Draft
 

MORAL TRUTHS AND MORAL PRINCIPLES

Curtis Brown

In recent years, a number of moral philosophers have held both that there are particular moral truths, and also that there are no general moral principles which explain these particular moral truths--either because there simply are no moral principles, or because moral principles are themselves explained by or derived from particular moral truths rather than vice versa. Often this combination of doctrines is held by philosophers interested in reviving an Aristotelean approach to ethics,1 but it has other defenders as well.2 I want to argue that these views cannot both be correct: if there are particular moral truths, there must be universal moral principles which explain why they are true.

I will argue that it is essential to distinguish between two uses of moral principles: to explain moral truths, and to guide or justify moral action. In arguing that, if there are moral truths, there must be moral principles, I am not arguing that moral principles are necessary as practical guides to action, nor even that an agent must appeal to principles in order to justify particular decisions. Like critics of the view that there are moral principles, I am skeptical about both these views. My claim is rather that moral principles are needed to explain particular moral truths: to explain what makes particular right actions right, what makes particular courageous actions courageous, and so on.

Most of the literature I take myself to be opposing does not make the distinction between explanation and guidance. Moreover, the examples and arguments in the literature focus almost exclusively on the action-guiding role of moral principles. So there is some danger that I am defending a view these writers do not mean to attack. But the writers with whom I am concerned at least appear to think that showing that we need not or should not rely on moral principles in acting morally suffices to show that moral principles are altogether unnecessary; indeed, they sometimes argue apocalyptically that nearly all of modern moral philosophy has consisted of a misguided effort to discover moral principles. I will argue that the conclusion that there are no moral principles does not follow from the critics' arguments about the role of such principles in moral reasoning.

A moral principle, in my usage, is simply a universal moral truth: a moral truth which makes no reference to particular people, places, or times. What might such principles look like? They might look like this: "one ought to perform whatever action, of those available to one, would maximize overall utility." Or like this: "one ought to act only on maxims one could will to be universal laws." Or like this: "one ought to refrain from acting in ways of which an ideal observer would disapprove." These are all principles of considerable generality. But there may also be more specific or local moral principles: "One ought to help people in distress," perhaps, or "one ought not to break one's promises." (I will use the terms "rules" and "principles" interchangeably. But they could be used to mark any of several distinctions; for example, one could reserve the term "principles" for very general principles, and use the term "rules" for more specific principles.3) Then what if the only way I can help someone in distress is by breaking a promise? It may be that in such a case there is simply no fact about what I ought to do. Conceivably the only moral principles are specific and local, and there are no facts about how to resolve conflict among principles. But if there is a fact about what I ought to do in a case of conflict, then, I shall argue, this can only be because there is a more general moral principle which explains why this is so.

Opponents of the view that there are principles in ethics, like opponents of general laws in history or general principles in aesthetics, can point to a long history of failed attempts to find such principles. This may be evidence that there are no such principles. But even if it is, this does not count against the conditional claim for which I will argue. For if I am right, then evidence that there are no principles is also evidence that there are no moral facts. (Similarly, evidence that there are only specific principles, and no principled way to resolve conflict among them, is evidence that there are fewer moral facts than we might have thought.)

I will proceed by exploring an extended analogy between ethics and the sciences. This is hardly a novel project, but the parallels I will draw and the uses to which I will put them are not the usual ones. In the process, I will both advance a positive argument for principles and undermine some points made in the anti-rule literature.

PREDICTION AND EXPLANATION

The fundamental distinction on which I want to insist is parallel to the distinction in the sciences between prediction and explanation.4 It is a commonplace that principles or laws in the sciences serve two functions. First, they have a predictive function: they enable us to make predictions about future events.5 It is useful to further distinguish between two sorts of predictive role principles may have. They may play a psychological role in enabling us to form beliefs we might not have formed otherwise. But there may also be occasions on which a scientist forms a belief about a particular event based not on deduction from general principles but on a sort of "hunch"--on a felt similarity to certain previous phenomena, perhaps. Scientific laws or principles may still play an epistemological role in justifying this belief, even though they did not play a psychological role in one's coming to have the belief.

In addition to the predictive function of scientific principles, whether psychological or epistemological, there is a second, explanatory function: such principles explain particular empirical facts. We might regard this explanatory role as metaphysical rather than psychological or epistemological, since it concerns what makes particular facts true rather than how we come to know them or justify our belief in them. Often the very same principle will both predict and explain the same event: the gas laws together with statements of particular fact will enable us to predict that if we heat a balloon tomorrow, it will expand; once we have heated the balloon, the same principles explain why it expanded. But though there is often a close connection between explanation and prediction, they are fairly clearly distinct notions.

We need to distinguish further between two sorts of explanation. The sort most familiar from the philosophy of science is event explanation: we explain the occurrence of a particular event by showing how it fits into a broader causal network. But there is another sort of explanation which is also important in science. In property explanation, we explain something's possession of a property by showing how to analyze its having the property into its (or its components') having more basic properties. As Robert Cummins writes:6 "Many scientific theories are not designed to explain changes, but are rather designed to explain properties. The point of what I call a property theory is to explain the properties of a system not in the sense in which this means 'Why did S acquire P?' or 'What caused S to acquire P?' but, rather, 'What is it for S to instantiate P?', or, 'In virtue of what does S have P?'" Cummins offers the following examples of property explanations in the sciences: "We know how temperature is instantiated, how inheritance is instantiated, how electricity is instantiated, how solubility is instantiated. I think we are close to knowing how life and intelligence are instantiated, though we are a long way from understanding how consciousness or intentionality are instantiated."

