The tenseless theory of time: insights and limitations.

Dolev, Yuval

THE TENSELESS THEORY OF TIME has enjoyed a great revival in the twentieth century. Prominent philosophers such as Russell, Ayer, Goodman, Quine, and Smart and, more recently, Mellor and Parfit, have turned their philosophical efforts and talents to its defense. It is proper to refer to their work as a "revival," for under different names (and, of course, with terms typical of other eras in the history of philosophy), the view has been at the center of the philosophical preoccupation with time for centuries. A version of the view can be found in Augustine's Confessions, which presents the essentials of the view in terms that are surprisingly familiar to a student educated in contemporary philosophical thought and jargon.

In this century, supporters of the tenseless view have been keen on quoting Einstein's famous assertion that "the distinction between past, present, and future is only an illusion." (1) Of course, each thinker has his own way of unpacking this statement, and the differences can be considerable. Nevertheless, the term "illusion" is invoked in connection with tense by practically all the view's sympathizers. This assertion will be one of my principle targets. Another closely related target will be the notion of "tenseless relations." It is claimed by tenseless theorists that, for example, the assertion that the American Revolution precedes (or preceded) the French Revolution describes a tenseless relation, that is, a temporal fact in which the notions "past," "present," and "future" do not figure in any way, explicitly or implicitly. I shall argue that, like the idea that tense is an illusion, the notion of a tenseless relation is an incoherent one.

A major challenge tenseless theorists have confronted was undermining the massive evidence against their view constituted by everyday language and thought. They have had to show that the fact that we speak and think in tensed terms can be reconciled with the tenseless view. For this purpose, many of them have resorted to the so-called token-reflexive (2) account of tense. In this paper I shall discuss this device as it is developed and presented by Mellor in his 1981 book Real Time. I choose this book because of its thoroughness and clarity. In his recent book, Real Time II (published in 1999), Mellor offers a modified version of the tenseless view, an "indexical theory," that does not rely on token-reflexive sentences. I shall explain in the course of what follows how the objections I develop against the token-reflexive account carry over to Mellor's current view.

To anticipate a likely misunderstanding, I wish to emphasize from the outset that my objections to the tenseless view must not be taken as implying support for the opposing tensed theory of time. The arguments developed in this paper do not assume and do not entail any of the tensed view's tenets. Since this remark is important for the understanding of what follows, I would like briefly to explain it.

The tenseless theory of time is held by many of its adherents to consists of two separate theses: a negative thesis, that there are no deep ontological differences between the past, present, and future; and a positive thesis that, as Parfit puts it, "tense is merely part of mental reality," (3) or that tense is an illusion, or as Mellor puts it, a myth. The first thesis is negative because it consists simply of the negation of the rival tensed theory of time--the view that the present is more real than, or ontologically privileged with respect to, the past and the future (4)--and is given in terms taken from the rival theory. The positive thesis consists of a picture of time that stands independently of any other proposal. Tenseless theorists take the two theses to be mutually implicated. For many of them, the only way to establish the correctness of the tenseless view is to demonstrate the correctness of the negative thesis, that is, the untenability of the tensed view. Thus Mellor writes: "Tense is so striking an aspect of reality that only the most compelling argument justifies denying it: namely, that the tensed view of time is self-contradictory and so cannot be true." (5)

For reasons that I can only hint at in this paper, I believe that this supposed equivalence between the theses does not in fact obtain, and that in particular, the negative thesis does not entail the positive one. The assertion that there are no ontological differences between the past, present, and future implies that the distinction between the past, present, and future is merely an illusion only if we are forced into choosing between the thesis that the present is ontologically privileged and the thesis that "tense is merely part of mental reality." (6) I think we should reject both positions as incoherent. Indeed, I agree with tenseless theorists that the attempts to articulate a metaphysical theory that ontologically or metaphysically privileges the present are doomed to fail. We cannot make sense of the claim that the present is metaphysically more real and, thus, we cannot accept the tensed view, or so I claim. But neither can we make sense of the tenseless positive assertion that the present is equally real as the past and the future, when it is taken to mean that the distinction between the past, present, and future is merely an illusion. Thus, I think we ought to reject the imposition, accepted by both the tensed and the tenseless camps, of the need to choose between them, an imposition grounded in the shared supposition that philosophical clarity concerning time must take the form of a reality claim. I therefore hold that it is possible to accept the negative thesis of the tenseless theory without thereby becoming committed to its positive thesis: we should reject the idea that the present is ontologically privileged, but we cannot endorse the thesis that tense belongs to mental reality.

