Moral Arguments

The three arguments so far took their starting point from the world: that there is one, that it is ordered, that a concept of it can be formed. This one starts from something closer in.

Almost everyone has the experience of being under a requirement they did not choose and cannot dissolve by deciding they would rather not be under it. Torturing a child for entertainment is not merely something one has decided against. It is forbidden, and it would remain forbidden if everyone alive came to approve of it. The moral argument takes that feature seriously and asks what sort of world could contain requirements of that kind.

Before any of that can be assessed, a version of the argument that circulates widely has to be removed, because it is not the argument, and confusing the two makes the real dispute impossible to see.

Clearing the ground

The version to discard first is the claim that people who do not believe in God cannot behave well or cannot tell right from wrong. That claim is empirical, it is false, and nobody who defends a moral argument in the philosophical literature asserts it. Whether the people around you behave decently has no visible relationship to what they believe about God, and every serious defender of these arguments concedes this without reluctance.

Three questions have to be kept apart, and the popular version runs them together.

Whether a person can know moral truths without believing in God is one question, and all parties answer yes. Whether a person can act well without believing in God is a second, and all parties answer yes to that too. Whether moral facts of a particular kind could obtain in a world with no God is the third, and it is the only one in dispute.

That third question is about grounding rather than about knowledge or behaviour. The parallel is with arithmetic. Someone can know that seven is prime, and use the fact reliably, while holding any view at all or no view about what numbers are. A dispute about what makes arithmetical statements true is not a dispute about whether anyone can count.

Within the third question there are two families of argument, and they work differently. Theoretical versions claim that certain moral facts exist and that God is the best explanation of them. Practical versions claim that a person committed to morality is rationally required to believe in God in order for that commitment to make sense, which is a claim about what an agent must assume rather than about what is the case.

The feature the argument leans on

Obligations have a character that other evaluations do not. It is good to be witty and unfortunate to be tone-deaf, and neither is a demand. An obligation is a demand. Failing to meet one leaves you guilty rather than merely mistaken or unlucky, and the guilt is not something you can escape by pointing out that you never agreed to the requirement.

Elizabeth Anscombe pressed this point in a paper that both sides of the dispute now cite. Writing in Philosophy in 1958, she argued that the concepts of moral obligation and moral duty are survivals from an earlier conception of ethics in which ethics was understood as law, and that they had a home in that setting because a law conception supplies a lawgiver. Detached from it, she thought, the words carry a force they have no longer earned, and continuing to use them produces a philosophy that trades on associations it has disowned.

Two things about Anscombe should be recorded, because the paper is regularly quoted in a misleading way.

Her recommendation was not that philosophers return to a divine lawgiver. It was that they stop using the terms until moral philosophy has an adequate account of human action and psychology, and the positive programme she launched was virtue ethics, which does without obligation as a basic notion entirely.

And her paper is used by both sides for exactly that reason. The theist reads it as showing that obligation requires a lawgiver. The secular reader takes it as showing that obligation was always the wrong central concept and that ethics is better built on something else.

The theoretical argument from obligation runs from the feature to a source. Obligations function like commands, in that they issue demands rather than merely describing what would be admirable. Commands require a commander with the standing to issue them. No human being or human institution has standing over all people in all places regardless of their consent. So the source is something else, and the candidate on offer is God.

Kant's version

Kant's argument in the Critique of Practical Reason, published in 1788, is the most influential of the practical family, and it is often reported inaccurately.

The structure is this. Morality directs us toward what Kant calls the highest good, in which virtue and a proportionate happiness are combined. We cannot coherently be commanded to pursue what we take to be impossible. Nature by itself provides no guarantee that virtue and happiness ever line up, and the observable evidence runs the other way. So a rational agent who takes the moral law seriously must postulate the conditions under which the highest good is possible, and those conditions include an immortal soul and a God who can bring the two into proportion.

What this is not is a theoretical proof, and Kant is explicit. He had spent the first Critique dismantling the theoretical arguments, including the ontological argument covered in the previous article, and he did not think he had smuggled one back. The postulates are things a moral agent must assume in order to be committed to morality without incoherence, and Kant regarded that as a different kind of warrant altogether.

The vulnerability is correspondingly different. A critic need not attack any premise about God. It is enough to say that a person can accept the moral law, act on it, and be resigned to the highest good never being realised, holding that the universe is simply not arranged to reward virtue. Whether that position is incoherent or merely bleak is the whole of the dispute, and most readers since Kant have found it merely bleak.

Divine command theory

Robert Adams gave the version that current discussion works with, in a paper of 1973 and then at length in Finite and Infinite Goods, published by Oxford University Press in 1999. Two features distinguish it from cruder ancestors.