Moral principles, like scientific ones, have both an explanatory or theoretical role and a "predictive" or action-guiding role. Moral rules provide us with practical guides to action; in conjunction with statements of particular fact they enable us to "predict" whether the actions we are considering would be morally acceptable. But moral principles also explain why certain actions are right or wrong, good or bad, morally acceptable or morally unacceptable. Typically the same moral principles will play both roles: "It is wrong to break a promise" not only guides the behavior of those of us who want to avoid acting badly, but also explains why certain actions are morally wrong.7 The sort of explanation offered by moral principles is, it is important to note, property explanation rather than event explanation.

Criticisms of the possibility or desirability of moral principles typically ignore this distinction between the explanatory or theoretical and the action-guiding or practical roles of moral principles,8 and focus exclusively on their role in guiding action.9 But while the great historical defenders of particular principles have been interested in providing practical guides to action, surely they have been at least as concerned to provide general explanations of particular moral truths. Some arguments against rules or principles are simply irrelevant to this explanatory need for principles.

For example, in a recent article John Waide10 attempts to refute the claim "that virtue judgments always presuppose moral principles and, therefore, that virtue ethics is always derivative from principle ethics." On one interpretation the view Waide criticizes is a special case of the view I will defend, since the judgment that someone is courageous (say) is, if true, precisely the sort of particular moral truth which I maintain requires explanation by moral principles.

Waide distinguishes between two versions of the view he rejects. The "psychological version," which Waide rightly regards as highly implausible, "would hold that a person has in mind . . . a moral principle each time a virtue judgment is made." Waide calls the second version of the view "metaethical," but I prefer to call it "epistemological," since it concerns the conditions under which a belief is justified; it holds that "if a virtue claim is to be warranted, one must be able to offer reasons for it, reasons which must include reference to some moral principle." This is the version of the view that virtues presuppose principles which Waide goes on to attack.

Now, I am strongly inclined to believe, with Waide, that both the psychological and epistemological versions of the view that virtues presuppose principles are false. Virtue judgments do not presuppose judgments of principle. But Waide does not even mention what I take to be the most persuasive version of the view he criticizes, namely the metaphysical version: if a claim about someone's possession or exercise of a virtue is true, then there must be moral principles which play a role in explaining why it is true.11

In some areas of philosophy, notably the philosophy of language and the philosophy of science, we have learned to be careful to distinguish between epistemological and metaphysical claims or notions: Kripke, for instance, has stressed the importance of distinguishing between the epistemological notion of the a priori and the metaphysical notion of necessity.12 We need to become more careful about these distinctions in ethics as well.13

A related failure to distinguish the epistemological from the metaphysical seems to be involved when various writers argue that there cannot be moral principles on the ground that we cannot codify our own views on moral matters. John McDowell, for instance, writes: "If one attempted to reduce one's conception of what virtue requires to a set of rules then, however subtle and thoughtful one was in drawing up the code, cases would inevitably turn up in which a mechanical application of the rules would strike one as wrong--and not necessarily because one had changed one's mind; rather, one's mind on the matter was not susceptible of capture in any universal formula."14 And Martha Nussbaum makes the related point that "a system of rules set up in advance can encompass only what has been seen before,"15 arguing a few pages later that "the only procedure to follow [in reasoning about what constitutes a good life] is . . . to imagine all the relevant features as well and fully and concretely as possible, holding them up against whatever intuitions and emotions and plans and imaginings we have brought into the situation or can construct in it" (180). To the extent that these writers mean only to write about moral decision making, I need not quarrel with them: but in both one gets the feeling that there is an implicit inference from "principles are unnecessary (and perhaps harmful) in moral reasoning" to "principles are unnecessary." This seems to presuppose that the only function of moral principles is their predictive or action-guiding role, and so to ignore their theoretical or explanatory role.

Simply drawing the distinction between the predictive and explanatory roles of moral principles, then, is enough to call into question some arguments against principles. Further attention to the relation between these two roles will provide the material for further criticisms of such arguments. First, however, I will offer a positive argument for principles.

THE NEED FOR PRINCIPLES

There is a very simple argument from three premises to the existence of moral principles. I will briefly sketch the argument, and then argue for two of the premisses by responding to some objections. The first premise is that there are some particular moral truths. For example, consider a case in which an agent has a choice between two possible actions, A and B.16 In many such situations, the choice between A and B will be morally indifferent--perhaps the choice is between eating chocolate ice cream and eating vanilla, or between giving a thousand dollars to a perfect stranger and giving it to her identical twin, or between breaking and entering the house on the left or the exactly similar house on the right. But the critics of principles with whom I am concerned will agree that there are also at least some cases in which one action, say A, is morally preferable to the other, meaning by this that A is genuinely morally better than B, not that someone prefers it or exhorts its performance or expresses approval of it.17

The second premise is that particular moral truths must be explicable. For example, if A is genuinely better than B, then there must be some reason for this. It is hard to see how the moral preferability of one action over another could be simply some sort of brute fact, incapable of further explanation. If A is better than B, it must be better in virtue of other facts about A and B, and thus there will be an explanation of why A is better which appeals to these other facts.

The third premise is that the explanation of a particular moral truth requires that there be a moral principle. For example, an explanation of why A is better than B must do more than appeal to certain features of A and B. It must also say why these features are relevant; to do this, it must say something about the connection between these features and moral preferability. Perhaps A is preferable to B because B does, and A does not, involve breaking a promise. But this fact about A and B can only explain why A is better if there is some link between keeping promises and moral preferability; and if there is such a link it must hold in general and not just in the particular case. So the explanation that A is preferable because it does not involve breaking a promise can only be successful if there is a moral principle linking promise-keeping with moral preferability. (It does not matter whether the principle must form a part of the explanation, or whether it is only required for the success of the explanation; either way there must be a moral principle if A's preferability is to be explicable.)

From these three premises, it clearly follows that there are moral principles.18 And from the second and third premises alone, there follows the claim I am particularly concerned to defend, namely that if there are particular moral truths, there must be moral principles. So critics of this conditional claim appear to have little room to maneuver: they must claim either that particular moral truths require no explanation at all, or that their explanation does not require general moral principles, or that, although the conclusion of the argument is correct, it is somehow trivial.

