What moral realism can learn from the philosophy of time * Heather Dyke University of Otago This is a preprint of an ar.cle whose final and defini.ve form is published in Heather Dyke (ed.) Time and Ethics: Essays at the Intersec2on. Dordrecht: Kluwer Academic Publishers, 2003: 11‐25. 1. Introduction It sometimes happens that advances in one area of philosophy can be applied to a quite different area of philosophy, and that the result is an unexpected significant advance. I think that this is true of the philosophy of time and meta-ethics. Developments in the philosophy of time have led to a new understanding of the relation between semantics and metaphysics. Applying these insights to the field of meta-ethics, I will argue, can suggest a new position with respect to moral discourse and moral reality. This new position retains the advantages of theories like moral realism and naturalism, yet is immune to many of their difficulties. 2. The tenseless theory of time The tenseless theory of time claims that there are no tensed facts. The old tenseless theory tried to prove this by showing that tensed expressions could be eliminated from natural language.1 It claimed that any tensed sentence (a sentence locating an event or state of affairs somewhere in the past, present or future) could be translated, without loss of meaning, by a * I am grateful to Alan Musgrave and Charles Pigden for extremely helpful comments on an earlier draft of this paper. 1 Proponents of the old tenseless theory include Goodman (1951) Quine (1960) and Smart (1963). tenseless sentence (a sentence locating an event or state of affairs in the static B-series). It concluded that, since tensed expressions were not needed to completely describe reality, there is no feature of reality to which they refer. That is, if tense can be eliminated from language without any loss of meaning, that shows that there is, in reality, no distinction between past, present and future, and no flow of time. As it turned out, the old tenseless theory of time was wrong about the possibility of eliminating tense from natural language. It is not possible to translate tensed sentences into tenseless sentences without some loss of meaning. 2 The tensed theory of time always denied that tensed expressions can be eliminated from natural language without some attendant loss of meaning.3 It concluded that, since tensed expressions are needed to give a complete description of reality, there is a feature of reality to which they uniquely refer. That is, there really is an objective distinction between past, present and future, and time really does flow. The new tenseless theory of time accepts that tense cannot be eliminated from natural language, but denies that time itself is tensed.4 Tensed expressions, those that reflect the distinction between past, present and future, and the associated flow of time, merely express features of our representations of temporal reality, rather than picking out features of temporal reality itself. Thus, according to the new tenseless theory, tense is a feature of language that picks out no feature of reality. If the new tenseless theory is right, it follows that tense is both irreducible (it cannot be eliminated from language without some loss of meaning) and nonreferring (there is nothing in reality to which it refers). 2 See, for example, Mellor (1981) and (1998) and Dyke (forthcoming, 2002) 3 Proponents of the tensed theory include Gale (1968) and Schlesinger (1980). 4 The new tenseless theory of time was first developed and put forward by Smart (1980) and Mellor (1981). Oaklander and Smith (1994) tracks and develops the debate between the tensed and the new tenseless theories of time. The debate between the tensed and the old tenseless theories of time can be illustrated by considering the following argument: Argument A 1. No tensed sentence can be translated by a tenseless sentence without some loss of meaning. (Premise) 2. Either tensed sentences are translatable by tenseless sentences without loss of meaning, or true tensed sentences reflect a tensed reality (i.e. are made true by tensed facts). (Premise) 3. Therefore, true tensed sentences are made true by tensed facts. (From 1 and 2) 4. Some tensed sentences are true. (Premise) 5. Therefore, there are tensed facts. (From 3 and 4) According to the tensed theory this argument is both valid and sound. If some true sentence makes an irreducible reference to the pastness (say) of an event, that must be because there exists a fact about the pastness of that event which makes that sentence true. So argument A establishes the existence of tensed facts, facts about the pastness, presentness and futurity of events. The old tenseless theory thought that argument A was valid, but unsound because premise 1 is false. Its proponents offered a variety of translation schemas, which purported to show how any tensed sentence could be replaced by some tenseless sentence without any loss of meaning. Their reasoning was that if tensed sentences are translatable by tenseless sentences, then they are not made true by tensed facts. Instead they are reducible to sentences that are made true by tenseless facts. So the only facts needed to account for everything that can be said by both tensed and tenseless sentences are tenseless facts. But these attempts failed, because premise 1 is true. The new tenseless theory also takes argument A to be unsound, but it rejects premise 2. Tensed sentences are not translatable by tenseless sentences, but it’s not the case that the only alternative to this is that they are made true by tensed facts. It is possible for there to be true tensed sentences that cannot be translated by tenseless sentences even if there are no tensed facts. A tensed