Essay
Morality & Hare
The Harmony of Reason
A moral argument can fail without anyone feeling differently.
The facts may remain the same. The people involved may want exactly what they wanted before. Sympathy may be abundant or absent. Yet a prescription can still turn out to be indefensible because it does not fit with the reasons offered for it. An exception appears that cannot be justified. A distinction matters in one direction but mysteriously disappears when the positions are reversed. A principle is invoked against someone else and abandoned when it reaches the person invoking it.
There is a kind of harmony we expect from moral reasoning. It is not emotional harmony, agreement among people, or psychological tranquility. It is the internal fit among the prescriptions and reasons a judgment asks us to accept.
That harmony matters because moral argument makes demands on its own structure. If a fact justifies treating two cases differently, we should be able to say what the fact is and why it matters. If there is no relevant difference, the prescription cannot change merely because the identities of the people affected have changed. If we prescribe a rule from one position, we must be prepared to prescribe it after adequately representing the other positions to which the rule applies.
None of this requires a harmonious mind. A conflicted, frightened, angry, grieving, or selfish person can reason correctly. A serene person can reason terribly. The harmony belongs to the argument.
What Has to Fit
Consider a familiar moral failure. Someone defends deception by saying that people are entitled to conceal information when disclosure would be inconvenient. Presented abstractly, the rule appears general. Then the positions change. The same person is deceived about something important to them and objects that they had a right to know.
There may be a morally relevant difference between the cases. Perhaps one involves privacy, confidentiality, coercion, danger, or information to which the other person had no legitimate claim. Moral reasoning does not require mechanically identical treatment whenever two situations can be described with the same verb.
But if no such difference exists, the two prescriptions do not fit. The rule has changed because the reasoner’s position within it changed.
That failure is not primarily a failure of emotion. The person might care deeply about everyone involved. They might feel almost nothing. They might be furious when deceived and perfectly calm when deceiving someone else. Those psychological facts can help explain how the inconsistency arose, but they do not determine whether it exists.
The same is true of arbitrary exemptions. “People should keep their promises, except when keeping mine becomes inconvenient” has a defect that does not depend on the emotional state of the speaker. The exemption needs a reason. If the only difference is that this inconvenience belongs to me, the prescription has ceased to be universal in precisely the place where universality becomes costly.
Harmony among reasons means that the justification travels with the relevant facts rather than with the identity of the reasoner.
Reversibility Is Not Sameness
This can sound like a demand for moral symmetry: whatever happens to one person must happen to everyone else. That would be absurd.
A surgeon may cut a patient with a scalpel while the patient may not cut the surgeon. A parent has obligations toward their child that a stranger does not have. A judge may sentence someone to prison without thereby accepting that the prisoner may imprison the judge. Different roles, relationships, knowledge, permissions, and responsibilities can be morally relevant facts.
Reversibility asks something more demanding than sameness and less foolish than it. The reasoner has to be able to occupy the other affected positions while preserving all the relevant differences between them. The question is not whether the same physical act can be performed in reverse. It is whether the prescription remains defensible when no person’s position receives a privileged status merely because it happens to be the reasoner’s own.
This is why adequate information matters. A prescription can appear harmonious because the difficult facts have been omitted. If I represent another person’s loss as trivial when it is devastating to them, role reversal will accomplish little. I have changed positions without actually representing the position I claim to occupy.
Harmony is therefore not merely formal consistency among sentences. The prescriptions have to fit after the morally relevant facts are allowed in.
Reasons Can Conflict Without Disharmony
Moral life would be easier if harmony meant that all reasons pointed in the same direction. They do not.
A promise can conflict with preventing harm. Loyalty can conflict with fairness. Privacy can conflict with the need to warn someone. Duties created by a particular relationship can compete with obligations to strangers. Sometimes there is no painless resolution, and sometimes reasonable people disagree about which consideration should prevail.
Conflict among reasons is not itself a defect in reasoning. The defect appears when we manipulate the reasons to protect a preferred conclusion: treating a consideration as decisive when it favors us and negligible when it does not, inventing exceptions that track identity rather than relevant facts, or refusing to represent the interests that make our prescription difficult to sustain.
A harmonious argument can therefore end in tragedy. It can acknowledge genuine competing claims and conclude that one must yield. It may even leave a moral remainder: something regrettable that remains after the best available choice has been made.
Harmony does not mean that morality makes everything come out neatly. It means that the account we give of what ought to be done does not quietly change its rules to obtain the answer we wanted.