Let us consider these objections in turn. First, someone might question whether moral truths require explanation at all. Indeed, one might suggest that my argument involves the Principle of Sufficient Reason run wild. Many things have no explanation: why is there something rather than nothing? Why did this radioactive atom decay today rather than tomorrow or yesterday? We no longer think that every phenomenon is such that there must be a sufficient reason for it. Why should moral truths in particular require explanation?

There are two main reasons for rejecting the Principle of Sufficient Reason; neither of them gives grounds for exempting particular moral truths from the need for explanation. The first reason is that the chain of explanations must end somewhere. In the case of causal explanation, it seems that there must be some causal question which has no answer: why did the Big Bang occur? Or why is there something rather than nothing? Similarly, in the case of property explanations, it seems that there must be some properties which are explanatorily basic, which do not require further explanation.

This first reason for the failure of the Principle of Sufficient Reason gives us no reason to doubt that moral properties must have sufficient explanations, since moral properties are fairly clearly not explanatorily fundamental or basic. If goodness were simply one of the basic properties of the universe, perhaps it would be senseless to ask in virtue of what A is better than B, as perhaps it is senseless to ask in virtue of what the Earth has a larger diameter than Mars. (We can explain how it came to have a larger diameter, but not what makes it the case that it has a larger diameter, in what its having a larger diameter consists.) But we do not think that goodness is ontologically basic. Moral facts are secondary in something like the sense in which color is a secondary quality. If one object is redder than another, there must be something in which this difference in color consists, some explanation of features of the objects in virtue of which the one is redder than the other; similarly, if one action is morally better than another, this must be explicable in terms of features of the actions. Moral properties are simply not part of the basic furniture of the universe. This is not so say, however, that moral terms must be definable in some more basic vocabulary, nor that moral concepts can only label regions of conceptual space already mapped out by means of other concepts.19 The dependence of moral properties is metaphysical, not linguistic or conceptual.

The second reason for the failure of the Principle of Sufficient Reason is that it now seems likely that determinism is false, that some events are irreducibly probabilistic. Suppose for simplicity that the outcome of a coin toss is like this. I toss the coin; it comes up tails. Why did it come up tails? There simply is no sufficient reason: it could equally well have come up heads. There may be an explanation of sorts: the fact that the coin had a 0.5 probability of coming up tails may have some explanatory power even though it does not enable us to predict beforehand which way the coin will come up.20 But although there is a kind of reason the coin came up tails, it is clearly not a sufficient reason. In similar cases, the coin will come up heads about half the time and tails about half the time, and there is no explanation of why a toss has a particular outcome on a particular occasion.

This failure of determinism provides counterexamples to what we might call the Principle of Sufficient Event Explanation. By itself the failure of determinism does not give any reason to doubt the Principle of Sufficient Property Explanation, the principle that if a certain property is instantiated there must be a sufficient explanation in terms of more basic properties. Still, although the failure of determinism gives us no reason to doubt that moral properties have sufficient explanations, it can provide a model of what it would be like for them to fail to have such explanations. The analogue in the case of moral (or other) properties to the failure of determinism would be something like this: a configuration, N, of non-moral properties of an action lends certain probabilities to the propositions that the action has each of two or more moral properties. To keep this case parallel to that of a coin toss, let us say that the action's having N lends probability 0.5 to the proposition that the action is virtuous and 0.5 to the proposition that it is vicious. If this is so, then roughly half the actions with N will be virtuous and half vicious. But there will be no saying why a particular action with N is virtuous as opposed to vicious.

This, I submit, is simply not believable. If one action with N is virtuous and another vicious, there must be some further nonmoral property which distinguishes them and which accounts for their difference in moral properties. Neither of the main reasons for the failure of the Principle of Sufficient Reason, then, provides reason to think that the possession of a moral property might lack a sufficient explanation.

The objection I have just considered is that particular moral facts might be the sort of facts which do not require explanation at all. A second objection is that, even if particular moral facts require some sort of explanation, there is no reason to suppose that this explanation must be of the sort which involves universal moral principles. The explanation may itself be particular or local. The answer to the question "Why was Joseph's action wrong?" may be "because it involved breaking his promise," without it being the case that any promise-breaking would be wrong.

We need to distinguish two ways in which it might happen that the explanation of the wrongness of Joseph's action might fail to involve universal moral principles. First, there might be the sort of failure of the Principle of Sufficient Reason we have just discussed: perhaps the fact that an action involves breaking a promise makes it likely that the action is wrong, but not all actions which involve breaking a promise are wrong, and there is no explanation of the difference between the ones that are wrong and the ones that are not. I have already suggested that this possibility cannot be taken seriously.

But there is a second way in which the explanation of the wrongness of Joseph's action might be local. It is true, I think, that explanations of particular moral facts will typically appeal either to principles with ceterus paribus clauses, or to something like the principles which specify W. D. Ross's prima facie duties. Let us consider the second, Rossian possibility.

Ross holds that there are principles which specify prima facie duties--roughly, principles each of which specify what would be our duty if that principle applied and no other principle did. Thus if "one ought not break one's promises" is such a principle, then if A involves breaking a promise one has a prima facie duty not to do A. And one also has a duty simpliciter not to do A, unless some other such principle applies, e.g., "one ought not to cause harm." So the explanation of the wrongness of Joseph's action might be that it involves breaking a promise and no other prima facie duties apply. (But then there is a relevant universal principle, namely that it is wrong to do something if one has a prima facie duty not to do it and no prima facie duty to do it.)