sentence can be irreducible, in that no tenseless sentence can capture the entire meaning conveyed by it, while still being made true by a purely tenseless fact. The implications of the new tenseless theory of time for the semantics and the metaphysics of tense are as follows. Metaphysically, temporal reality is constituted by the temporal relations of ‘earlier than’, ‘later than’, and ‘simultaneous with’. Reality is temporally ordered according to these, and only these, relations. Any tensed sentence (one which appears to locate an event somewhere in the past, present or future), if true, is made true by some tenseless fact (a fact about the temporal relations that obtain between events). For example, a token of the tensed sentence “It rained yesterday,” if true, is made true by the tenseless fact that the token is produced one day later than a day on which it rained. All tensed sentences, according to this theory, have tenseless truth conditions and truthmakers. Their truth conditions make no reference to tense, and their truthmakers include no facts about anything’s pastness, presentness or futurity. It follows that there is no observerindependent feature of reality that corresponds to tense in language. But it does not follow from this that tense can be eliminated from language. Any tensed sentence has tenseless truth conditions, but it does not have the same meaning as the tenseless sentence that states its truth conditions.5 For example, a token, u, of the sentence “It rained yesterday” is true if and only 5 See, for instance, Dyke (forthcoming, 2002). if rain occurs one day earlier than the day on which u is produced. But “It rained yesterday” does not mean the same as “it rains one day earlier than u.” So, no tensed sentence can be translated by a tenseless sentence. It follows that tense constitutes a significant and irreducible aspect of language and thought that has no ontological counterpart. 3. Moral realism: its motivation and two problems Moral realism is a theory about the status of moral discourse that has both semantic and metaphysical components. Its metaphysical component is that there are distinctively moral facts and moral properties. Its semantic component consists in cognitivism about moral judgements. Cognitivism is the doctrine that moral judgements express propositions that are capable of truth and falsity. According to moral realism, moral judgements are not only capable of truth and falsity, some of them are actually true. The principal source of motivation for moral realism is the idea that there are correct answers to moral questions, and that when we argue with each other about the correct answer to a particular moral question, we are engaged in a genuine disagreement. If there are correct answers to moral questions, it seems to follow naturally that this is because the correct answers correspond to the way things are, independently of what anyone happens to think. This in turn suggests that there is a way things are morally, not just a multitude of moral opinions, and that we can be genuinely either correct or mistaken in our moral beliefs. Thus, the domain of moral discourse is treated by the realist in a way similar to ordinary, factstating discourse. If I ask “What is the population of New Zealand in 2000?” there is a correct answer to this question. I can discover that the correct answer to this question is that the population of New Zealand in 2000 is 3.8 million. The proposition that New Zealand’s population in 2000 is 3.8 million is a true proposition that corresponds to the facts, independently of what anybody happens to think. Similarly, according to moral realists, if I ask “Is euthanasia ever morally permissible?” there is a correct answer to this question, and I can discover what it is. There is a fact of the matter about the moral permissibility of euthanasia that corresponds to the way things are independently of what anybody happens to think. There is an obvious attraction in thinking that moral discourse is objective in the way just described. If moral disagreements express mere differences of opinion, then there can be no hope of resolving them. Neither could there be any grounds for attributing praise or blame. I could not be praised for rescuing a drowning child, just as I cannot be praised for liking strawberry ice cream. However, one of the most serious objections to moral realism is that it severs the link between moral judgements and our motivations to act in accordance with them.6 For while discovering facts about the world can cause me to change my beliefs, it cannot cause me to change my desires, but it must do so if discovering moral facts is to play a role in motivating me to act.7 If I believe that it is right to rescue drowning children, and I find myself nearby a drowning child, I will be motivated to rescue that child. Having the moral belief just is my motivation to act in accordance with it. But simply having a certain belief is not normally sufficient for having a motivation to act. I can discover any number of facts about the world, but by themselves they will not cause me to act in one way rather than another. It is only in conjunction with a desire that a belief can provide me with a motivation to act one way rather than another. Thus, if moral judgements correspond to moral facts, we are left without an explanation as to how they motivate us to act in accordance with them. 6 7 See Smith, (1994) See Smith (1991). Another problem for moral realism, indeed a central problem in meta-ethics quite generally, is to present an account of the relation between fact and value. Hume (1738) is credited with establishing that there is a logical gap between statements of fact and those of value. Essentially, what this amounts