1Over the last two decades, one proposal has breathed new life into meta-ethical debates about the status of normative properties. [1] The challenge is to get away from the age-old conflict between realism and subjectivism about values, and to dissociate the adoption of cognitivism—the position that there is such a thing as moral knowledge—from any claim about value realism. Traditionally, the discussion stood as follows: either we hold that there is such a thing as moral knowledge, in which case we must be realists about values and normative properties; or else we are not realists about them, in which case we must reject the possibility of moral knowledge.

2The situation has changed with the defense of a form of cognitivism without value realism, proposed in similar forms by a number of authors, chief among them T.M. Scanlon, Derek Parfit, and John Skorupski. These authors subscribe to the buck-passing account of value [2]—that is, to the theory that what ultimately gives reasons for being in favor of x is not the fact that x has value, understood positively (we are dealing here with moral, prudential, or esthetic value); on the contrary, x is valued because x has natural properties [3] that, under such circumstances, give us reasons to have (evaluative, conative, practical) attitudes in its favor. [4] For instance, if some artifact in a museum deserves our admiration, this is not just because it is “good” or “interesting”—for, while it is these things, it is so in virtue of structural and functional properties that, while artistic, are not evaluative. When we say that these properties, which are beyond (or below) values, give reasons to value the thing in question in these circumstances, we must understand them as justifying reasons and not (or at least not necessarily) as motivating ones.

3Scanlon’s theory combines positive and negative considerations about the relationship between values, reasons, and natural properties. In his positive considerations, he explains values using something more fundamental than them, that is, the relation of justifying certain attitudes that is supported by reasons given by lower-order properties. In his negative considerations, he eliminates values as higher-order properties, or at least denies that they can provide reasons. [5]

4Parfit offers substantial nuance to Scanlon’s theory on at least two points. First, while he leaves Scanlon’s positive considerations untouched, he moderates the negative ones. For Parfit, an esthetic or ethical value can count as a reason to appreciate something or to act in a certain way. For instance, the fact that a film is good is a reason to go see it again. But this only happens derivatively, and the normative cost must ultimately be paid by properties of the film that are not esthetic values stricto sensu, but instead artistic properties. Parfit makes non-eliminative use of the buck-passing account (hereafter BPA). [6] There is no contradiction between his adoption of BPA and the importance that he grants to the sort of reasons he calls “value-based.” [7]

5Second, Parfit often privileges the vocabulary of facts rather than that of properties, and presents BPA in a (so to speak) de-ontologized, epistemologized way. He conceives of facts as truths. BPA mobilizes natural (non-evaluative) facts that provide reasons to adopt certain attitudes. So we might say that normative facts are, once again, part of the ethical furniture, and of the philosophy of values more generally—but in the form of a justificatory function establishing a relation between natural facts, agents, actions or attitudes, and so on. It is a fact (1) that the fact (2) that this person (or animal) is suffering gives me a reason to try to help them. Fact (1) is normative. But fact (2) is non-normative—something that does not stop it from having “normative importance.” [8]

6As Skorupski remarks, there are at least two major objections to BPA. [9] The first concerns “the wrong kind of reasons.” [10] The second accepts that there is an equivalence between characterizations in terms of values and of reasons, but claims that BPA interprets this equivalence incorrectly. [11] The present article discusses this second objection, in order to defend a particular version of BPA.

7Let us take the example of a tool. [12] On the second objection, if I have good reason to judge that this is a good wrench, and to buy it, this is because this wrench is truly good. And, because the fact that this is a good wrench explains why I adopt this favorable attitude towards it (something we will happily admit if we abandon BPA’s negative thesis), the explanation must work in this direction—because the other way round it would explain nothing at all. And to say (following BPA’s positive thesis) that this wrench is good because we have reasons in favor of it explains nothing at all, because all we have done is repeat that this wrench is a good wrench.

8But this objection is missing an important point. According to BPA, when I judge that this is a good wrench, one that is worth the money, it is ultimately because of its finish, materials, ergonomics, and shape that I have a sufficient reason to view it as such. My appreciation can be explained by the facts that justify it. The fact that the wrench has these properties gives a reason to count it as a good one, and an explanation of the fact that I consider it a good one. [13] It is not the fact of having a reason that is explanatory, but rather the fact(s) that give this reason. What would we think of someone who said, “I’ve got a reason to start volunteering as an after-school tutor,” but who will not tell you what the reason is? And what would we think of someone who said, “I’ve got a reason: it’s good”? The first response does not explain anything at all. The second one really explains very little. BPA explains much more, and matches the way in which we can justify a belief, an action, or a judgment by the content of our reasons.