So far there is no violation of the demand that moral explanations require the truth of universal moral principles. But now what about cases in which prima facie duties conflict? On Ross's view if A involves both keeping a promise and harming someone, then "there is no principle by which we can draw the conclusion that it is on the whole right or on the whole wrong."21 On the other hand, Ross also believes that, although there are no moral principles to settle which action is preferable in cases of a conflict between prima facie duties, nevertheless in at least some cases one action is genuinely preferable. He writes, for example, that "this sense of our particular duty in particular circumstances, preceded and informed by the fullest reflection we can bestow on the act in all its bearings, is highly fallible, but it is the only guide we have to our duty."22 If our sense of duty is fallible, then there must be moral truths for it to discern or fail to discern.

But this combination of views simply invites a meta-application of our argument. It may be that in at least some cases where local or prima facie principles conflict there simply is no fact about what one ought to do: in fact I think it probable that there are some cases like this. But it also seems clear that in some cases in which prima facie principles conflict, there is a fact about what one ought to do. (Perhaps the only way I can keep my promise to meet you at noon is to murder the large, harmless but determined person blocking my way.) If there are such facts they must be explicable. And their explanation will require moral principles.23 (The required principles could be of either of two sorts. It could be that the principles about prima facie duty can be expanded into principles about duty all things considered, or it could be that there are meta-principles about how to resolve conflict among lower-level principles.) So the objection that the explanation of particular moral facts might appeal to local moral principles, while in one sense correct, does not refute the claim that the explanation of a moral fact requires the truth of a universal moral principle.

Now let us turn to a third objection to the argument. This objection holds that, although it is true that there are universal moral principles, nothing in the argument shows that these will be anything other than trivial and uninteresting. There is an important distinction between universality and generality: it may be that moral truths require universal moral principles and yet be the case that these principles are highly specific, perhaps so specific that they apply only to a particular case.24 The objection is related to claims that Kant's requirement of universalizability is trivial, since a description of the relevant features of an action may be so detailed and specific that the universalized maxim applies only to a single case.

In the worst possible case, the only available principles would be something like this. Why is A morally preferable to B? Well, A has total nonmoral properties N, B has total nonmoral properties N*, and for any actions x and y, if x has N and y has N* then x is morally better than y. This is a universal moral truth, and together with the facts of this particular case it logically implies the moral truth we wanted to explain. But although the principle is universal, it in fact has only one instance, since A and B are the only actions which have N and N* respectively. Why does my argument show the existence of principles in anything more than this empty sense?

Because these moral "principles" do not explain anything. An explanation must say something about which properties are relevant. Certainly, anything with all the nonmoral properties of A will be preferable to anything with all the nonmoral properties of B. But some of these properties matter to the explanation and some do not; some play a role in A's being better and some do not. To explain A's preferability by appealing to all the nonmoral properties of A and B is rather like explaining why a particular dropped stone falls by appealing to the total physical state of the universe in the previous moment. If we want to know why the stone fell, or why A is better, we want to know which facts or properties were instrumental in the stone's falling or A's being better.

This response to the objection still leaves open the possibility that the correct moral principles are highly specific. Not all of an action's properties are relevant to its moral status, but it may be that very many of them are. Consider again A's moral preferability to B. Let the total relevant nonmoral properties of A be R, and of B, R*; it may still be that A is the only action with R and B the only action with R*, and that the moral principle which "explains" A's preferability is that for any actions x and y, if x has R and y has R* then x is preferable to y--a principle whose only application is to A and B.

I have two replies. First, I want to say that in this case, as in the previous case, we do not in fact have an explanation of A's preferability at all. If there are millions of moral principles, one for each particular moral truth, and these millions of principles have no interesting connections with one another, then it seems to me they tell us nothing about why one particular action is preferable to another. Moral "facts" which can only be explained by such principles, I suggest, are moral "facts" which have no explanation and hence, according to my argument, not really facts at all.

Second, the idea that all moral principles are extremely specific may gain a specious plausibility by means of a faulty inference. It may well be that the totality of morally relevant features of an action A in fact suffices uniquely to determine A. And it may seem to follow from this that the only moral principles we can hope for will be highly specific. But this does not in fact follow. The morally relevant features of an action may be arranged hierarchically. If utilitarianism were true, for instance, then any feature of an action which affects its overall utility would be a morally relevant feature, and the sum of these features (e.g. making one person deliriously happy, hurting the feelings of three others, making a fifth feel slightly disgusted, and so on) might well be specific to that particular action. But these features are relevant because they affect the action's total utility; the total utility of the action is more likely to be shared by other actions than the specific features in virtue of which it has that total utility. And the action's total utility again is relevant only because of its contribution to a still more general feature, that of having more total utility than any alternative action available to the agent at the time; and this feature will surely be shared by many other actions.

EXPLANATION WITHOUT GUIDANCE

As I have mentioned, much of the literature which opposes principles seems to argue for conclusions about the role of principles in moral reasoning--that principles are unnecessary, or that they are pernicious, or that devices other than principles can serve moral reasoning as well or better--and then infer that there simply are no principles. Perhaps the inference is simply fallacious, a result of ignoring the explanatory role of principles. But one might suppose that the explanatory and action-guiding roles of principles are related in such a way that if principles are not needed for the latter role, they are also not needed for the former. In this section and the next, I will try to undermine two arguments for such a view.

One argument might run as follows. First premise: Moral principles are not useful, in fact they are often pernicious, as guides to what one ought to do. Second premise: If moral principles explained moral truths, then the principles would be useful in determining the right thing to do. Conclusion: moral principles do not explain particular moral truths.

I want to attack this argument at its second premise. But it is worth noticing first another difficulty with the argument: it seems to require an equivocation between two senses of 'principle'. The first premise makes the point that when people in fact try to arrive at decisions about what to do, they find sometimes that the principles they accept do not determine any particular course of action, and sometimes that these principles determine what seems intuitively or on other grounds to be a morally wrong course of action. 'Principle' in this context refers to whatever principles people happen to adopt: it means something like 'proposition believed by the agent to be a universal moral truth.' On the other hand, if the second premise is to be at all plausible, a moral principle must be taken to be, not something merely believed to be a universal moral truth, but something that actually is a universal moral truth. So unless the propositions we believe to be universal moral truths actually are universal moral truths, the argument involves a fatal equivocation.