to is that no conclusion about what one ought to do can be derived, in a valid argument, from premises just about what is the case. Or, to put it another way, no moral conclusion can be derived from non-moral premises; no “ought” from “is.” G. E. Moore (1903) relied on this logical gap in his “open question argument” that moral propositions cannot be reduced to natural, or non-moral, propositions. To think otherwise, he argued, is to commit the “naturalistic fallacy.” Moore believed that ethics is autonomous; that truths of morality cannot be reduced to truths of any other, non-moral kind. He argued that no moral predicate, such as “good” or “right,” is identical with, or reducible to, any non-moral predicate, for instance, “pleasant” or “desirable.” Suppose we took the predicate “good” to denote the same property as that denoted by “pleasant.” It would follow from this identification, Moore argued, that finding out that something is pleasant would be sufficient to establish that it was good. But it is not. Even if something is pleasant, it remains an open question whether or not it is good. Thus, according to Moore, no moral property is identical with any non-moral property; moral properties are non-natural, sui generis properties. Moore’s “open question argument” has certainly been challenged, 8 but any moral realist must address the charge that her position commits her to a metaphysically unattractive realm of real, unanalysable, non-natural moral properties. An alternative response to the is/ought distinction is to argue as follows. If moral propositions cannot be reduced to non-moral propositions, and the truth of some moral 8 See, for instance, Pigden (1991). propositions requires there to be a range of distinctively moral properties, then it simply follows that no moral propositions are true. This response can be developed in two different ways. Firstly, one could take the line of Ayer (1936), Stevenson (1944), and Hare (1952), and argue that moral discourse does not play the same kind of role in our language as ordinary, fact-stating discourse, and so should not be assessed in the same kind of way. Moral propositions do not describe reality, but rather, they prescribe, in some way or another. To say of some state of affairs that it is good, or of some action that it is right, is not to describe it, and thus attribute the property of goodness or rightness to it. Instead, it is to express one’s approval of it, or one’s desire that it be brought about. Thus, the entire thrust of Moore’s open question argument is avoided by claiming that moral discourse should not be taken to be describing reality, so it is not making any claims about what sorts of facts there are. Instead, moral discourse plays a different kind of role in our language. It expresses approval and disapproval; it is prescriptive rather than descriptive. The second kind of response along these lines argues that moral discourse should be interpreted as purporting to describe reality, but since reality is not as it describes, it is systematically in error. This also involves rejecting the premise that some moral propositions are true. However, rather than reinterpreting moral discourse, it claims that, since there are no distinctively moral properties, any proposition that makes reference to such properties must be false. This is the line taken by Mackie (1977), who goes on to present an account of our moral discourse and practices that makes no reference to moral facts. Yet another response to the is/ought distinction is to challenge Moore’s contention that it establishes that if moral properties are real, they must be non-natural. This involves responding directly to Moore’s arguments that any naturalistic account of moral properties commits the naturalistic fallacy. Pigden (1991) carefully dissects the assumptions supporting Moore’s arguments and argues that they do not establish the falsity of naturalism. He distinguishes between three forms of the autonomy of ethics: logical, semantic and ontological. Logical autonomy is the thesis that no “ought” can be derived from an “is.” Semantic autonomy is the thesis that moral terms do not have the same meaning as any nonmoral terms. Ontological autonomy is the thesis that what makes any true moral judgement true is some sui generis moral fact (Pigden (1991) 427). Pigden argues that only ontological autonomy entails the falsity of naturalism, but that Moore’s arguments, at best, merely establish semantic autonomy. So, it seems that there is still room for a moral realist who denies that there is a realm of moral facts over and above the facts that can be described using non-moral language. However, any naturalist moral realist faces a challenge. She must give a reductive analysis of the moral to the non-moral. This will involve presenting an account of the nature of the naturalistic properties to which moral predicates refer. To sum up then, the is/ought distinction, together with the threat of committing the naturalistic fallacy, seem to leave available only the following positions on the conceptual landscape with regard to moral discourse and reality. Firstly, one can hold that moral propositions are to be taken as describing reality, and that reality is as they describe. Moral propositions make reference to distinctively moral properties and facts, which cannot be identified with natural properties and facts, so reality contains a realm of non-natural moral facts. Secondly, one can hold that moral propositions, despite appearances, are not to be taken as describing reality, but instead merely express our attitudes towards various actions and states of affairs. Thirdly, one can hold that moral propositions are to be taken as describing reality, but since there are no moral properties, moral predicates are empty and propositions containing them are false. Lastly, one can argue that the logical and semantic autonomy of ethics does not establish the falsity of naturalism; some kind of reduction of moral properties to non-moral properties is possible. Of these options, only the first and last are available to a moral realist. So the problem for moral realism is that it seems forced either to countenance strange, non-natural moral properties, or to provide an adequate analysis of moral facts and properties in terms of non-moral facts and properties. 