Philosophy has always been concerned with the largest questions of what exists and what is the case. Not specific questions like “Is there extraterrestrial life?” or “What is the speed of light in a vacuum?” but maximally general questions about what kinds of things exist and what kinds of facts there are—basic kinds that cannot be defined in terms of anything still more basic, and from which everything else is constituted. This is the subject matter of metaphysics. A related question, how we can know about such things, is the subject matter of epistemology.
There is disagreement about how far a philosophical account of existence, truth, and knowledge can depart from the assumptions that underlie ordinary, commonsense thought and discourse about what there is, what is the case, and what we know. Radical doubt is not unusual: the basic reality of many things—matter, mind, space, and time—has been called into question at one time or another; but among the domains or areas of thought that are perennially believed to pose problems, and that continue to be the focus of much discussion today, are morality and mathematics.
We frequently make moral judgments—that it was wrong of X to accuse Y of a theft of which X was in fact guilty, or that helping someone to commit suicide is not always wrong—but what, if anything, could make these judgments true or false? They seem not to describe what exists or happens in the world but to say something further. What kind of fact could this be? We also have mathematical beliefs, for example, that 7+8=15. Of course we use these beliefs in practical calculations about the material world, but we also think they are true in themselves—timelessly true. But what are they about, and what makes them true? The number 7 itself doesn’t seem to be part of the natural world we see around us.
It is characteristic of our age that moral truth and mathematical truth are found philosophically puzzling because they seem so different from scientific truth about the natural world. Existence, truth, and knowledge in the natural sciences are regarded as relatively straightforward, though they depend on taking for granted the validity of the observational evidence and logical or mathematical reasoning used in the confirmation of scientific theories. Natural science is widely thought to provide a standard for robust reality that is difficult for the objects of moral judgment and pure mathematics to meet. That is why it is so easy to wonder whether numbers or wrongness really exist, outside of our thoughts.
Such doubts depend, however, on a crucial assumption. They assume that there is a universally applicable standard of existence and truth that determines the reality or unreality of any type of thing or fact. This means that the reality of what we are talking about when we talk about morality or mathematics must be established by appealing to a standard from outside those domains of thought—a standard that determines whether numbers and wrongness are really part of the fabric of the universe. Since the natural sciences are assumed to satisfy this supposedly universal standard, that leads to the search for things and facts analogous to but distinct from those described by natural science, which make the statements of morality and mathematics true. And that seems a fairly hopeless enterprise.1
In his short, powerful book Being Realistic about Reasons the distinguished philosopher T.M. Scanlon sets out, among other things, to refute this assumption and to present an alternative approach to the question of what, ultimately, is real—what kinds of things most basically exist and what kinds of facts are most basically true.2 I have introduced the examples of morality and mathematics because they are important in Scanlon’s discussion, but his main topic is that of normative reasons—reasons of the kind that specify what should or should not be thought or done. He is particularly concerned with reasons for action, of which morality and moral reasons form a special subpart. The comparison with mathematics is present here, as it has been in earlier writings of Scanlon, because of parallels between the philosophical problems posed by these two areas, and parallels between what he regards as the solutions to those problems.
The general category of the normative encompasses everything about what people should or shouldn’t do or think, for reasons of any kind. These include reasons of self-interest, of sympathy, and of morality in the sphere of action, and reasons of empirical evidence and deductive logic in the sphere of belief. The life of a rational being is pervaded by normative thought about what to do and what to think, in light of all the data and considerations bearing on the choices available. Normative thought is an essential part of our lives, and Scanlon holds that the irreducible core of such thought is found in the concept of a reason—a reason in favor of or against doing or thinking something.
Consider the following statement:
For a person in control of a fast moving automobile, the fact that the car will injure and perhaps kill a pedestrian if the wheel is not turned is a reason to turn the wheel.