This criticism may seem rather unfair to the intent of the argument. Critics of the view that there are moral principles often think that principles can at best serve as rules of thumb: that there are no exceptionless moral principles, only generalizations that happen to give the correct answer for a certain range of cases. That is, they think that there are no moral principles in the sense in which I am using the term. But now it seems that I am arguing that, in order to show that there are no moral principles, the critic of moral principles must assume that we know what the correct moral principles are!

We could present an argument truer to the intentions of the critic as follows: the best approximations we have to genuine moral principles are unhelpful or pernicious; if there were moral principles which explained particular moral truths, then they would be useful in determining the right thing to do; so there are no moral principles. This argument does not involve the self-defeating presupposition that we know the correct moral principles, but it does require the related assumption that our best approximations to moral principles are the best approximations there are, that there are no moral truths which are more like genuine principles than the truths we now know. Otherwise it may be that better approximations than we now have will be more useful as guides to action.

Now let us turn our attention to the second premise. It is easy to see why someone might think it to be true. A true moral principle provides something like an algorithm for determining particular moral truths: feed empirical facts in, get particular moral truths out. Such an algorithm seems almost to be a kind of machine for providing moral guidance. Then how could there be moral principles unless these principles were useful as guides to action? It seems all one would need to do is apply the principle. Nevertheless, the second premise is false, and the analogy with scientific principles will help explain why.

It should be clear that there may be scientific principles which explain without being useful for making predictions. How can this happen? In several ways. First, the principles may simply be too complex to be useful. There are many cases in the sciences in which a false approximation to the correct theory is more useful in prediction than the correct theory itself. In a wide range of cases Newtonian mechanics is more useful than relativity theory; older accounts of chemical bonding may be more useful for many purposes than the more complex account in terms of the number of electrons in the outermost orbital, and so on. Chaos theory provides intriguing examples in which the mathematical equations which explain a certain phenomenon, while fairly simple, may still be impossible to solve without superhuman resources.

Could ethics be like this? Certainly. It is entirely possible, for instance, that something like Robert Nozick's account25 of the various relations between right-making and wrong-making features of actions (e.g. overriding, outweighing, undercutting) is a part of the truth about how prima facie obligations influence one another, but I am not sure that I am capable of mastering his apparatus fully enough to find it useful in practice. An analogy with the law may also be helpful. The laws which determine whether my actions are legal or not occupy huge stacks of volumes. If I act illegally, the correct explanation of this will appeal to these laws, but it would be enormously difficult, if possible at all, to use an exhaustive knowledge of the law in trying to live without acting illegally. (It is hard enough to determine whether you have filled out your income tax return correctly.)

A second way in which the correct scientific principles might fail to be useful in prediction is that their use in prediction might require empirical information we do not have and perhaps could not obtain. I suppose the laws of physics that determine when California will be devastated by a huge earthquake are well enough understood, but applying them would require much information we do not have about conditions underground. (The correct application of these principles would also, of course, be impossibly complex.) Similarly, if utilitarianism were the correct moral theory, there might still be many occasions on which it was not useful because of the difficulty of determining the consequences of a particular action (or even assigning probabilities to the possible consequences).

There are also some more far-fetched ways in which a scientific theory might fail to be predictively useful. Perhaps one can imagine that the truth about human psychology is so awful that anyone who believed it would go mad, and thus be unable to use it to make predictions. Or perhaps if one were really convinced of the truth of behaviorism, one would become listless and inert, no longer interested in making predictions. These wild possibilities may be analogous to the following sort of case, due to Bernard Williams.26 It may be that benevolence is better if it is spontaneous; nevertheless attempting to follow the rule "be spontaneously benevolent" can only lead to failure.

I conclude that it is not the case that a correct set of moral principles would have to be useful in guiding moral action and hence that, even if it were true that our present moral theories were the best we could get, and that they were often not useful in guiding moral action, it would not follow that there are no true moral principles.

GUIDANCE WITHOUT EXPLANATION

Now let us consider another sort of argument from moral psychology or epistemology to the conclusion that moral principles are unnecessary. This argument might go as follows. First premise: there are mechanisms other than principles which enable us to "predict" the right thing to do (and which may even justify us in the belief that a particular action is right). Second premise: any mechanism which enables us to determine that (or justifies us in the belief that) a particular action is right must also explain why the action is right. Conclusion: moral principles are not needed to explain why particular actions are right (since something else can do the job).

Let us begin by considering the first premise. It is often argued that in fact people reason less from rules than from concrete examples. Bernard Mayo, for instance, writes that, in an ethics of character, the question what I ought to do "is answered, not by quoting a rule or set of rules, but by describing a quality of character or a type of person. . . . The heroes and saints did not merely give us principles to live by (though some of them did that as well): they gave us examples to follow."27 And Harold Alderman writes that "determining what one ought to do involves neither knowing and applying a rule nor specifying a good and predicting which course of action will most efficiently yield access to it. Rather one resolves moral ambiguity and makes choices by envisioning what some paradigmatic individual would do in the same situation."28

Is it true that moral reasoning does or should more often involve exemplars than rules?29 Much that is said about reasoning in other contexts supports the general claim that reasoning relies more heavily on exemplars than on rules. For example, Thomas Kuhn, in his classic The Structure of Scientific Revolutions, argues at length for his thesis of "the priority of paradigms": his view, that is, that scientists agree much more easily on paradigms of scientific rationality than on the rules which they exemplify, and thus that what constitutes good scientific practice is more a matter of similarity to exemplary cases than of conformity to rules.30 One solves problems less by applying rules than by coming to see the problems as like others previously encountered.31 Although there may be good reasons for adopting one rival scientific theory over another, "there is no neutral algorithm for theory choice, no systematic decision procedure which, properly applied, must lead each individual in the group to the same decision."32

It is an empirical question whether Kuhn is right about science, and how far similar remarks are true of our moral knowledge. But it certainly seems plausible that a good deal of our familiarity with the demands of morality consists in a collection of exemplars--of morally good people and of actions which are paradigms of integrity, benevolence, sympathy, courage, and other virtues. Surely in judging actions, and people, a large part of what we do is measure them against these exemplars. Our attempts to formulate more generally what makes these exemplars good or right may be crude and in many situations less morally helpful than the exemplars themselves.