4. The argument for moral realism Characterisations of moral realism typically suggest that one arrives at it from an adequate understanding of the implications of its semantic thesis. For example, Geoffrey Sayre-McCord (1988) characterises moral realism9 as a commitment to the following two theses: (1) Moral claims, when literally construed, are literally true or false (cognitivism), and (2) Some moral claims are literally true. (Sayre-McCord (1988) 5) From (1) and (2) it is taken to follow immediately that, since moral claims purport to state facts, and some of them succeed in doing so (i.e. are true), there exist in reality some moral facts. A very tight connection is taken to hold between the semantic and metaphysical theses of moral realism.10 We might present the argument for moral realism as follows: Argument B 9 Sayre-McCord presents a characterisation of realism in general, which consists of claims (1) and (2) bereft of any reference to morality. I have added the references to morality in order to keep my discussion focussed on moral realism. 10 See also Smith (1991) p. 402 and Brink (1995) p. 511. 1. No moral predicate is identical with, or reducible to any non-moral predicate. (Semantic autonomy of ethics) (Premise) 2. If (1), then no assertoric sentence containing a moral predicate (a moral sentence) can be translated by any sentence that contains no moral predicates without some loss of meaning. (Premise) 3. Therefore, no moral sentence can be translated by any non-moral sentence without some loss of meaning. (From 1 and 2) 4. Moral sentences are capable of truth and falsity. (Cognitivism) (Premise) 5. Either moral sentences are translatable by non-moral sentences without loss of meaning, or true moral sentences reflect a moral reality (i.e. are made true by moral facts). (Premise) 6. Therefore, true moral sentences are made true by moral facts. (From 3, 4 and 5) 7. Some moral sentences are true. (Semantic thesis of moral realism) (Premise) 8. Therefore, there are moral facts. (From 6 and 7) The first thing I should note is that I introduce the term ‘moral sentence’ here to refer to sentences that contain moral predicates or expressions. The introduction of this term is intended to preserve the parallel with argument A. Elsewhere I refer to moral judgements, and by this I mean the act of asserting a moral sentence. The inference from 1 and 2 to 3 in argument B is a distillation of my discussion in section 3 of the impact of the is/ought distinction on moral realism. If moral properties are not reducible to non-moral properties, then moral discourse cannot be eliminated from natural language without some loss of meaning. But then, given that moral sentences are fact-stating, and that some of them are true, it follows that there is some feature of reality which moral discourse uniquely picks out, viz., moral facts. In the next section I will discuss the similarities between arguments A and B, and between the available responses to them. 5. The semantics and metaphysics of tense and morality I shall first comment on the differences between arguments A and B. Firstly, Argument B contains a sub-argument from B1 and B2 to B3, to which there is no counterpart in argument A. B3, the claim that moral sentences cannot be translated by non-moral sentences without some loss of meaning, is analogous to A1, the claim that no tensed sentence can be translated by any tenseless sentence without some loss of meaning. The fact that there is no corresponding sub-argument in A merely reflects a difference in focus in each of these areas of debate. In meta-ethics the focus has been on the is/ought distinction and the import of the naturalistic fallacy. Debate in the philosophy of time has focussed directly on whether or not it is possible to eliminate tense from natural language. Secondly, there is no counterpart in A to premise B4. This is because it has not seriously been questioned that when someone says, for instance, “It is now 4 o’clock,” what one says has a determinate truthvalue. What has been up for debate is what makes such sentences true or false. The nature and function of moral discourse in natural language and in our lives, by contrast, has led some to question whether or not cognitivism about moral sentences is appropriate. Thus, this difference between the arguments is a result of the different contents and functions of moral and tensed discourse. Leaving aside these differences then, we can restate argument B so that it is structurally analogous to argument A as follows: Argument B’ 1. No moral sentence can be translated by any non-moral sentence without some loss of meaning. (Premise) 2. Either moral sentences are translatable by non-moral sentences without loss of meaning, or true moral sentences reflect a moral reality (i.e. are made true by moral facts). (Premise) 3. Therefore, moral sentences are made true by moral facts. (From 1 and 2) 4. Some moral sentences are true. (Premise) 5. Therefore, there are moral facts. (From 3 and 4) In the debate about moral realism, just as in the debate about tense, it has been unquestioningly assumed that acceptance of cognitivism about moral