From Theory to Cases

A patient with a few months to live has not been told. The family has asked the doctor not to tell them, on the grounds that the news will destroy what time is left. The patient has not asked directly and has not asked not to be told. A decision has to be made this afternoon.

Someone who has worked through a course in moral theory can now state three positions on what makes actions right. None of the three hands over an answer. Each of them will do something to the question, and what each does is different, and knowing which one is correct would settle less than it looks as though it should.

This tier is about the gap between having a theory and knowing what to do. The gap is real, it is not a defect in anyone's education, and most of the interesting work in moral philosophy over the last fifty years has been done inside it.

The subject and its raw material

The word applied in the name of the subject is doing a lot of work and most of it is misleading. It suggests that the theories were worked out first and are now being brought to bear on cases, in the way that a structural engineer brings mechanics to bear on a bridge. That picture has a name in the literature, the application model, and almost nobody now defends it in the strong form.

Three routes from a question to a verdict

The theories are not competing answers to a single question so much as different machinery for converting a situation into a judgement, and watching them run on the same case shows what each demands.

Consequentialism asks what the available options are, what would follow from each, and which set of outcomes is best by the standard of value the theory adopts. On the case above it requires an estimate: whether this patient, told, would spend the remaining months in terror or would use them, and whether the family's account of what the patient can bear is reliable or is about what the family can bear. The theory converts a moral question into a prediction, and the prediction is one that nobody in the room can make with confidence.

Deontology asks what the act would be, under a description, and whether an act of that kind is permitted. On the case above everything turns on the description. Withholding a diagnosis when the patient has not asked may be a failure to volunteer information, or it may be a deception, and it becomes a lie outright the moment the patient asks a direct question. The theory converts a moral question into a question about which description is the right one, and it does not supply the answer to that.

Virtue ethics asks what a person of good character would do here, and what the available actions would express. On the case above it directs attention to things the other two skip: whether the doctor is protecting the patient or avoiding a conversation they find unbearable, and what a courageous and compassionate practitioner would say. The theory converts a moral question into a question about a person, and different virtues point in different directions, since honesty and compassion are both virtues and here they pull apart.

Each route is genuinely different and each ends short of a verdict. That is not because the theories are badly formulated. It is because each requires an input that the theory itself does not contain: a prediction, a description, or a judgement about which trait is in play.

Why the application model fails

Three problems sink it, and they are independent of each other.

The first is that theories underdetermine cases. As the example shows, a theory tells you what to look for and then requires a further judgement about the particular situation before it delivers anything. That further judgement is where the disagreement usually is, and the theory has no more to say about it than anyone else does.

The second is that the theories are contested. Consequentialism, deontology, and virtue ethics have been argued over for centuries by people who understood the arguments perfectly well, and none has won. A verdict derived from one of them inherits all of that instability. If applied ethics had to wait for the theoretical disagreement to be settled before saying anything, it would have nothing to say about a patient who needs an answer this afternoon.

The third is a point about where confidence actually lies. Almost everyone is more certain that torturing a child for entertainment is wrong than they are of any general principle from which that follows. When a theory and a settled judgement about a case conflict, the reasonable response is very often to abandon the theory. Deduction can run in either direction, and the premises are not always the part we should trust.

The history of the field records the retreat. When Tom Beauchamp and James Childress published Principles of Biomedical Ethics in 1979, the first edition presented something that looked like a hierarchy running from theory down through principles and rules to judgements about cases. The objection came from people who worked on actual cases, and the objection was to the one-directional movement. The authors abandoned the model in later editions.

Reasoning from cases instead

The method is much older than the objection to the application model. It takes its name from the Latin for a case, it was developed in detail in early modern Catholic moral theology for the guidance of confessors, and it fell into disrepute after Pascal used the Provincial Letters, published between 1656 and 1657, to portray it as the art of constructing an excuse for anything a client wanted to do. The word survived mainly as an insult and still carries that sense in ordinary English.

Albert Jonsen and Stephen Toulmin recovered it in The Abuse of Casuistry in 1988, and their claim was that the method Pascal had ridiculed is what careful practical reasoning actually looks like when it is done well.

The procedure runs like this. Begin with cases whose resolution is not in doubt, where every reasonable person agrees and the reasons for agreement are clear. Arrange the surrounding cases in a series, moving away from the paradigm one feature at a time, so that it becomes visible which feature is carrying the weight. Locate the new case in that series. Where it sits close to a paradigm, the paradigm settles it. Where it sits between paradigms that point in opposite directions, the reasoning has to say which resemblance matters more and why, and the answer will be a judgement rather than a derivation.

Two features of this deserve notice. It does not require agreement on any theory, which is why it works among people who disagree about everything general. And its conclusions come with an honest grading: some are certain, some are probable, and some are the best that can be said, which is closer to how practical judgement actually feels than a derivation from principles ever is.

Where reasoning from cases runs out

Casuistry has its own failure and it is the mirror image of the other one.

Paradigm cases can be corrupt. Every society has cases it regards as settled, and some of those settled cases have been appalling. A method that reasons outward from what everyone agrees on has no way to say that what everyone agrees on is wrong, and the moral advances that matter most have usually consisted of exactly that claim. Reasoning by analogy from accepted cases will reproduce the assumptions embedded in them with great fidelity.

Analogy needs something to work with. Saying that this case resembles that one is empty until it is said in which respect, and a claim that a respect is the relevant one is a general claim about which features carry moral weight. That is what a principle is. Casuistry does not escape principles; it uses them without stating them, which makes them harder to examine rather than absent.

And Pascal's charge has not been answered so much as accommodated. A method that proceeds by finding a resemblance to some accepted case, in a stock of cases large enough, can find a resemblance to something for almost any conclusion. The safeguard is the honesty of the person doing it, which is not much of a safeguard.

The two-way street

The position most practitioners now occupy takes something from each direction. Judgements about cases are evidence about principles, principles are evidence about cases, and each is revised in the light of the other until the whole set hangs together. Rawls gave the procedure its name, reflective equilibrium, in a work of political philosophy, and it has been taken up in ethics far more widely than the theory it was designed to support. Beauchamp and Childress adopted it in place of the hierarchy they began with.

The obvious complaint is that this is a description of how people revise their views rather than a test of whether the resulting views are true, and that a coherent set of beliefs can be coherent and wrong. That complaint is serious and the next section is a version of it.

A more radical position holds that the search for principles was misconceived from the start. Jonathan Dancy's moral particularism rests on a claim about how reasons behave: a feature that counts in favour of an action in one case can count against it in another, and can be irrelevant in a third. If that is right, no finite list of principles can capture what makes actions right, and the competent moral judge is someone with the perception to see what matters here rather than someone equipped with rules. The position is a minority one and it is difficult to refute.

Next

Nothing so far has said which beings the whole apparatus is about. Every case in this article involved a human patient, and the assumption that a human being is the kind of thing that can be wronged went unexamined because it is not usually questioned. It becomes urgent as soon as the beings in a case are animals, or embryos, or people whose capacities have been permanently lost, or machines. The next article asks what it takes for something to count.