There is at least some empirical work which makes this view plausible. Carol Gilligan has suggested that men are more governed by rules in their moral deliberation than are women--and that it would be a mistake to treat this as evidence that men have more highly developed moral sensitivities than women.33 Well-known work in cognitive psychology by Eleanor Rosch and others suggests that we typically remember concepts by remembering prototypes or exemplars rather than by remembering sets of features;34 if this is true in general then presumably it is true in particular of moral concepts. Also relevant may be evidence gathered by Hubert and Stuart Dreyfus that the development of expertise in a great many areas is a "progression from the analytic behavior of a detached subject, consciously decomposing his environment into recognizable elements, and following abstract rules, to involved skilled behavior based on holistic pairing of new situations with associated responses produced by successful experiences in similar situations."35

There are, then, reasons to think that reliance on concrete exemplars is an important component of much moral reasoning. It would be a mistake to move too quickly from this point to the view that our moral decision-making does not rely on rules or principles: it may be that our behavior is guided by sophisticated principles even though we are consciously entirely unaware of these principles; this seems fairly clearly to be the case with respect to our production of grammatical sentences. But let us waive this worry and grant the first premise of the argument we have been considering, namely that in much moral reasoning principles simply play no role, conscious or otherwise. The second premise of the argument holds that whatever mechanisms we successfully use in moral reasoning must explain particular moral facts. In my view this is completely mistaken, and a return to the analogy with science reveals why.

In the scientific case, it often happens that we can predict events of a certain kind without being able to explain them. As Bas van Fraassen puts it, "giving good grounds for belief does not always amount to explanation."36 Examples are numerous; van Fraassen mentions some familiar ones: for instance, that a barometer reading may enable us to predict the coming of a storm without explaining it, and that we can deduce the height of a flagpole from the length of its shadow and the angle of elevation of the sun, although this in no way provides us with an explanation of the height of the flagpole. I suggest that this is precisely the situation of moral reasoning which makes essential use of exemplars. If I know that Susan is a moral paragon, and I know that, were she confronted with my current moral situation, she would (say) return the money, then I have good reason to think that returning the money is the best thing to do. But the fact that Susan would do it in no way explains why it is the right thing to do. It is not the best thing to do because it is what Susan would do. Consequently the usefulness of moral exemplars in moral reasoning in no way obviates the need for explanatory moral principles.

CONCLUSION

This has hardly been an exhaustive sampling of the literature against rules. But the arguments I have considered are interesting and representative, and their failure illustrates the difficulty of countering the powerful intuition we have in favor of moral principles. Furthermore, my remarks can be extended in straightforward ways to apply to other attacks on rules; for instance, when J.B. Schneewind objects that one argument that rules are necessary falsely presupposes "that the reasoning needed in morality is purely deductive,"37 we may respond that our argument for moral principles, at least, presupposes nothing about moral reasoning, but only about the structure of moral explanation.

Someone who wants to combine the view that there are particular moral truths with the view that there are no general moral principles must explain how it is possible either that there could be moral truths which were simply inexplicable, brute facts, or that the explanations of particular moral facts are themselves inevitably particular. In the absence of an account which would make either alternative plausible, we are justified in our conviction that, to the extent that there are particular moral truths, there are moral principles which explain them.38

NOTES

1. Some examples: Harold Alderman, "By Virtue of a Virtue," Review of Metaphysics 36 (1982): 127-53, reprinted in Robert B. Kruschwitz and Robert C. Roberts, ed., The Virtues: Contemporary Essays on Moral Character (Belmont, California: Wadsworth, 1987), 47-65; Hans-Georg Gadamer, Truth and Method (New York: Seabury Press, 1975; a translation of Wahrheit und Methode, Tübingen: J.C.B. Mohr, 1960), especially 274-89; Eugene C. Hargrove, "The Role of Rules in Ethical Decision Making," Inquiry 28 (1985): 3-42; John McDowell, "Virtue and Reason," The Monist 7 (1979): 331-50; McDowell, "Are Moral Requirements Hypothetical Imperatives?" Proceedings of the Aristotelian Society, Supplementary Volume, 1978, 13-29; McDowell, "Non-Cognitivism and Rule-Following," in Steven H. Holtzman and Christopher M. Leich, eds., Wittgenstein: To Follow a Rule (London: Routledge & Kegan Paul, 1981), 141-62; Martha Nussbaum, "The Discernment of Perception: An Aristotelian Conception of Private and Public Morality," in J. Cleary, ed., Proceedings of the Boston Area Colloquium in Ancient Philosophy (New York: University Press of America, 1985), 151-201; Nussbaum, The Fragility of Goodness: Luck and Ethics in Greek Tragedy and Philosophy (Cambridge: Cambridge University Press, 1986), Chapter 10; Edmund Pincoffs, "Quandary Ethics," in Pincoffs, Quandaries and Virtues: Against Reductivism in Ethics (Lawrence, Kansas: University of Kansas Press, 1986), 13-36 (originally published in Mind 80 (1971): 552-71); John Waide, "Virtues and Principles," Philosophy and Phenomenological Research 48 (1988): 455-72.