sentences, together with the claim that some moral sentences are true, automatically commits one to the existence of a realm of moral facts. When this is combined with Moore’s arguments about the irreducibility of moral predicates, the moral realist has found herself committed to a profligate ontology of non-natural moral facts. The conceptual landscape surrounding argument B’ has traditionally offered the following habitable positions. One can accept the argument as valid and sound, perhaps arguing that the conclusion is not as unpalatable as it seems. One can reject premise 4 as well as the suppressed premise that moral sentences are capable of truth and falsity, thus taking a non-cognitivist approach to moral discourse. Alternatively one can reject premise 4 while retaining cognitivism, thus arguing that moral sentences are all false, and developing an error theory of moral discourse. Lastly, one can reject premise 1, which is the strategy of naturalism. Naturalism adopts a cognitivist approach to moral discourse, holds that some moral sentences are true, but argues that there are no facts or properties over and above those that can be specified using non-moral language. So, any true moral sentences must be analysable in terms of non-moral sentences. Premise 2 has not yet been called into question. Compare the responses to argument B’ with those to argument A. The tensed theory accepts argument A as valid and sound, and is thus the conceptual counterpart to the first response to B’ noted above, which we might call genuine moral realism. According to the old tenseless theory, the argument is valid, but premise 1 is false, because tensed sentences are reducible to tenseless sentences. This response is the conceptual counterpart of the last response to argument B’, viz., naturalism. There are no counterparts to either the second or third responses to B’. The reason for this, as I noted above, is that no one has seriously questioned either the claim that tensed sentences are capable of having a truth-value, or that some of them are actually true. 11 It is hard to think what alternative there could possibly be to cognitivism about tensed discourse, and an error theory for tensed discourse would just about have to be an error theory for natural language, since almost everything we say is tensed. There was a third response to argument A, which was to deny the disjunctive premise. The new tenseless theory accepts the truth of both A1 and A4, but denies A2. There can be true tensed sentences that cannot be translated by any tenseless sentence even if there are no irreducibly tensed facts. Irreducibly tensed sentences have tenseless truth conditions. If any of them are true, their tenseless truth conditions are fulfilled by tenseless facts. Could an analogous position be developed with respect to argument B’? Such a position would deny B’2. It would hold that it is possible for there to be irreducible, truth-assessable moral sentences, and for some of them to be true, while there are no moral facts serving as their truthmakers. It would hold that it is possible for irreducible moral sentences to be made true 11 This is not entirely true. In a forthcoming article L. Nathan Oaklander has recently defended the thesis that all tensed sentences are logically false. See Oaklander (2002). by non-moral facts, despite the fact that they are not reducible to non-moral sentences. This position is to naturalism what the new tenseless theory is to the old tenseless theory, so might appropriately be dubbed “new naturalism”. However, I don’t think the name “new tenseless theory” is particularly helpful as it doesn’t describes the position it names. A more helpfully descriptive name for this meta-ethical position is: non-reductive naturalism. 12 In the next section I will say some more about how non-reductive naturalism might be filled out. 6. Non-reductive naturalism Non-reductive naturalism is a framework for an ethical theory that can be filled out in different ways, depending on the sorts of non-moral facts one takes to be the truthmakers for moral propositions. It is beyond the scope of this paper to consider which sorts of non-moral facts might best do this job. My aim here is merely to establish that the framework is a viable one, and furthermore to show that it has some advantages over its rivals. The first thing to say about non-reductive naturalism is that it is a realist position with respect to morality. A typical characterisation of moral realism is that given by Sayre-McCord (1988), who states that it involves commitment to two theses, viz., (1) Moral claims, when literally construed, are literally true or false (cognitivism), and (2) Some moral claims are literally true. Non-reductive naturalism accepts both these claims. Where it differs from traditional accounts of moral realism is in its rejection of any peculiarly moral facts. There is an assumption rife in the literature that commitment to (1) and (2) automatically commits one to the existence of moral facts. After all, the assumption seems to be, if moral claims are irreducible to non-moral claims, truth-assessable, and some of them are true, doesn’t that 12 I thank Charles Pigden for suggesting this name to me. necessitate the existence of moral facts to make the true ones true? It is just this assumption that is rejected by non-reductive naturalism. Secondly, despite their obvious similarity, there is an important difference between naturalism and non-reductive naturalism. Naturalism is a reductive doctrine. It denies the existence of moral properties (goodness, rightness, wrongness etc.), and reduces talk of moral properties to talk of other, non-moral properties. Similarly, the old tenseless theory is a reductive doctrine, arguing that