The normative character of this reason becomes clear if we contrast it with an explanatory reason. We can say that the reason why the driver turned the wheel was that he wanted to avoid hitting a pedestrian. That explains his action. But the normative reason—the reason to turn the wheel—is present whether the driver turns the wheel or not. It is a reason why he should turn the wheel. The normative reason enters into an explanation of action only if the driver recognizes it and is motivated by that recognition to act on it.
Scanlon calls his view Reasons Fundamentalism. He says the normative statement above is obviously true (who could deny it?), and he contends that it is true in itself and is not equivalent to any other statement that does not itself make use of the concept of a reason. Scanlon holds that we all understand what it means for a reason of this kind to exist: the circumstances of the driver count in favor of turning the wheel. To ask what in the world makes this the case, demanding an answer that does not itself include a reference to reasons, is simply confused. Nothing else makes it the case, nothing that is not itself a normative truth—though often one can derive specific reasons from more general ones. Reasons are sui generis, and can be understood only from within the domain of normative thought. Much of the book is taken up with patient and trenchant refutations of the many ways philosophers have found of resisting this conclusion.
One such way is to interpret truths about normative reasons as truths about human motives—for example, it has been claimed that for someone to have a reason to do a is for a to be an action that would contribute to satisfying some desire that the person has—or perhaps a desire that he would have if he were fully informed about the facts. As Scanlon says, the trouble with this analysis is that it leaves out the specifically normative character of reasons—their “should” content. Reasons don’t just say how people are or would be motivated: they say how we should be motivated, and that is simply a different kind of truth. If, on the other hand, the desire theory is interpreted as a genuinely normative claim, to the effect that what we have reason to do is what would satisfy our (factually informed) desires, then, first, it employs the concept of a reason and does not provide an analysis of reasons in purely psychological terms, and, second, it is not a credible normative claim, since a mere desire for something worthless does not give a reason to do what will satisfy it (unless satisfaction of the desire would produce a further effect, such as pleasure, which itself provides a reason).
The analysis of reasons in terms of desires is an attempt to show that normative statements are not basic by finding conditions for their truth in the sort of natural facts that a science like psychology can identify. A different strategy for analyzing the normative in naturalistic terms is to abandon the search for truth conditions altogether, and to interpret normative statements in such a way that they are neither true nor false because, despite appearances, they do not assert that anything is the case. This theory, which goes back to Hume, is called expressivism; it holds that normative statements, about what reasons there are, what we should and shouldn’t do, think, or feel, are expressions of the speaker’s attitudes, rather than claims that could be true or false. In other words, when you say that the driver has a reason to turn the wheel, or that it would be wrong not to turn the wheel, you are expressing your endorsement of his turning the wheel or your distress at the thought of his not doing so, or something of the kind—whether the driver is you or someone else.
Scanlon discusses sophisticated contemporary versions of expressivism such as that proposed by Allan Gibbard.3 He argues that these views are incapable of accounting for the role of normative discourse in giving advice or justification to others, or for the process of trying to decide what reasons we ourselves have. First, if someone objects to my having done a, and I reply by pointing to the fact that p, and asserting that p was a good reason to do a, this will not serve as a justification if it merely expresses my personal feeling or attitude.
For example, suppose I turn the wheel of a borrowed car to avoid hitting a cat, and as a result I hit a tree and crumple a fender. Then p is the fact that if I hadn’t turned the wheel, I would have killed the cat, and my justification to the car’s disgruntled owner is that that was in fact a good reason to turn the wheel. But this normative claim is a justification only if, in Scanlon’s words, I am
calling that person’s attention to what I claim to be a fact, independent of both of us, about what one has reason to do…. Whether the other person agrees with it or not, my claim that p was a reason for me to do a responds to his challenge, in a way that expressing my acceptance of a norm, or a plan, or an attitude of approval does not.
Second, whatever attitudes we ourselves may have (with regard to the value to be put on the lives of nonhuman animals, for example), we can always go on to ask whether we should have those attitudes, whether they are justified by good reasons. And that question is not answered just by noting that we have them (nor by finding that we have other attitudes that endorse them, since the same question can be asked about those higher-order attitudes). The search for justifying reasons, whether for ourselves or for others, has to be the search for something that is true independent of us and the attitudes we happen to have. The aim is to get it right—to bring our attitudes into conformity with the truth about what we should do and think, if we can discover it. That is the point of normative thought.