9The facts that provide certain reasons may, however, be in competition with other facts. And the reasons themselves may consequently be compared and contrasted. It is important to note that other facts can give me a reason to consider the wrench to be a good one and, if it is for sale, to buy it: for instance, the fact that it is much less expensive that a slightly more finely made wrench from another brand. Or, for instance, because it was made in Europe—a fact which, as a patriotic consumer, I am particularly sensitive to. Or, say, the fact that this wrench will complete my collection of tools made by this brand.

10In any case, it is true, as Parfit claims in On What Matters, that facts provide reasons. An interesting question which then arises is how they do so. We may say that reasons consist in facts, and that some fact is a reason to believe that p, or to act in a certain way, but this is an elliptical formulation, and we should not thereby conclude that reasons are (natural) facts, but instead that they are provided by facts.

11For instance, there is an objective, very important reason not to smoke: the fact that tobacco causes cancer. We should not smoke for the reason that tobacco is carcinogenic. The reason and the fact seem to be identical. But there is an important difference. The fact F is a reason for the agent A (whether they currently recognize it or not) to φ (to carry out an action or to adopt an epistemic attitude) in particular circumstances. This relation between F, A, and φ is supported by a reason. Saying that F is this reason is just shorthand for saying that F provides this reason. Why should I not smoke? Because tobacco is carcinogenic. That tobacco is carcinogenic is a fact, but it only provides a reason insofar as it allows us to answer the question, “Why should I not smoke?”

12Defenders of reasons fundamentalism are happy sometimes to write that “reasons are facts.” [14] And this is certainly sometimes true. One reason to go to the doctor when I am sick is the fact that I am worried about my health. These facts “about me,” which are subjective insofar as they involve my beliefs or desires, are indeed reasons for me. If we only had reasons of this sort, the defenders of reasons internalism (who reduce reasons to psychological states) would be quite right. The fact that there are reasons to x, and the fact of having such psychological states, would be identical. But there are also facts—even facts “about me”—that are not subjective like this, and that give me reasons. So the fact that I am sick has an objectivity that is distinct from the fact that I am worried about my health, or from the fact that I want to get better. We could say, elliptically, that facts of this sort are also reasons, [15] but what this means is that, as natural facts, they have (in Parfit’s words) “normative importance.”

13If we follow Parfit in distinguishing between normative facts that are reasons and non-normative but normatively important facts that give these reasons, a question inevitably arises: by what mysterious process do they provide reasons? Are we required to think that normative facts “supervene” on natural facts, or that while reasons are primary they are “grounded” in non-evaluative properties? Or is there some other way to understand this relationship? The question here is not about the content of reasons or of facts—for instance, figuring out whether the contribution to collective well-being is a more important fact than the satisfaction of my desires—but instead about formal aspects of the relationships between facts, reasons, and appropriate attitudes. [16]

Do facts alone provide reasons?

14A first question is: do the facts alone provide reasons? The question looks a little like a Humean one: can Reason (with a capital R, in the sense of the faculty of reasoning) alone influence actions and passions? But the resemblance is only superficial. For the Humean, since Reason is the mind’s power to represent to itself facts about the world, and to reason about them, it can have such an influence only derivatively. Reason tells me that tobacco is carcinogenic. But, according to the Humean, this only has authority over my behavior if one of my motivations is already the fear of becoming sick. (Big-R) Reason does not by itself provide a (little-r) reason to act, whether we understand the latter as a motivating or a justifying reason.

15Against this, Parfit argues that in order for natural facts that can have normative importance (that is, those possessing the property of giving reasons) [17] to speak to us, we must have, not desire or feelings, as Humeans think, but the normative concept of reasons, [18] without which we could not access normative truths. In this sense, Parfit is a rationalist and not a sentimentalist.

16With Parfit, and against Hume, we must subscribe to the thesis that facts, which can be described as natural facts, provide little-r reasons that have authority over behavior. But, against Parfit—or, more charitably, against the silences and gaps in his account—we should argue that they do not provide these reasons all by themselves. Such facts “count in favor” (as Scanlon and Parfit say) of an action. But do they count by themselves, all alone, or, as Scanlon puts it in one of his questions to Parfit, “atomistically”? [19] No. It is not that they only count subject to our desires, which would be a variant of the Humean thesis. According to Scanlon, they count in a practical context, in a particular situation of action that is not “internal” in the way a set of desires or motivations is. It is facts in context—in the context of a decision, or one that may lead to a decision—that provide reasons.