talk of the pastness, presentness or futurity of something is reducible to its standing in some temporal relation to something else. The new tenseless theory, by contrast, denies that there is any sense in which tense is reducible to anything else. It holds instead that tense is an irreducible feature, not of the world, but of our representations of the world. If non-reductive naturalism follows the new tenseless theory in this regard, it will hold that moral properties are not reducible to non-moral properties, but that morality is somehow an irreducible feature, not of the world, but of our representations of it. To make the distinction between naturalism and non-reductive naturalism more apparent, consider the following naturalist position. A semantic naturalist might hold that, while the moral predicate “good” cannot be replaced in all contexts salva veritate by the nonmoral predicate “pleasant” (for example), because the two terms are not synonymous, nevertheless both predicates stand for the same property. So the kind of reduction from the moral to the non-moral endorsed by semantic naturalism is the same as that between the ordinary language term “water” and its scientific counterpart “H2O”. In a scientific reductions such as this, it is recognised that the two terms do not have the same meaning, but they have a common referent. So the ordinary language term is reducible to the scientific term in so far as it refers to nothing over and above what the scientific term refers to. However, no such claims for a reduction are made by the new tenseless theory of time, or by non-reductive naturalism as I am developing it. The new tenseless theory of time does not claim that “past” and “earlier than” share a common referent. Rather, tenseless relations figure in the truthmakers for true tensed sentences. A component of tenseless reality is the sole truthmaker for some tensed sentence, but it does not fully account for the sentence’s meaning. 13 Similarly, non-reductive naturalism says that non-moral properties figure in the truthmakers for moral sentences, but they do not fully account for their meanings. Non-reductive naturalism says that moral expressions contribute, at a semantic level, to the meanings of moral sentences. It is a naturalist doctrine, in that it postulates nothing in the world over and above what can be described using nonmoral expressions, but it is not a reductive doctrine, as naturalism usually is. The peculiar ‘tensedness’ of tensed language and thought emerges out of the fact that we experience the world and talk about it from some perspective on tenseless reality. Since our perceptions of and interactions with reality are always from some temporal perspective, we tend to locate other events and states of affairs by implicit reference to our own temporal location. Thus, the peculiar ‘tensedness’ of tensed language consists in its being implicitly and essentially temporally self-locating. This is one reason why tensed sentences cannot be translated by tenseless sentences, since the latter are not self-locating. It is open to nonreductive naturalism to argue that the peculiar ‘moralness’ of moral language and thought emerges out of the fact that we bring a moral perspective to bear on a non-moral reality. The feature of tensed language that is not captured by a statement of its tenseless truth conditions is, to use Kaplan’s (1989) terminology, its character. The character of a tensed sentence is a function from its context of utterance to its content in that context. A proponent 13 See Dyke (forthcoming, 2002). of non-reductive naturalism could argue that moral sentences have a character as well as a content.14 The character of a moral sentence might be a function from a feature of its context of utterance to its content in that context, where the salient feature of context for moral sentences is a normative system. If the character of a moral sentence is a function from normative systems to moral propositions with determinate truth values, then when a person asserts a moral sentence she commits herself to a number of normative systems, viz., those that the function takes to true propositions. If this suggestion were taken up by the nonreductive naturalist, then she would be in a position to maintain that no moral sentence means the same as any non-moral sentence, even though they may have the same content. Furthermore, the aspect of the meanings of moral sentences over and above the non-moral facts that are their truthmakers would be a fact about the sentence and its use rather than about extra-linguistic reality. One advantage of non-reductive naturalism is that it coheres with the usual motivation for realism that I discussed earlier. The intuitive appeal of moral realism, recall, is that we think there are correct answers to moral questions, and that when we engage in an ethical dispute, there is a genuine disagreement at issue, and not a mere difference of opinion. Since non-reductive naturalism endorses both the thesis that moral sentences are truth-assessable, and that some of them are true, it is consistent with the claim that there are correct answers to moral questions. However, it is not committed to the conclusion that there must be peculiarly moral facts. So it retains the advantage of moral realism without the attendant commitment to dubious metaphysical excesses that has plagued many other versions of moral realism. In the following section I will argue that non-reductive naturalism has two particular advantages that render it considerably more attractive than any of its rivals. 14 This idea derives from a suggestion by James Dreier (1999). 