This brief and general survey of the discussion in the metaphysics of time touches upon many claims, the argumentation for which would exceed the scope of this paper. I shall limit myself in it to an examination of the main points in Mellor's rendition of the tenseless view. The next section concerns the role of tense in language and discusses Mellor's reasons for holding that "tense is an inescapable mode of perceiving, thinking and speaking about reality." (7) In the third section Mellor's token-reflexive account of tensed truths is presented, as well as his current, modified view. These sections prepare the ground for the critical evaluation, offered in the fourth section, of the notion of "tenseless relations" on which Mellor's picture of time is founded. Finally, in the fifth section, I outline my reasons for questioning the intelligibility of the claim that tense is an illusion.

II

The Semantically Reductive Analysis of Tense. According to Dummett, "McTaggart took it for granted that reality must be something of which there exists in principle a complete description," (8) by which he meant a description that is independent of the perspective of the describer (he called it "observer independent"). Tensed properties, which seem to be perspectival--the 1996 presidential elections are past for us, but were future for the voters in 1995--cannot, therefore, be part of a complete, objective description of events and things. In the same vein Parfit writes: "In a complete description of reality, no [tensed statements] would be needed." (9) For tenseless theorists, saying of tensed sentences that they are perspectival is tantamount to asserting that, as Mellor puts it, "there is in reality no difference between the past, present, and future." (10) Very similarly, Parfit says that there are no deep ontological differences between that past, present, and future. Mellor seems to be speaking for both, or indeed, for all tenseless theorists, when he asserts that "tense is not the way that in reality [things] are," (11) and, what amounts to the same, that "tense is really just a mode of thought." (12)

Driven by this notion of "a complete description," early versions of the tenseless view were reductive, that is, they consisted of attempts to eliminate tense, in principle at least, from language and from thought, so as to furnish, again, in principle at least, such a description. Various translation manuals of tensed sentences into tenseless ones that were taken to mean the same were suggested for that purpose. For example, in The Structure of Appearance, Goodman writes: "The `past', `present', and `future' name no times. Rather the `is past at', the `is present at' and the `is future at' are tenseless two-place predicates that may respectively be translated by the tenseless predicates `is earlier than', `is at', and `is later than'." (13) On this suggestion, "Kennedy's assassination is now past," means the same as "Kennedy's assassination is earlier than the utterance of this sentence (or the tokening of this thought)."

The term "translation manuals" is somewhat misleading, for a translation usually imports a thought into a language which we command. The above translation manuals, on the other hand, were conjured up in order to translate something, namely tense, away--to export it out of any language we might possess. Their function was not to help us understand something we do not, but to make a metaphysical picture palatable despite its apparent conflict with the language we already understand and depend upon. Thus, the sophisticated intervention with language that such manuals constitute was also somewhat violent, but it was deemed necessary by early defenders of the tenseless view because of the remarkable and pervasive presence of tense in our experiences, language, and thought.

Indeed, the manifestations of tense with which such a theory must deal are ubiquitous. All of our experiences are thought and spoken of by us not as simply being in time, that is, as occurring on a certain date, but as being past, present, or future. Moreover, not only are there grammatical differences in the way we talk about past, present, or future events and experiences, there are also phenomenological disparities that correspond to the tensed locations of events and experiences. Anticipating what it will be like to see Ingres's "La Grande Odalisque" for the first time was an experience qualitatively different from actually seeing it, and these two experiences are phenomenologically different from remembering that first time. Internal sensations, as Prior showed, make the case even more impressively: "If I am glad that the pain will be over in five minutes, this is not the same thing I shall be glad about in five minutes' time when I say `Thank God it's over now!'" (14)

To be sure, initially at least, experiential manifestations of tense do not establish anything about its nature, for example, they do not by themselves constitute proof of the reality or objectivity of tense. But given their predominance, they do seem to place the burden of proof on those who are after a tenseless description of the world. The translation manuals mentioned above were supposed to meet this challenge by dispensing with tense, at least insofar as language and thought are concerned. However, despite the stock and effort invested in them, further work soon proved their inadequacy as means for tenselessly handling tense, thus confirming the worries of those skeptical of the tenseless enterprise, that tense is too deeply and inextricably rooted in language and thought for the idea of removing it from them to so much as make sense. I shall now review Mellor's arguments which are aimed at establishing decisively that translation manuals of tensed sentences into tenseless ones that mean the same cannot be formulated. Tensed sentences, in most cases, do not have, and cannot have, tenseless equivalencies.