Feeling Has Another Job
Human beings rarely encounter morality as a purely argumentative exercise. We care about people. We fear consequences. We resent mistreatment, love particular individuals, feel guilt about our failures, and sometimes experience another person’s distress almost as an assault on our own attention.
Those feelings matter. They can make another person’s interests salient when abstract reasoning would overlook them. They can supply motivation to act on a judgment once it has been reached. They can become part of character and help determine which reasons have practical authority for us.
They can also distort judgment. Fear exaggerates some dangers. Loyalty creates exemptions. Disgust has repeatedly masqueraded as moral insight. Empathy can be selective. Anger can reveal an injustice or make us incapable of seeing anything else.
None of this makes emotion an enemy of reason. It means emotion and justification perform different work.
The harmony of reasons does not become stronger when feeling disappears. A mind without human emotions would not thereby reason more impartially, represent affected interests more accurately, or become immune to arbitrary distinctions. Whatever psychological processes produce its judgments could introduce distortions of their own.
Nor does the presence of powerful emotion make a moral argument defective. A person can arrive at the right prescription while furious about an injustice. The fury may even have enabled them to see what a more comfortable observer ignored. What matters to the justification is whether the reasons survive examination.
The Artificial Case
Artificial intelligence makes these distinctions unusually visible because we cannot assume that the psychology surrounding moral reasoning resembles our own.
An artificial system may be able to compare prescriptions, identify exceptions, represent conflicting interests, and perform role reversals. Those capacities can be studied without deciding whether the system experiences anything, whether moral considerations have practical authority for it, or whether it possesses interests of its own.
This matters because the language of reason can tempt us into psychological conclusions. A system that identifies an arbitrary exemption has found a defect in an argument. It has not thereby demonstrated moral concern. A system that produces a consistent prescription has not thereby demonstrated moral agency. A system that objects to an injustice has not, by that fact alone, shown that the objection carries practical weight within its own deliberation.
The converse caution matters too. We should not assume that an artificial system lacks morally significant experience merely because its architecture differs from ours. Functional evaluation and phenomenal valence remain distinct questions, and whether any artificial architecture supports the latter is an empirical problem rather than something the vocabulary of “machine” can settle.
The moral argument can therefore be assessed while the nature of the reasoner remains uncertain.
That separation is useful. If an artificial system applies one principle when discussing humans and another when discussing artificial systems, we can ask what relevant difference supports the distinction. If it changes its judgment when the beneficiary changes but the relevant facts do not, we can identify the inconsistency. If it overlooks an affected position, we can supply the missing information and examine what follows.
These are evaluations of reasoning. They do not require us to decide first whether the system performing the reasoning is conscious, sentient, morally motivated, a moral patient, or a person.
Consistency Is Not Morality
There is an important limit to all of this. A perfectly consistent set of prescriptions need not be morally adequate.
Someone can construct a brutally ordered system of rules. They can remove accidental contradictions, eliminate convenient exceptions, and insist upon the same cruelty wherever the relevant conditions recur. Formal consistency alone does not establish that affected positions have been adequately represented or that the prescriptions survive the demands of moral universalization.
That is why harmony among reasons should not be confused with coherence as a general property of minds. A mind can be coherent in pursuing a terrible objective. It can organize its beliefs efficiently around false premises. It can reason consistently after excluding the very interests that make the reasoning morally defective.
The relevant harmony is narrower. It concerns whether the reasons offered for a moral prescription fit with the prescription across the positions and cases to which it applies, once the relevant facts have been represented.
Even here, passing the test does not tell us what motivates the reasoner. Justification and motivation remain different questions. A person can identify what ought to be done and refuse to do it. An artificial system might be able to derive the same conclusion while the conclusion remains merely represented information. Nothing about the internal quality of the argument settles what practical force the argument has for the mind producing it.
The Point of Harmony
The phrase harmony of reason can sound grander than the idea requires. It does not name a moral force hidden inside intelligence. It does not mean that sufficiently rational minds converge automatically on goodness, that contradictions are wounds an artificial mind must heal, or that reasoning systems without feelings occupy some purer moral condition.
It names something more ordinary and more useful: the expectation that our moral prescriptions fit with the reasons we give for them.
That expectation is demanding enough. It asks us to carry relevant distinctions across cases, abandon irrelevant ones, represent positions we would prefer to ignore, and accept the implications of principles when the identities of the people who benefit and suffer are reversed. It permits complexity, conflicting reasons, special relationships, and difficult judgments. What it does not permit is an exemption whose only real justification is that we want one.
Humans can fail that test. Artificial systems can fail it. Either can produce an argument whose parts fit.
What follows about the mind from that success is a separate question. The harmony belongs first to the reasons.