensual utilitarianism would have nothing more to say on the matter at this point, and the life of the healthy person would be safe.