But what kind of truth is this? Scanlon’s positive conception of the normative is part of a larger pluralistic conception of reality, i.e., of existence, truth, and knowledge:
Accepting science as the way of understanding the natural world entails rejecting claims about this world that are incompatible with science, such as claims about witches and spirits. But accepting a scientific view of the natural world does not mean accepting the view that the only meaningful statements with determinate truth values are statements about the natural world….
I believe that the way of thinking about these matters that makes most sense is a view that does not privilege science but takes as basic a range of domains, including mathematics, science, and moral and practical reasoning. It holds that statements within all of these domains are capable of truth and falsity, and that the truth values of statements about one domain, insofar as they do not conflict with statements of some other domain, are properly settled by the standards of the domain that they are about. Mathematical questions, including questions about the existence of numbers and sets are settled by mathematical reasoning, scientific questions, including questions about the existence of bosons, by scientific reasoning, normative questions by normative reasoning, and so on.
The phrase “insofar as they do not conflict with statements of some other domain” is crucial. Scanlon holds that statements about the existence of reasons, numbers, and the collections called sets do not conflict with any true statements of natural science. It is not one of the claims of natural science that natural science contains all the truths there are, and statements about reasons and numbers, unlike statements about witches and ghosts, do not contradict anything that science tells us about the natural world and its causal order.
However, one might ask, if reasons and numbers, moral and mathematical truths are not part of the natural world in which we live, how can we know anything about them? If they are not in space and time, like stars and molecules, how can we make contact with them, how can they affect us? But this is just the old illusion raising its head again, the illusion that all existence, truth, and knowledge must conform to a single model: if these things have no spatial or temporal location they must exist “outside of space and time,” and it becomes a mystery how we can have access to them. “But here,” as Scanlon says, “the spatial metaphor has simply gotten out of hand.” Instead of conjuring up a mysterious region to whose spooky occupants we could gain access only by a magical analogue of perception, what we need to understand is how we come to know certain familiar kinds of truth just by thinking about them in the right way.
Scanlon illustrates the systematic investigation of an autonomous domain with the examples of arithmetic and set theory. We employ precise concepts such as “successor” and “member” to formulate rules for the construction of numbers or sets, and axioms that seem clearly correct and whose extensive further implications seem undeniable even after careful reflection. The concepts identify the kinds of truth involved, but most of the truths are not conceptual—i.e., they do not follow from the meaning of the terms in which they are stated.
Though the domain of reasons is entirely different in subject matter, and much less orderly, we also learn about it just by thinking about it in the right way—clarifying the relevant concepts, identifying statements that seem clearly true on reflection, formulating general principles that explain them and that seem true in themselves, and testing those principles against the fruitfulness and credibility, upon careful reflection, of their further consequences. Scanlon provides examples: in this book he offers a convincing account of the concept of a reason and the logical structure and boundaries of the normative domain, and in his 1998 book, What We Owe to Each Other, he presented a theory of the place of morality in the larger domain of reasons. But for reasons of space I will discuss only his general epistemological account of the method of thought involved, which applies to both mathematical and normative judgments.
The method was given the name “reflective equilibrium” by John Rawls, who adopted it for his work in moral and political philosophy but noted that it has far wider application. It is a way of extending and rendering systematic one’s understanding of a domain, by searching for an “equilibrium” in which convincing general principles account for the more specific judgments that one finds undeniable on reflection.
An example from moral thought is the complex process of reflection prompted by apparent counterexamples to the utilitarian principle that it is always right to do what will produce the best overall outcome. Would it be right, for example, to frame and execute one innocent man for an alleged interracial or interreligious crime if that is the only way to pacify an enraged mob that will otherwise attack an entire community, killing many innocent people? The considered judgment that this would be wrong even though it would result in fewer innocent deaths suggests several different hypotheses at the more general level:
(1) Perhaps every harmless individual has an inviolable right not to be killed merely as a means to some desirable end, and perhaps the general principles of morality have to include universal rights of this kind, and not just requirements to minimize the number of deaths.