17This context is not just that of the particular situation of the action. It also takes in other reasons. It seems to me that there is also something that may occupy a place between facts and epistemic or practical responses: considerations about facts. These considerations may even potentially be integrated into instances of reflection, argument, and deliberation (whether inter- or intrapersonal), which may themselves be informed by substantive moral views. In no way is this a challenge to reasons fundamentalism, that is, the thesis that the concept of reason is both simple [20] and primitive, something shown by our inability to analyze it without using it. [21] Considerations respond to reasons and depend on them. Inversely, however, the importance granted to certain facts insofar as they provide reasons depends on these considerations. Particularly when they form part of an instance of practical reflection, considerations give them different weights which may vary depending on the course that deliberation has taken. We may call this the deliberative context of the force of reasons. The force of a reason—the fact that one reason can override others—is not a property that a reason possesses by itself, independently of any deliberative context. This gives us a motivation for resituating the question of reasons in a deliberative perspective which Parfit neglects (in favor of a more demanding model of moral knowledge), or which he reduces to a purely psychologistic interpretation of Bernard Williams’ position.

18Parfit argues that facts provide reasons in the same way that truths may justify attitudes, beliefs, and actions. Isabelle gives all her change to someone who is clearly very down on their luck, for the reason that they are clearly very down on their luck. Paul approves of her action, and would have done the same thing if he had had any change, but this time for the reason that there is something very appealing about this down-on-their-luck person which satisfies Paul’s social needs and desire for interaction. Isabelle and Paul do not have the same reasons. That means that they do not justify their actions or attitudes by invoking the same reasons, but instead using different reasons. But we may argue that it is not true that the person’s appealingness constitutes a normative reason. Paul has a subjective justification that does not correspond to the objective justification Isabelle employs, and that may mean he would not give money to an unappealing person. Moreover, if we adopt Parfit’s position, we can claim that Paul is wrong about the facts, and that Isabelle is right. Paul does not take the important facts into account.

19Against this Parfitian view of moral knowledge as a form of non-probabilistic knowledge which is potentially as certain as mathematical intuition, we may object that the context in which facts can provide reasons is very often one of deliberation, particularly when it is a question of moral or prudential reasons in a situation that demands a degree of reflection. That does not mean that facts only provide reasons in such a context, or always do so through deliberation. That is wrong. The sharp pain I feel in my foot may give me a reason to stop walking which is outside any deliberative context. Nor does it mean that reasons only have authority over an agent insofar as they are currently or potentially integrated into the agent’s deliberation. And nor does it mean that reasons are reducible to psychological entities. Deliberation itself is, from one point of view, a psychological process, and more than that from other points of view, like those of logic and rhetoric.

20To borrow an argument from Thomas More’s sailor-philosopher, Raphael Hythloday: [22] for highwaymen who do not kill those they rob, the prospect of nonetheless being sentenced to death may encourage them to leave no witnesses, and to systematically kill all their victims; this fact is a reason for the authorities not to impose the death penalty or, if it is already in place, to modify it; but it is not a decisive reason in and of itself, independently of any deliberation. Cardinal Morton’s remark is not be dismissed lightly: “They would look on a mitigation of the punishment as an invitation to commit crimes, almost a reward.” [23] Rule consequentialism, act consequentialism, deontological ethics, and so on, are repertoires of arguments in favor of the normative importance of certain facts which, during practical reflection, can affect the relative weight of the reasons provided by the facts.

How do facts provide reasons? In search of a foundation

21We have an opportunity here to answer a more difficult question: how do facts provide reasons? What is the status of the relationship between facts and reasons? Some argue that reasons supervene on facts, in virtue of the undeniable truth that, in identical circumstances, if nothing else has changed, a variation in reasons presupposes a variation in the facts that provide them. At best, reasons supervene on facts in their contexts, and not on facts all by themselves. What should concern us is the question of whether the supervenience relation means that facts, or natural properties, are the ontological basis for reasons, or whether they are just their epistemological basis, that is, the source of the justificatory function that they fulfill.