It applies to obligation only. Goodness is not constituted by commands on Adams's account; it is grounded in God's own nature, and finite things are good by resembling it. So the theory is a theory about a part of morality, and Adams leaves the rest to accounts that are not voluntarist at all.

And the commands in question are those of a loving God, not of any commander whatever. This is not a decoration. It does most of the work in answering the objections below.

Adams also embeds the view in a general account of obligation rather than treating divine commands as a special case. Obligations, he argues, arise from demands actually made within relationships that matter, which is why a promise creates one and a stray wish does not. A divine command account then extends a familiar structure rather than inventing one for this occasion.

The Euthyphro dilemma

The question comes from Plato's Euthyphro, at the passage marked 10a, where Socrates asks whether the pious is loved by the gods because it is pious or is pious because the gods love it. Two features of the original are worth noting before it is transferred. It concerns piety rather than morality generally, and it concerns many gods who may disagree rather than one. The transfer to a single God and to obligation is standard and it is not automatic.

Take the first horn, that an act is right because God commands it. Then had God commanded cruelty, cruelty would have been obligatory, and the content of morality is settled by a will that could have gone either way. A second problem follows the first. If goodness is whatever God commands, then calling God good says only that God conforms to God's own commands, which is not praise but a tautology. The claim that God is good is meant to be substantive, and this horn makes it empty.

Take the second horn, that God commands things because they are right. Then rightness is prior to and independent of God's will, and God stands to morality as an authoritative reporter rather than a source. Whatever else follows, the moral argument for God's existence is finished, since the moral facts were there anyway.

Adams's answer is to divide the question rather than pick a horn. For goodness, he takes the second: goodness is not commanded into being, and God's own nature is the standard, which is why calling God good is not trivial. For obligation, he takes the first: what makes an act obligatory is that a loving God has commanded it. Because God is essentially loving, the commands imagined in the arbitrariness objection are not merely unissued but impossible, so the horn's sting is removed.

That answer has held up better than its critics expected, and two objections to it remain live.

The first concerns what the theory still commits its holder to. Even granting that God cannot command cruelty, the theory appears to make the statement true that if God commanded cruelty, cruelty would be obligatory. Defenders reply that a conditional with an impossible antecedent tells us nothing, since anything follows from an impossibility and we should not read theoretical commitments off such statements. Critics reply that willingness to accept the conditional is evidence about what the theory actually says, and that the standard treatment of impossible antecedents is itself disputed. Neither side has closed this, and describing it as settled in either direction misreports the literature.

The second is more interesting and Adams accepts it. If God's nature constrains which commands can be issued, then something is prior to the commands after all, and that something is goodness. The second horn has been taken for a large part of morality. Adams's response is that this was always his position and that obligation is the only thing he claimed to ground in commands. The concession is real, and it makes the moral argument narrower than its popular presentations suggest: what would be established is not that morality depends on God but that one feature of it, the bindingness of requirements, does.

The alternatives

Two secular positions occupy the ground the argument wants.

The first accepts that moral facts are objective, irreducible, and not natural facts, and denies that they need a ground outside themselves. Erik Wielenberg develops this in Robust Ethics, published by Oxford University Press in 2014, holding that ethical features are necessary and depend on nothing for their existence, supervening on non-moral features in a way that requires no explanation because necessary truths do not have makers.

The theist's complaint is that this leaves brute necessities lying about unexplained. Wielenberg's reply is that the theist has one too. On the theistic account God's own nature is not made by anything and could not have been otherwise, so both views terminate in something necessary and unexplained, and the dispute is over where to stop rather than whether to. That reply is hard to answer, and it is the reason the moral argument is less often presented as decisive now than it was fifty years ago.

The second alternative is explanatory rather than metaphysical. Our disposition to make moral judgements has a natural history, and the shape of that history was not selected for tracking moral truth. Sharon Street pressed the resulting difficulty in 2006, and it is aimed at moral realists generally rather than at theists in particular, which means the theistic realist inherits it along with everyone else and has to answer it too.

Whether an account of why we hold a belief undermines the belief is a separate matter, and treating explanation as refutation is a recognised mistake. The last article in this tier examines that pattern of argument directly, where it is applied to religious belief rather than to moral belief, and the analysis there transfers.

Next

Every argument in this tier so far has been public, in the sense that its premises are available to anyone willing to consider them. The next takes up a class of evidence that is not public at all: the reports of people who take themselves to have encountered God directly, and the question of what such a report can establish for the person who had it and for anyone else.