2. Examples include Jonathan Dancy, "Ethical Particularism and Morally Relevant Properties," Mind 92 (1983): 530-47; Albert R. Jonsen and Stephen Toulmin, The Abuse of Casuistry: A History of Moral Reasoning (Berkeley and Los Angeles: University of California Press, 1988); Iris Murdoch, "Vision and Choice in Morality," Proceedings of the Aristotelian Society, Supplementary Volume 30 (1956): 32-58; Iris Murdoch, "Against Dryness: A Polemical Sketch," Encounter 16 no. 1 (January 1961): 16-20, reprinted in Stanley Hauerwas and Alasdair MacIntyre, eds., Revisions: Changing Perspectives in Moral Philosophy (Notre Dame: University of Notre Dame Press, 1983), 43-50; J. B. Schneewind, "Moral Knowledge and Moral Principles," in Knowledge and Necessity, Royal Institute of Philosophy Lectures, volume 3 (1968-69): 249-62, reprinted in Hauerwas and MacIntyre, 113-26. (For an interesting response to Dancy see Robert Gay, "Ethical Pluralism: A Reply to Dancy," Mind 94 (1985): 250-62.)

3. Compare e.g. Marcus Singer, "Moral Rules and Principles," in A. I. Melden, ed., Essays in Moral Philosophy (St. Louis: Washington University Press, 1958), 160-97.

4. For a classic statement of this distinction, see Carl G. Hempel, Aspects of Scientific Explanation (New York: The Free Press, 1965), especially 364-76. Hempel argues that "every adequate explanation is potentially a prediction" (a claim I will criticize) but regards it as an open question whether "every adequate prediction is potentially an explanation."

5. More generally, scientific laws enable us to draw conclusions about events past, present or future: prediction is not importantly different from retrodiction.

6. Robert Cummins, The Nature of Psychological Explanation (Cambridge: The MIT Press, A Bradford Book, 1983), 14-15. Related distinctions are drawn by others: Robert M. Armstrong contrasts laws of nature, which are relations between universals, with "scientific identifications," which are a matter of "exploring the geography" of a single universal, in What is a Law of Nature (Cambridge: Cambridge University Press, 1983), 137-40; and David Lewis distinguishes between the explanation of events, which is entirely causal, and the explanation of facts or truths, which is (in part) noncausal, in his Philosophical Papers, Volume II (New York: Oxford University Press, 1986), 223-24, 267-69.

7. There may seem to be something suspect about principles like "One ought not to break a promise." There may arise a sort of conflict between, say, "One ought not to break a promise" and "One ought to relieve suffering." Perhaps I can save the life of a drowning man by jumping into the water and helping him to shore, but only if I break my promise to meet a friend for lunch. It seems that I both ought to save the man, according to the second principle, and ought not to save him, according to the first. There are many possible responses to this difficulty. But it does not strike me as any more troublesome than the following, parallel "difficulty": If one heats a balloon, it will expand. If one lets air out, it will contract. Then what if one both heats it and lets air out? It seems to follow that it will both expand and contract! Of course in both the moral and the scientific case, the principles that lead to the apparent conflict are prima facie principles, or have implicit ceterus paribus clauses, or something of the sort. In both cases we hope that there are more general principles which explain the lower-level principles, and that these more general principles will also explain why there are exceptions to the more local ones.

8. Related distinctions have, however, been discussed in other regions of the ethics literature. See, for example, David O. Brinkís very similar distinction between moral theories as decision procedures and as criteria of rightness in Brink, Moral Realism and the Foundations of Ethics (Cambridge: Cambridge University Press, 1989), 216-17, 256-62. See also the references to Mill, Sidgwick, and others in Brinkís note 4, p. 216.

9. There are several recent discussions of the extent to which moral principles are explanatory, but these usually focus on whether moral facts are needed to explain our experiences or observations, not on whether moral principles are needed to explain particular moral facts. Examples include Geoffrey Sayre-McCord, "Moral Theory and Explanatory Impotence," Midwest Studies in Philosophy 12 (1988):433-57, reprinted in Sayre-McCord, ed., Essays on Moral Realism (Ithaca, New York: Cornell University Press, 1988), 256-81, and Nicholas Sturgeon, "Moral Explanations," in David Copp and David Zimmerman, eds., Morality, Reason, and Truth (Totowa, N.J.: Rowman and Allanheld, 1985), reprinted in Sayre-McCord, ed., 229-55. Both articles respond to a problem posed by Gilbert Harman in The Nature of Morality (New York: Oxford University Press, 1977), Chapter One, also reprinted in Sayre-McCord.

10. Waide, "Virtues and Principles" (see note 1).

11. Waide, perhaps following the lead of G. E. M. Anscombe's "Modern Moral Philosophy," Philosophy 33 (1958), sees the view that principles are central to morality as based on a "legal model": principles are like statutes. This may help explain his neglect of the metaphysical version of the view he rejects: laws are made up in response to a variety of sorts of pressures, so it is odd to think of them as explanatorily fundamental in the way I am suggesting moral principles are. In this respect I think the comparison developed here, between moral and scientific principles, is more illuminating than the comparison between moral principles and statutes. (Of course there are disanalogies in both cases.)

12. Saul Kripke, Naming and Necessity (Cambridge: Harvard University Press, 1980).

13. Compare Jonathan Dancy's illuminating remark about W. D. Ross: "In Ross, the drive to principle is not epistemological but metaphysical. He feels that an individual act cannot be right without there being some principle behind it, as it were" (Dancy, 543). This precisely expresses my own reason for thinking there must be moral principles. Oddly, having put the point so succinctly, Dancy recommends abandoning Ross's metaphysics without, so far as I can see, giving any reasons for this.