22We might talk about these reasons supervening on facts or natural properties. But such language nudges the epistemological problem of the content of a justification towards ontology. If we subscribe to a form of reasons fundamentalism that entirely assimilates reasons to second-order (normative) properties, it is appropriate to talk about supervenience. But if we maintain a relational conception of reasons, in which facts provide justifications for practical or doxastic responses, such talk seems less appropriate.

23The important point is that the question of the supervenience of reasons arises in completely different terms from that of the supervenience of values or, more generally, of ethical properties. There is an apparent affinity as, in all these cases, we are describing a relation of anchoring in the natural, in the sense of the non-moral. But facts provide reasons. What we mean is that having a reason is being able to justify—to others, necessarily—a decision, attitude, practice, and so on, using these facts. The fact that an animal is suffering before our eyes is a reason to try to help it. The fact that animals suffer in abattoirs is a reason to think that we must reform or even close such places. Here, the fact that p is not just the non-ethical basis on which an ethical entity—in this case a moral reason—supervenes or is grounded. It is not a matter of facts providing a foundation [24] or a mere explanation, but of facts providing a justification.

24Reasons fundamentalism leads to non-naturalism insofar as it avoids the question of the foundation of reasons themselves, and pays little attention to the question of their supervenience as non-natural properties on natural ones. It would be bizarre to ask about reasons the questions that we normally ask about ethical properties and their links with non-ethical properties. Yet Peter Railton and other authors successfully tempted Parfit towards naturalism, a position for which he showed some sympathy. [25]

25However, Parfit claims that truths about facts that provide reasons have no “ontological” implications, in a sense of the term that is very restricted and primarily involves current existence in some time and place. [26] If we take “ontology” in the sense it has in the philosophy of mathematics, reasons fundamentalism certainly does have ontological implications. But what are they? According to Parfit, facts do not have a causal role, because they belong to the order of truths and not to that of events or objects in the world. But it is tempting to base the normative relation on one of supervenience, rather than a causal one. Parfit seems to hesitate on this point [27]—a hesitation encouraged by his concerns about reconciling his views with other meta-ethical positions.

26Let us change tack and approach the question from a logical point of view. We can treat claims about reasons as premises the conclusion of which may be evaluative judgments or practical preferences, and do so without necessarily crossing the Humean divide between “is” and “ought.” [28] If a non-normative premise constitutes a reason, it is because there is a normative element, in the broad sense of the term, before the conclusion. The conclusion derives from the conjunction of the non-normative premise and this normative one, based on the following scheme:

27Non-normative premise P1 (the object of empirical observation, assuming happiness can be measured): in these circumstances, this policy maximizes happiness.

28Normative premise P2: in these circumstances, the fact that this policy maximizes happiness is a sufficient reason in its favor.

29Normative conclusion C: this policy should be adopted in these circumstances, unless there are stronger reasons to the contrary. [29]

30What ontology corresponds to these two premises? One solution is Parfit’s non-naturalistic cognitivism. The non-normative premise corresponds to a natural fact that mobilizes only natural properties. These natural properties are “normatively important,” of course, but the scheme above only comes about if this natural fact is taken into consideration by a being that is sensitive to reasons. The normative premise corresponds to a normative fact that mobilizes an irreducibly normative concept, that of a reason. Parfit’s ontology is best expressed in the language of facts, understood as truths. But it can also be reconciled with the thesis that normative properties supervene on non-normative ones.

31But Railton’s moderate naturalism explains the above scheme just as well. P1 presupposes natural properties. P2 mobilizes the irreducibly normative concept of a reason, without presupposing the existence of non-natural properties or primitive normative facts. Normative distinctions supervene on natural facts. [30] At the lowest level, we have (natural) properties; and there is no space for properties in the ontological sense of the word at the higher level because normative concepts are enough. Moderate naturalism can even accommodate talk about normative facts at the higher level, but only in the sense that facts are truths. [31] In Parfit’s own view, these normative facts do not have the normative importance that natural facts providing reasons have. [32]

32This is why a compromise between Railton and Parfit that relegitimizes the language of supervenience and foundations will also serve. As Railton writes:

33

After all, as Parfit agrees, the things of normative importance include plainly natural things like pain, happiness, accurate belief, and so on. Fix these things, and you have fixed what the reasons-making facts are. [33]

34The least we can say is that the ontology of BPA is flexible, and that firming it up is not one of Parfit’s core concerns. We should ask, then, if debates about whether reasons are grounded on facts obscure an equally important issue, about the connection between reasons and practical, action-oriented reflection.