7. Two advantages of non-reductive naturalism Non-reductive naturalism provides a solution to both of the problems typically faced by moral realism outlined in section 3. It provides a solution to the apparent clash between the objectivity of morality on the one hand, and the fact that it provides us with motivations to act on the other. It also offers a plausible account of the is/ought distinction. Both of these advantages are suggested by parallel considerations in the new tenseless theory of time. John Perry (1979) first argued that, while tensed beliefs are irreducible to tenseless beliefs, they are nevertheless indispensable for timely, successful action. In other words, tensed beliefs are indispensable to us even though there are no tensed facts to which they correspond. This has been taken up by proponents of the new tenseless theory, for example, Mellor (1998). According to Mellor, we need true tensed beliefs in order to act successfully because we need beliefs that are not true at all times. For example, if I want to be on time for my dentist appointment, and I believe both that my appointment is at 12.30pm and that it takes me half an hour to get to the dentist’s, why do I leave my office at noon? My action cannot be explained in terms of my holding the tenseless beliefs that my appointment is at 12.30pm, and that it takes me half an hour to get there, as I have had both of these beliefs all morning, yet I only leave my office at noon. No, the belief that motivates me to leave my office is the belief that it is now noon. This is an irreducibly tensed belief, one that is only true at noon. Mellor argues that what causes me to act is not the fact that it is now noon, (there are no such facts) but the fact that I come to believe that it is now noon. This can be borne out by the observation that if my watch had stopped and it was actually 1pm, but I believed it to be noon, I would still leave my office. However in this scenario my action would not be successful because the relevant tensed belief is false. So, true tensed beliefs are not reducible to tenseless beliefs, but they are indispensable for successful timely action, even though they are not made true by tensed facts. What makes my belief that it is now noon true (if it is true) is the fact that I hold the belief at noon, which is a tenseless fact. The salient point here is that discovering tenseless facts is not sufficient to provide me with a motivation to act. It is only when I acquire an appropriate tensed belief, a belief that is nevertheless made true by a tenseless fact, that I am motivated to act. The motivational force is supplied by the fact that the belief is a tensed one, since any tenseless belief lacks the appropriate motivational force. But the fact that the belief is tensed is a fact about the belief, and not a fact about its truthmaker. The first objection to moral realism that I outlined in section 3 was that, while it safeguards the objectivity of morality, it severs the link between moral judgements and our motivations to act in accordance with them. The reason for this is that discovering facts does not normally provide us with any motivation to act, but discovering what appear to be moral facts does provide us with a motivation to act. But non-reductive naturalism has a response to this problem. In the case of tensed beliefs, we saw that their motivational force is supplied by the fact that they are tensed, which is a fact about the beliefs, and not about the world. In the case of moral beliefs, non-reductive naturalism could make a parallel case that the motivational force of a moral belief is a feature of the belief, rather than of the non-moral fact which makes the belief true. In this way, acquiring a moral belief could be sufficient to provide me with a motivation to act even though coming to believe the non-moral fact that is its truthmaker would not be. The motivational force of a moral belief would be supplied by the fact that it is irreducibly moral, which is a fact about the belief, and not about the world. The second problem for moral realism was that it must provide an account of the relation between fact and value. Genuine moral realism regards the two domains as irreducible to one another, and is thus committed to a realm of non-natural moral facts. Naturalism regards the moral as reducible to the non-moral. Non-reductive naturalism, as I have argued, makes no claims that the moral is reducible to the non-moral, but neither does it countenance an ontology of irreducibly moral facts. So it must provide an account of the relations between moral and non-moral expressions, and between moral expressions and nonmoral reality. Before developing such an account, I will briefly present the new tenseless theory’s parallel account of the relations between tensed and tenseless expressions, and between tensed expressions and tenseless reality. According to the new tenseless theory, the truth conditions for tensed sentences can be stated by tenseless sentences, but the former are not reducible to the latter. So, for example, a token, f, of the tensed sentence “It rained yesterday” has truth conditions that can be stated by the tenseless sentence, g, “Rain occurs one day earlier than f.” f does not have the same meaning as g, but it does have the same truth conditions. Both f and g are true if and only if rain occurs one day earlier than f. So the existence of the same tenseless fact is necessary and sufficient for the truth of both the tensed and the tenseless sentence. In spite of this, f and g do not have the same meaning, and cannot translate one another. So, tenseless facts are all the facts that are needed to account for the truth of both tensed and tenseless sentences, but there is an aspect of the meanings of tensed sentences that is left unexplained by this account of their truth