(2) Perhaps the execution of this one innocent man would corrupt the legal system and lead to further abuses, so that the balance of deaths now may not reflect its true cost; in that case the judgment that the act would be wrong may be supported by utilitarian principles after all.
(3) Perhaps our aversion to killing the innocent man is an instinctive reaction that for biological or cultural reasons is found in human beings because of its general value in preventing violence; and perhaps if we recognize the general usefulness of the instinct, we should be content to act on it without trying to calculate in each case whether this will have the best consequences.
(4) Perhaps our instinctive reluctance to kill the innocent man, whatever its causes, is just wrong in this case, and should be overridden by utilitarian calculation, provided we can really be confident of the results.
There are also other possibilities, and this is only the beginning. Each general hypothesis must be evaluated for its intrinsic plausibility and the acceptability of its further consequences. How would a right not to be killed have to be modified to allow self-defense, or to allow causing one death as a side effect of an action that saves many lives? Does the general usefulness of a disposition against killing, or of the institution of due process of law, justify our adhering to it case by case regardless of consequences? Can we come to regard our intuitive judgment of wrongness in this case not as an apprehension of the truth, but as a psychological quirk, akin to a phobia or an optical illusion? One tries to arrive at a stable conclusion by going back and forth between these judgments at different levels of generality, constructing alternative general hypotheses and comparing them in the light of their different implications.
The search involves repeated adjustments and revisions, at both general and specific levels, and each step is an exercise of judgment within the moral domain. The process is open-ended, and equilibrium is an ideal, rather than a static end point we can expect to reach. But someone who remains convinced of the wrongness of framing the innocent man will be led by this process to accept the general principle that seems to offer the most plausible justification of his judgment, on reflection. Sometimes, as in this case, people will disagree in their judgments of the correct result; but disagreement in judgments can be found in almost any area of inquiry, not just the normative.
The pursuit of scientific knowledge through the interaction between theory and observation has an analogous structure: we test theories against their observational consequences, but we also question or reinterpret our observations in light of theory. (The choice between geocentric and heliocentric theories at the time of the Copernican revolution is a vivid example.) The difference is that in mathematical or moral thought, instead of perceptual observations of events spatially distant from us, we start from considered judgments about specific statements within the domain (like a statement about the reason to turn the wheel, or not to frame the innocent man).
The method depends, as Scanlon says, on the assumption
that we have reason to believe the things that seem to us to be true about these subjects when we are thinking about them under the right conditions, and hence that it is appropriate for us to treat these beliefs as considered judgments in a process of seeking reflective equilibrium if we have no apparent reason to doubt them.
This seems like a platitude about how to pursue the truth, but it can be extraordinarily difficult to accept, or to understand. How can the mere fact that something seems to us to be true be a ground for believing that it is true? If seeming is all we have to go on, doesn’t that mean that objective truth is unreachable or nonexistent, and all we can do is to explore the insides of our own minds?
But note that this form of skepticism is so general that it would destroy all thought, since the entitlement to believe what on careful reflection seems to be the case, where there is no reason to doubt it, is the necessary condition for being able to form any justified beliefs at all. Science would be impossible if we could not form warranted beliefs about the world around us on the basis of perception, which is just the way things seem when we use our senses. How things seem is the starting point for all knowledge, and its development through further correction, extension, and elaboration is inevitably the result of more seemings—considered judgments about the plausibility and consequences of different theoretical hypotheses. The only way to pursue the truth is to consider what seems true, after careful reflection of a kind appropriate to the subject matter, in light of all the relevant data, principles, and circumstances.
