Moral Intuitions and What to Do With Them

The previous article ran on an assumption it never stated. Almost nobody says you may push the man off the footbridge, and that fact was treated throughout as something a theory has to accommodate rather than something a theory might overrule. Say why and you have said what the method of moral philosophy is.

The thing being relied on

Three misreadings of the word are worth clearing out, because each turns the topic into something easier and less interesting.

An intuition is not a feeling. Revulsion at a photograph is a feeling; the judgement that what it shows was wrong is an intuition, and the two can come apart in both directions. People have intuitions with no accompanying affect at all, such as that a promise made under duress is not binding, and they have strong reactions to things they judge to be permissible.

An intuition is not a hunch, in the sense of a guess with low confidence. Some are held with more confidence than anything else the person believes. That gratuitous torture of a child is wrong is an intuition, and a person surer of the premises of an argument against it than of it has miscalibrated something.

And an intuition is not a report about the person having it. "Pushing him would be wrong" is a claim about the pushing, and treating it as a description of the speaker's psychology changes the subject.

This is what the trolley pair was doing. A principle was proposed, a case was constructed on which the principle gives an unacceptable verdict, and the principle was counted as damaged. The whole of the previous article is an application of the method, and so is most of the moral philosophy published in the last century.

Why anything is built on this

The obvious complaint is that a reaction to a made-up scenario is a poor foundation for anything. The reply is that there is no alternative on offer, and it is worth seeing how thoroughly there is not.

A moral theory is a general normative claim. It cannot be derived from descriptive premises without the addition the first article flagged. It cannot be established by observation, since observation delivers facts about what people do and believe rather than about what is right. It cannot be proved from self-evident axioms, because the candidate axioms are themselves things that seem right on consideration, which is to say intuitions with a formal presentation. Anything that gets a moral theory started has to start somewhere, and the somewhere is judgements that strike people as correct.

Compare the situation in logic, where the same structure is visible and less controversial. A proposed rule of inference is tested by whether it licenses an argument we can see is bad. That test treats a judgement about a particular argument as evidence about the general rule, and nobody thinks logic is thereby founded on prejudice.

The method stated properly

John Rawls gave the method its name and its standard statement in 1971, in the course of arguing for a theory of justice rather than a theory of right action, and the machinery has since been applied everywhere. Nelson Goodman had made the same structural point earlier about the rules of inference, arguing that rules and accepted inferences are justified by being brought into agreement with each other.

Two features of the picture are easy to miss and do most of the work.

The adjustment runs both ways. Reflective equilibrium is not the method of preserving all your reactions and finding a theory that fits them, which would be a machine for entrenching whatever you started with. A sufficiently well-supported principle can override a confident judgement about a case, and this has repeatedly happened. Positions on slavery, on the standing of women, and on the treatment of animals have all been changed by arguments that ran against firmly held reactions.

And the inputs are filtered before they are used. Rawls did not take all moral judgements as data but only considered judgements: those made without distraction or emotional distortion, without a stake in the outcome, in conditions where the person is likely to judge well. A reaction produced by disgust at unfamiliarity, or by self-interest, does not enter the process at all.

A refinement is standard and worth naming. Narrow reflective equilibrium seeks agreement between your judgements and the principles that systematise them. Wide reflective equilibrium adds a third element: the background theories, philosophical and empirical, that bear on whether those principles could be true. The narrow version is vulnerable to the charge of tidying up a prejudice, and the wide version is the one its defenders actually hold.

What the method cannot do

Two limits should be stated by anyone defending the method, because its critics will state them anyway.

Equilibrium is not uniqueness. Two people who both proceed carefully can reach coherent systems that conflict, if they started from different judgements and made different adjustments. Nothing in the method guarantees convergence, and this is a real cost rather than a technicality, since a method that lets careful reasoners end up anywhere is not obviously a method for finding anything out.

And coherence is not truth. A person whose principles and judgements agree perfectly has achieved consistency, and a well-constructed error is consistent. The wide version is a response to this, since bringing in background theories imports constraints from outside the moral system, and whether that is enough is disputed.

Stepping back

Every question so far has been about which moral claims are true. A different set of questions has been visible at the edges throughout: what a moral claim is, whether there is anything for it to be true of, and how anyone could find out. The last article of this tier introduces them.