conditions. “Yesterday” does not mean the same as “one day earlier than,” nor, if uttered on Tuesday, does it mean Monday. However, it has a function such that it picks out the day preceding the day of its utterance, so if uttered on Tuesday it refers to Monday. But all this describes the semantics of the term, and it remains semantically irreducible to anything tenseless.15 So, the relation between tensed and tenseless expressions is that they are wholly irreducible to each other. The relation between tensed language and tenseless reality is that the latter serves as truthmaker for any true sentence of the former. Non-reductive naturalism provides non-moral truth conditions for moral sentences, while denying that moral sentences are reducible to non-moral sentences. So, for example, the sentence “Rescuing that drowning child is the right thing to do,” will have truth conditions statable in non-moral terms, so its truthmaker will be some non-moral fact, such as, for example, “Rescuing that drowning child will maximise happiness.” But non-reductive naturalism denies that the expression “the right thing to do” is synonymous with “the thing that maximises happiness” (or whatever). The moral aspect of a moral sentence is a feature only of the sentence, and not of the fact that makes the sentence true, which is entirely nonmoral. So, the relation between moral and non-moral expressions is that they are wholly irreducible to each other. The relation between moral language and non-moral reality is that the latter serves as truthmaker for any true sentence of the former. 8. Conclusion What moral realism can learn from the philosophy of time, then, is this. Moral realism has, up to now, faced the following dilemma. It must either concede that the existence of moral truths entails the existence of moral facts, or attempt a reduction of moral discourse to non-moral facts. This is exactly analogous to the state of play in the philosophy of time before the emergence of the new tenseless theory. The parallel dilemma was to think that either the existence of tensed truths entails the existence of tensed facts, or that some kind of reduction of tensed discourse is required. But the new tenseless theory showed that the 15 A fuller account of tensed meaning is developed and argued for in Dyke (forthcoming, 2002). entailment doesn’t hold, so the reduction is not necessary. Irreducible tensed truths are made true by tenseless facts. Tense is a feature of our beliefs about tenseless reality. Non-reductive naturalism takes up the moral counterpart to this position, and holds that moral beliefs are irreducible, but are made true by non-moral facts. Morality is a feature of our beliefs about non-moral reality. The new tenseless theory of time has taught us a valuable lesson on the dangers of being too quick to draw conclusions about the nature of reality from the nature of our representations of it. It may be that the only way in which we can represent temporal reality to ourselves is in an irreducibly tensed way, but it is invalid to infer from this that reality itself is irreducibly tensed. So perhaps the lesson for moral realism is that morality is an irreducible feature of our language and thought about non-moral reality, and not a feature of reality at all. And I think this is how it should be. Moral discourse says more about us than about the reality it seems to describe. References Ayer, A. J. (1936), Language, Truth and Logic (London: Gollancz). Brink, D. O. (1995), ‘Moral Realism’ in R. Audi (ed.), The Cambridge Dictionary of Philosophy (Cambridge: Cambridge University Press): 511-12. Dreier, J. (1999), ‘Transforming Expressivism’, Noûs, 33: 558-72. Dyke, H. (forthcoming, 2002), ‘Tensed Meaning: A Tenseless Account’, Journal of Philosophical Research. Gale, R. M. (1968), The Language of Time (London: Routledge and Kegan Paul). Goodman, N. (1951), The Structure of Appearance (Cambridge: Harvard University Press). Hare, R. M. (1952), The Language of Morals (Oxford: Clarendon Press). Hume, D. (1738), Treatise on Human Nature, L. A. Selby-Bigge (ed.) (Oxford: Clarendon Press, 1978). Kaplan, D. (1989), ‘Demonstratives’, in J. Almog, J. Perry, and H. Wettstein (eds.) Themes from Kaplan (New York: Oxford University Press): 481-614. Mackie, J. L. (1977), Ethics: Inventing Right and Wrong (Harmondsworth: Penguin). Mellor, D. H. (1981), Real Time (Cambridge: Cambridge University Press). Mellor, D. H. (1998), Real Time II (London: Routledge). Moore, G. E. (1903), Principia Ethica (Cambridge: Cambridge University Press). Oaklander, L. N. (forthcoming, 2002) ‘Two Versions of the New B-theory of Language’, in A. Jokic and Q. Smith (eds.) Time, Tense and Reference (MIT Press). Oaklander, L. N. and Smith Q. (1994), The New Theory of Time (New Haven: Yale University Press). Perry, J. (1979), ‘The Problem of the Essential Indexical’, Noûs, 13: 3-21. Pigden, C. R. (1991), ‘Naturalism’, in Peter Singer (ed.) A Companion to Ethics (Oxford: Blackwell Publishers Ltd.): 421-31. Quine, W. V. O. (1960), Word and Object (Cambridge: MIT Press). Sayre-McCord, G. (1988), ‘The Many Moral Realisms’, in G. Sayre-McCord (ed.) Essays on Moral Realism (Ithaca: Cornell University Press): 1-23. Schlesinger, G. N. (1980), Aspects of Time (Indianapolis: Hackett). Smart, J. J. C. (1963), Philosophy and Scientific Realism (London: Routledge and Kegan Paul). Smart, J. J. C. (1980), ‘Time and Becoming’, in Peter van Inwagen (ed.), Time and Cause (Boston: Dordrecht): 3-15. Smith, M. (1991), ‘Realism’, in Peter Singer (ed.) A Companion to Ethics (Oxford: Blackwell Publishers Ltd.): 399-410. Smith, M. (1994), The Moral Problem (Oxford: Blackwell Publishers Ltd.). Stevenson, C. L. (1944), Ethics and Language (New Haven, Conn.: Yale University Press).