Another way out: How facts provide reasons in practical deliberation

35Even if we turn away from the metaphysical temptations of “grounding” reasons, we can account for the role of the concept of reason and its connection with facts by attending to its function within instances of practical argumentation and deliberation. Why should a justification have as its counterpart non-natural properties, in addition to some normative fact? Deliberation uses natural facts as material for arguments which can justify its practical conclusion. This is a new motivation for resituating reasons within a deliberative perspective—which is not necessarily the same thing as internalizing reasons or making them into psychological entities.

36Parfit cannot accept this argumentative reduction. He claims that moral knowledge consists in taking the normatively important natural facts correctly into account, such that it is guided towards its conclusion by truths and, more specifically, by the conjunction of synthetic a priori truths (for instance, that suffering is to be avoided) and empirical truths (for instance, that the person in front of me is suffering greatly). We may object that, on this conception, fundamental normative truths guide deliberation, and the concept of reason becomes almost superfluous, at least if we think of it in the classical (notably Lockean) sense of a reason to believe. On Parfit’s conception, the normative truth that another person’s suffering gives us a reason to help them results from the conjunction of a synthetic a priori truth about suffering and truths about natural facts. In this case, the practical conclusion is reached without the mediation of any supplementary reason for holding that this conclusion is appropriate—or, at least, without the mediation of a merely probable reason. Taken to an extreme, such a position runs the risk of contradicting reasons fundamentalism and justifying the replacement of the concept of reason with that of inference or deduction—something that is of course undesirable.

37Would it not be preferable—without falling back into reasons internalism—to consider the domain of reasons as not entirely external to a deliberative perspective? Lying between the Platonic objectivity of truths and the subjectivity of motivations, reasons may have the intermediate status of arguments, which, in themselves, are neither reducible to psychological entities, nor simply deducible from first truths, even though the best of them are taken from the facts.

38One advantage of this way of thinking is that it meets one requirement of the triviality argument that Parfit uses against naturalism. [34] According to this argument, [35] if the naturalist is right—that is, if having a normative property (for instance, being socially just) could be identified with having a natural property (for instance, the disposition to reduce inequalities in means of living)—then any policy that possessed the natural property would obviously and trivially also possess the normative property. This is precisely the sort of identification that people often challenge. This remark on the controversial nature of the concept of justice points towards an argumentative conception of reasons.

39But such a conception is still compatible with the reasons externalism (or “objectivism”) that Parfit so vigorously and unwaveringly defends. [36] He is extremely wary of “intersubjective” varieties of subjectivism, which, he thinks, erect a mere facade of objectivity. And he is even more wary of deliberative theory, which he reduces to a psychologistic version of Williams’ position. Parfit does not think a different version of deliberative theory can avoid reasons internalism—which, fatally, does justice to the concept of a motivation, but not to that of a (normative) reason itself. Nonetheless, insofar as it mobilizes a “substantial” rather than a “procedural” conception of rationality, this other version can maintain reasons externalism.

40Parfit attacks Williams for reducing normative truths to psychological facts by integrating them into a conception of deliberation in which rationality is viewed only procedurally—whereas it should instead be conceived of in an equally substantial way. [37] The purely procedural view indicates that what we have reason to prefer is simply that which we would discover that we prefer over the course of rational deliberation. The substantial view admits that over the course of rational deliberation we can discover what we have reasons to prefer, but it specifies that this is not what makes it true that we have these reasons. Indeed, the opposite is true: as Parfit writes, “when it is true that we have decisive reasons to act in some way, this fact makes it true that if we were fully informed and both procedurally and substantively rational, we would choose to act in this way.” [38] I understand this distinction as follows. If we are both to defend reasons externalism and the integration of such reasons within a deliberative perspective, or within practical deliberation, we must understand the thesis that correct deliberation is guided by reasons de re (meaning: there are reasons, and correct deliberation will recognize them), rather than de dicto (meaning: whatever correct deliberation recognizes as a reason is a reason). That is, reasons must be identifiable independently of deliberation and, in this sense, must be external to it. I do not see why a deliberative theory of reasons could not meet this requirement.

41On the question of the relative weight of these reasons—for instance, whether some reason is decisive when set alongside other considerations—should we not claim, against Parfit, that this question is internal to deliberation, or at least internal to that reflection that guides our normative considerations and debates? What is important is determined outside of practical reflection, but the degree of importance is not. [39]