The first step toward realizing the impossibility of translating tensed sentences into tenseless ones is noting that tokens of tensed sentences do not have the same truth conditions as any tenseless sentence that is a candidate to be their translation. Consider a token of the tensed sentence "The lecture is beginning now." Let us call this token T. It is true if and only if it is tokened--uttered, read, thought--at the same time that the lecture is beginning. This tenseless fact, the temporal relation between the two events, is (at least according to tenseless theorists) the fact that gives T its truth-value. If any tenseless sentence can serve as a translation of it, that is, can have the same meaning it has, it is the sentence that describes this tenseless truth condition, namely, the sentence "The tokening of T, this very tokening of it, and the beginning of the lecture occur at the same time." Let us call a token of this sentence S (I shall contend shortly that this sentence is not really tenseless, but for the moment let us assume that it is). Now, whereas S is true (or false) regardless of when it is tokened, the truth of T depends on when it is tokened. T's date, its location in time, is part of the conditions in virtue of which it is true. But it is not part of the conditions in virtue of which S is true. So the two sentences do not have the same truth-conditions, and as Mellor says, "two sentences can hardly mean the same if, as here, they are true in quite different circumstances." (15)

It might be instructive to repeat this argument without appealing to the debatable association of meanings with truth-conditions. If you wonder when the lecture is about to begin, then T, but not S, can provide you with the information you need. If you ask someone when the lecture is beginning and the response is a token of S, you will still not know when the lecture is beginning. Say the response is "The lecture is beginning at the same time in which you are hearing this sentence." You might still be unsure, for it might not be clear which sentence the word "this" in the response refers to (imagine the response is given to you by a foreigner who you suspect has not yet mastered verb conjugation and might be using the present tense but means a sentence you heard in the past). (16) You will know when the lecture begins if the response is "The lecture is beginning at the same time in which I utter the sentence you are hearing now." But this response, employing the word "now," is of course tensed. Perhaps the tenseless response is simply: "The lecture is beginning at 6:00 P.M." Again, this cannot serve as a substitute for T, for it is useless unless you know that it is now 6:00 P.M., that is, unless you employ a tensed sentence in conjunction with it.

The point is that "no sentence could possibly mean the same as another if, as here, it cannot be used at all as the other one standardly is." (17) John needs to take a medication at one o'clock in the afternoon. If a reliable source--his wrist watch, the radio, and so forth--provides him with a token of the tensed sentence "It is now one o'clock in the afternoon," he will be able to take his medication on time. Otherwise, he can only try to guess the right moment. Obviously, if all he is provided with is a token of a tenseless sentence of the kind we have considered as a possible translation, he will not know when to take his medication--tenseless sentences say that at one o'clock in the afternoon it is one o'clock, but not whether it is one o'clock now, which is what John needs to know. Since no tenseless sentence can be used for the same end as "It is now one o'clock," no tenseless sentence can translate it. In general tensed sentences have daily uses that cannot be fulfilled by any tenseless sentences. Thus, whichever way the relation between use and meaning is construed, in light of such examples, we must conclude with Mellor that there cannot be tenseless sentences that have the same meaning as certain tensed sentences.

This also shows that tensed sentences are indispensable--to act successfully we must have tensed beliefs. To take his medication on time, John needs to know what time it is now, that is, have a tensed belief of the form "It is now ..." Insofar as we act in the world, the success of our actions depends in part on our possession of tensed beliefs. I think these claims are pretty much straightforward and indisputable. That is why attempts to formulate and defend the tenseless view by demonstrating the dispensability, in principle at least, of tensed sentences were doomed from the outset.

III

The Semantically Nonreductive Token-reflexive Account. The failure of the reductionist program did not by any means signify, however, the end of the tenseless endeavor. Having abandoned the project of eliminating tense from language, a new generation of theorists found promise in a version of the view which, rather than resisting it, allows as part of its tenseless picture of time the fact that "tense is an inescapable mode of perceiving, thinking and speaking about reality." (18) Once this fact was recognized, the task for the tenseless theorist was reconceived to consist, not of explaining tense away, but of accounting for it in a manner that, while registering its inescapability and indispensability, is congenial to the tenseless picture. What needed to be abandoned, according to the new plan, was the theory of meaning which claims that, and presumes to specify how, the meaning of tensed sentences can be captured by tenseless statements. In place of it, defenders of the tenseless view had to offer accounts of the manner in which tensed statements, whose meanings cannot be given in tenseless terms, receive their truth-values from tenseless conditions, or explications, in tenseless terms, of how an understanding of tensed terms comes about and what it consists of.