If someone objects that this shows that we are never talking about objective truth, but only about the realm of appearances, the reply is that that skeptical claim, too, is presented as something we should believe because it seems true on reflection. It relies on the very principle of belief that it denies, making an exception for itself from the general skepticism it proposes. Furthermore, it does not survive the test of seeming to be true on careful reflection, for when we reflect on the skeptical hypothesis it proves unbelievable: the things that seem to be the case—about the natural world, about morality, or about mathematics—continue to seem to be true in themselves and not just as descriptions of the inside of our minds.
The internal pursuit of reflective equilibrium is the positive aspect of Scanlon’s method for discovering the truth in a given domain. But there are also, importantly, two ways in which we may find that there is much less truth there than we had thought. This is important because it means that the standards for truth and knowledge are not too easily met.
The first possible ground for doubt is internal: it may turn out that the effort at systematic elaboration and explanation of our judgments within a domain fails repeatedly to arrive at stable conclusions—that nothing coherent emerges when we try, in discussion with others, to adjust general principles to particular judgments and vice versa. In that case we may be forced to conclude that we are not talking about an objective subject matter at all, but are merely giving voice to a set of subjective responses. (Something like this seems to have happened with the attempts in the nineteenth century, by figures as eminent as William James and Henry Sidgwick, to put communication with the dead on a scientific basis.)
This is not a problem for mathematics, given its extraordinary fruitfulness, order, and capacity to command unanimous conviction, but it must be taken seriously as a problem for morality, and for some aspects of practical reason in general. Scanlon’s view is that we cannot hope for the kind of universal order and agreement in moral thought that we find in mathematics, and that the basis for confidence in determinate truth-values for statements about reasons will have to be piecemeal. He rejects any single all-encompassing moral principle like utilitarianism or Kant’s categorical imperative. There are many claims we can be confident about, but we have no grounds for confidence that questions about what there is reason to do always have determinate answers. By contrast, we are confident that questions of arithmetic always have correct answers, even though we can’t always discover them. (As Kurt Gödel proved, no consistent set of axioms can yield as theorems all the truths of arithmetic.)
The second challenge, alluded to earlier, is that claims that seem to be supported by judgments internal to one domain may conflict with the truths of another domain. Not all domains are self-contained in the way that mathematics is. Systems of religious belief, for example, may include claims about the natural world—e.g., about the creation of the universe—that are subject to refutation by the standards appropriate to that other domain.
Scanlon maintains that normative statements, including those of morality, do not entail or presuppose claims about the natural world that science gives us good reasons to reject. He says that the significance of the normative domain depends on there being rational agents—people who can act for reasons—but he believes “that rational agents are just a kind of natural organism, and that organisms of this kind do exist.” Rational agents are capable of being motivated by their beliefs about normative facts, but this “does not involve causal interaction with these facts.” Driving home the separation of the normative from the causal, Scanlon says:
When a rational agent does something that he or she judges him or herself to have reason to do, this judgment makes sense of the action in normative terms and explains it, because the action is what one would expect of a rational agent who accepted that judgment. Presumably there is also a causal explanation of this connection, and of the more general uniformities…in virtue of which a being is a rational agent. But this causal explanation is another story, for neuroscientists to fill in.
I believe this carries the project of demystification too far. Normative rationality seems to require a kind of explanation of action and belief that consists neither of their merely “making sense” nor of their having certain neural causes. A rational being, unlike a calculating machine, responds to reasons because they exist and he recognizes them. How this explanation is to be understood is a deep question; but wherever the truth on this question may lie, Scanlon’s defense of the reality of reasons is a great advance in our understanding of the subject.
A leading example of this type of skepticism is J.L. Mackie’s Ethics: Inventing Right and Wrong (Pelican, 1977). ↩
His general outlook has much in common with those of two other important recent works of moral theory, Ronald Dworkin’s Justice for Hedgehogs (Harvard University Press, 2011) and Derek Parfit’s On What Matters (Oxford University Press, 2011). ↩
In Wise Choices, Apt Feelings: A Theory of Normative Judgment (Harvard University Press, 1990) and Thinking How to Live (Harvard University Press, 2003). ↩