14. McDowell, "Virtue and Reason," 336.

15. Nussbaum, "The Discernment of Perception," 175; cf. The Fragility of Goodness, 302-3.

16. I am thus beginning with a moral quandary. It is sometimes suggested that it is a failure of modern ethics to take moral problems or quandaries as central: see e.g., Edmund Pincoffs, "Quandary Ethics," cited in note 1. But notice two things: (1) Whatever else ought to be included in our thought about moral philosophy, there are moral quandaries, situations of moral choice which require a reflective and reasoned decision, and discussion of these must be a part, whether or not a central part, of moral philosophy. (2) My focus on moral quandaries is not on how the agent is to choose the right action, but on what makes one action preferable to others.

17. The anti-rule theorists I am concerned with all seem to assume that there are objective moral truths, and some of them are explicit about this: e.g., Iris Murdoch lays great stress on the importance of truth, and John McDowell describes his view as "anti-non-cognitivism." See Murdoch, "Against Dryness," 46, 49; McDowell, "Non-Cognitivism and Rule-Following," 154.

18. I do not claim that this argument is particularly original. It is not hard to find related arguments: consider, for example, A. Phillips Griffiths, "Ultimate Moral Principles: Their Justification," in Paul Edwards, ed., The Encyclopedia of Philosophy (New York: Macmillan, 1967), vol. 8, 177-82. Griffiths writes: "If saying that someone ought to do something commits one to claiming that there is some fact in the situation which is a reason for doing the thing in question, then this reason must be subject to the requirement which reasons in general must satisfy: that anything which is a reason in any one case must be a reason in every case unless there are other special reasons for ignoring it. . . . It is this which leads to the claim that moral principles must be universal" (177). The term 'reason' may be ambiguous between an epistemological and a metaphysical sense, however; unlike other arguments with which I am familiar, the one I offer is explicitly metaphysical, and free of epistemological issues about what is required for someone to be justified in believing a particular moral claim.

19. Iris Murdoch argues persuasively against this latter view in "Vision and Choice in Morality."

20. See Richard C. Jeffrey, "Statistical Explanation vs. Statistical Inference," in Nicholas Rescher, ed., Essays in Honor of Carl G. Hempel (Dordrecht: D. Reidel, 1969), 104-113.

21. W. D. Ross, The Right and the Good (Oxford: Oxford University Press, 1930; Indianapolis: Hackett, 1988), 31.

22. Ross, p. 42.

23. I am being a bit careless about the details of Ross's views here. Ross always speaks of principles in their role as guides to action, so it is possible that he would agree that there must be universal moral principles, but deny that these principles will prove to be of use in concrete situation of moral difficulty.

24. For discussion of this distinction, see R. M. Hare, "Principles," Proceedings of the Aristotelean Society 72 (1972), and Hare, Moral Thinking: Its Method, Levels and Point (Oxford: Clarendon Press, 1981), 41. Hare's way of drawing the distinction is criticized in Marcus G. Singer, "Universalizability and the Generalization Principle," in Nelson Potter and Mark Timmons, eds., Morality and Universality (Dordrecht: D. Reidel, 1985), 49-53. John Rawls uses the same terms to mark a different distinction in A Theory of Justice (Cambridge: Harvard University Press, 1971), 131-33.

25. Robert Nozick, "Moral Complications and Moral Structures," Natural Law Forum 3 (1968): 1-50; compare Nozick, Philosophical Explanations (Cambridge, Mass.: Harvard University Press, 1981), 474-94.

26. Bernard Williams, "A Critique of Utilitarianism," in J.J.C. Smart and Bernard Williams, Utilitarianism: For and Against (Cambridge: Cambridge University Press, 1973), section 6.

27. Bernard Mayo, Ethics and the Moral Life (London: Macmillan, 1958), 213-214. Mayo qualifies his apparent defense of exemplars as against rules, however, writing: "I am not suggesting that the subordination of exemplars to principles is incorrect, but that it is one-sided and fails to do justice to a large area of moral experience" (214-15).

28. Alderman, "By Virtue of a Virtue," 61.

29. Strictly speaking, this lumps together two distinct but related questions: the empirical question of how we in fact reason, and the normative question of how we ought to reason. But for my purposes here it is unnecessary to disentangle these issues.

30. Thomas Kuhn, The Structure of Scientific Revolutions, Second Edition (Chicago: University of Chicago Press, 1970), Chapter V.

31. Kuhn, 189.

32. Kuhn, 200.

33. Carol Gilligan, In a Different Voice (Cambridge: Harvard University Press, 1982).

34. See, e.g., E. Rosch, "On the Internal Structure of Perceptual and Semantic Categories," in T. Moore, ed., Cognitive Development and the Acquisition of Language (New York: Academic Press, 1973); E. Rosch and C. Mervis, "Family Resemblance Studies in the Internal Structure of Categories," Cognitive Psychology 7 (1975). There is an interesting discussion of Rosch's views in George Lakoff, Women, Fire, and Dangerous Things (Chicago: University of Chicago Press, 1987), 39-54. For a recent survey of work on concepts, including Rosch's work, see Edward E. Smith, "Concepts and Induction," in Michael I. Posner, ed., Foundations of Cognitive Science (Cambridge: MIT Press, 1989), 501-26.

35. Hubert L. Dreyfus and Stuart Dreyfus, Mind Over Machine (New York: Macmillan, The Free Press, 1986), 35.

36. Bas C. van Fraassen, The Scientific Image (Oxford: Oxford University Press, 1980), 104.

37. J. B. Schneewind, "Moral Knowledge and Moral Principles," 114.

38. I have been helped by discussion with my colleagues at Trinity University, especially Steven Luper; by written comments from Martin Gunderson and Jonathan Dancy; by Peter Winch's response to the paper at an APA meeting; and by discussion with John Waide. I am especially grateful to Kadri Vihvelin for extremely helpful comments and correspondence. I am also indebted to Trinity University for a Summer Stipend which supported the writing of an ancestor of this paper.