The great advantage of this approach is that, while it removes the distinction between past, present, and future from objective reality, it leaves tenses untouched precisely in the domains in which we are accustomed to encounter them and in which we need them most--language and thought. As Mellor puts it, "tense is not being banished altogether, merely replaced where it belongs--in our heads." (19) I shall now present briefly Mellor's version of the tenseless token-reflexive account of truth, by means of which this taming of tense is executed.

Two preliminary remarks need to be made. First, it is necessary to adopt a terminological convention before we start. When one says that the next presidential elections will take place just over three years from now, one is locating these elections in time by means of a tensed description. Tensed sentences come in many forms. In addition to the notions "past," "present," and "future," adjectival words and phrases such as "now," "yesterday," "one week ago," "one year hence," can be used to construct tensed sentences. When the temporal locations of events and objects are given by means of such words and terms, we shall refer to them as their tensed location. The other manner of locating events in time is by means of dates. If I say that the date of the presidential elections is November 2004, I am locating them by their date. Following Mellor, I shall assume this ordinary distinction between tenses and dates. (20)

Second, it needs to be emphasized that the assertion that tense is an inescapable mode of thinking and speaking about reality, is about language in relation to extralinguistic events and objects. It states that when we think or speak about an event or an object, we necessarily think or speak about it as tensedly located in time, that is, as past, present, or future. When perceiving the setting of the sun, it is not my perception that I judge to be present, rather, I perceive the sun setting now. We may not pay attention to this fact about the event or the object, but if we do, it seems to be there to be apprehended by us. It is due to considerations such as these that Parfit remarks that when there is a dispute about the role of tense in language, "the disagreements [are] not about our language, but about the nature of time." (21)

Now, "the first step in explaining away tensed facts is to trade them in for tensed truths." (22) To facilitate the trade-in, Mellor marks several features of truth. First, he asserts that truth is a property of sentences. In the context of disputes concerning the nature, not of time but of truth, this claim can open up a Pandora's box of disagreements. Mellor, however, does not intend to make with it a metaphysical claim about the nature of truth, and so we shall grant him this harmless reference to truth as a property. Next, Mellor points out that truth is a property of sentence-tokens and not of sentence-types. Since the discussion that follows trades heavily on this claim, let us clarify it. The claim is that in thinking and speaking, we always make use of tokens of sentences, rather than of types. Again, in using the notions "type" and "token" Mellor does not have in mind any metaphysically charged distinctions but simply a very ordinary distinction between, for example, your particular copy of the bible, and what is sometimes referred to as the best-seller of all times; or between the sentence "Socrates is mortal" and all the particular events that consisted of someone's thinking, reading, or uttering this sentence. In the present context, the claim that truth is a property of sentence-tokens rather than of types can be easily demonstrated with tensed sentences such as "Today is Monday."

The useful observation Mellor makes about tokens--uttered tokens, read tokens, thought tokens, and so forth--is that they all have determinate dates: "Particular thoughts strike particular people at particular times: the token judgments they make, therefore, like the token sentences they speak and write, all have dates of greater or lesser duration." (23) There is a slight complication in the use of the terms "token" and "type" which should be noted. A particular copy of a book may be counted as a token, in which case the sentences printed in it should be counted as tokens as well. In another context, however, the sentences printed in a book are not tokens; rather, the event of someone reading them is. Thus, whether something counts as a token or as a type may depend on the context. But the cases that will concern us will not involve such complications. For our purposes, we can treat having a determinate date as a defining feature of tokens. The hour your watch is showing now (including the day of the month, the month, the year) is the date of the thought you are having now, and that suffices to distinguish it from the type we may associate with that thought. All we need to remember is that particular thoughts and utterances have a determinate date.

Finally, Mellor states that descriptive sentences that are true or false have "facts that give them the objective truth or falsity we know they have." (24) Mellor does not specify in any detail the nature of the relationship between facts on the one hand and meaning and truth on the other, nor does he assume any theory that explains it, for example, some correspondence theory of truth or some form of verificationism. Again, being economic with explanations on this matter is part of Mellor's policy of avoiding, when possible, philosophical muddles that are not related directly to the claims concerning the nature of time that he is arguing for. We will, nevertheless, have to examine somewhat closely the details of the association Mellor makes between truths and facts, but at this level of generality no particularly objectionable claim has yet been put forth.

Now, tenselessly accounting for the truth of tensed sentences consists in finding "tenseless facts that give [tensed sentences] the objective truth or falsity we know they have.... The facts I am looking for must be undeniable facts, objective and tenseless, that will make tensed judgments true when they are true, and false when they are false." (25) Examples such as the following indicate exactly where the sought after facts are to be found:

 Suppose someone says or thinks sometime in May 1984 that the Queen is
 fifty-eight, and suppose he absent-mindedly makes the same judgment again a
 year later. That is, he produces two tokens of the sentence type `The Queen
 is fifty-eight', each with an objective truth value. But not the same truth
 value. The first token is true, because ... in May 1984 the Queen is
 fifty-eight. The second token is false, because the inexorable change in
 tense of the Queen's birth during the intervening year has by then made her
 fifty-nine.... Token sentences, we have remarked, are things and events
 with more or less definite dates. These dates (May `84 for the first token,
 May `85 for the second), together with the date in April 1926 when the
 Queen was born, are enough to make the first of these tokens true and the
 other false. (26)

From examining such cases Mellor concludes that

 The truth-value of tokens of any particular tensed type are--so far as time
 is concerned--a definite function of how much later or earlier the tokens
 are than the events they are about. (27)

If among the facts that give a token of a sentence its truth-value there are facts about the token itself, for example, its date, the account of what makes that token true is called a token-reflexive account. Mellor's is such an account: rather than a nonrelational property of being past or present or future, a tenseless relation of precedence, in which the date of the token figures, is the objective fact that gives tokens of tensed sentences their truth-values. Tokens of future tense sentences are true if and only if the events they are about have a date which is later than the date of those particular tokens; tokens of present tense sentences are true if and only if they occur at the same date as the event they are about; and tokens of past tense sentences are true if and only if they occur after the event they are about.

Since according to this account only permanent, never changing tenseless facts about dates figure in giving tensed sentences their truth-value, it clearly paves the way for the elimination of tensed facts: if tensed facts need to be invoked at all, it is only in order to account for the truth of tensed sentences, but if other facts suffice for this purpose, which is the case if the token-reflexive account is correct, then no use can be found for tensed facts and there is no reason for insisting on their reality.

As mentioned, Mellor has modified his position, and in his new book, Real Time II, replaces the token-reflexive account with what he calls the "indexical theory." The reason for the modification is that, under certain circumstances, the token-reflexive account may give wrong truth values:

 Imagine a time t when no one has written or is thinking or saying anything,
 i.e. when there are no tokens of any proposition; and consider the
 now-propositions `There are tokens now' `There are no tokens now', which I
 shall call `u' and `-u' respectively. At t, `u' is false and `-u' is true.
 Yet, obviously, no token of `u' can be false at any time, and no token of
 `-u' can be true. For whenever there are any tokens of any proposition, `u'
 and its tokens are true and `-u' and its tokens are false. So when `u' is
 false and `-u' is true, the token-reflexive tale of what makes these
 now-propositions true or false will give the wrong answer. (28)

In order to circumvent this difficulty, the token-reflexive account, according to which

 any token of `P' is true if and only is it is as much earlier or later than
 e as `P' says the present is than e (29)

is replaced by the following "indexical theory" (in Real Time II Mellor adopts McTaggart's original terminology, and calls tensed sentences "A-propositions"):

 any A-proposition `P' about any event e is made true at any t by t's being
 as much earlier or later than e as `P' says the present is than e. (30)

Evidently, the function fulfilled by the token-reflexive account, namely, providing an analysis, in terms of tenseless relations, of what makes tensed propositions true, is just as successfully fulfilled by the indexical theory. The only difference is that tokens are replaced by times: a tenseless relation holds between an event e and a time t, rather than between an event e and a token of "P," as is the case on the token-reflexive account.

To sum up, the nonreductive tenseless view consists of two claims that, on the face of it, are at odds with each other, but that are reconciled by means of the token-reflexive or indexical theories. The first is that tense is inescapable and ineliminable from language, thought, and experience--that is, that tense necessarily appears as part of the reality which we experience, speak about, and think about. The second is that, as Mellor puts it, "tense is really just a mode of thought," (31) or as he says elsewhere, that "tense is not the way that in reality [things] are," (32) or, to use Parfit's formulation, tense "is merely a part of mental reality." (33) Now, when the way we perceive reality is not the way that reality is, we can say that we are subject to an illusion. When the illusion pervades not only perception but also language and thought, that is, when the way we think and speak of reality is also not the way that reality is, and when this predicament is one which, in principle, cannot be escaped from, then we are dealing with an illusion which is both global--it effects all aspects of our cognition--and perpetual--it is operative in all situations, all places, and at all times. We can bring together what has been said hitherto about the tenseless view by saying that, according to it, tense is such an illusion.

I shall now proceed to the critical part of the paper. First I shall present my objections to the token-reflexive account, with which the tenseless view is still widely associated, despite the existence of alternatives, such as the indexical theory. I shall then explain why this objection also applies to the indexical theory. Finally, I shall discuss the claim, implied by these theories, that tense is a global and perpetual illusion.

IV

The Incoherence of the Notion of "Tenseless Relations." Let me start with a summary of the claims I shall develop in this section. The token-reflexive account of the truth of tensed sentences is given in terms of temporally tenseless relations and tenseless sentences that describe them: what makes the sentence "The Queen is fifty-eight" true (if it is true) is the tenseless fact that fifty-eight years separate the Queen's birth and the date on which it is tokened, a fact which is given by the tenseless sentence "The Queen was born fifty-eight years before the tokening of this sentence," or "Fifty-eight years separate the Queen's birth and the tokening of this sentence." I shall argue that Mellor justifies the distinction between tenseless and tensed relations by an appeal to this very distinction, which renders his argument question begging. More seriously, as will come out in the course of analyzing the structure of Mellor's argument, the idea of "tenseless relations" is unintelligible.

Let us recall that Mellor begins the argument for the elimination of tensed facts by invoking the supposedly innocuous distinction between facts and truths--the function of the token-reflexive account is to facilitate the trade-in through which tensed facts are discarded while tensed truths are retained. Now, there is a perfectly ordinary sense in which events are distinct from the sentences with which we think and speak about them. When the witness in court testifies "I was at home when he was murdered," no difficulty arises in distinguishing this true or false statement from what it is about, which is not part of language at all. However, Mellor, under the guise of a natural distinction between tensed truths and tensed facts, sneaks in the heavily charged distinction between the mind, the dwelling place of the subjective tense, and the world of real time. Let us pause for a moment to consider the ordinary distinction between facts and truths. We can distinguish between chemical facts and the truths of chemistry, between the fact that [H.sub.2]O is water and the true sentence "Water is composed of [H.sub.2]O molecules"; or between the fact that Kennedy was assassinated in 1963 and the sentence "Kennedy was assassinated in 1963." These are instances of the ordinary distinction between facts and truths. One reason we call it ordinary is precisely because the question of a trade-in does not come up in connection with these examples. We cannot, and do not wish to enact a trade-in between these facts or events and the true sentences that state them. If a trade-in is the issue, it is plain that for it to accomplish its purpose, that is, (paraphrasing Mellor) "replacing something where it belongs--in our heads," the items traded must correspond to reality, on the one hand, and to what goes on inside our heads on the other, and these latter terms, used in this way, are no longer ordinary but are the products of a philosophical position which is wedded to weighty metaphysical assumptions. So the dichotomy is not the innocent distinction it appears to be.

Now, Mellor needs to provide a justification for treating tense separately, for metaphysically compartmentalizing tensed truths and facts in a manner that sets tense apart from other ordinary instances of the fact/truth distinction. He has to say why tensed facts can be eliminated by the invocation of this distinction, while, for example, chemical, historical, or tenseless facts are left unharmed. The following passages contain the crucial components of Mellor's reasons, and of his interpretation of the distinction between tenseless and tensed sentences:

 [contingent tensed sentences] may not have the same meanings as the
 tenseless sentences that give their truth conditions, but those truth
 conditions surely give their meaning ... anyone who knows that, for all
 dates T, `It is now T' is true during and only during T knows what `It is
 now ...' means. (34)

The word "give" works in this key passage in both directions: sentences give truth conditions, and truth conditions give the meanings of sentences. There is an asymmetry, however, between tensed sentences and tenseless sentences in this regard. While both are given their meaning by truth conditions, truth conditions are given only by tenseless sentences. This distinction between tenseless sentences that also give truth condition, and tensed sentences that only get them, needs to be examined. So let us try to get clear on what "giving" is all about. In the above quotation, to say of a certain sentence that it gives truth conditions is to say simply that it describes them, that is, that it can inform a listener as to what are the facts, the states of affairs, in virtue of which a sentence is true (or false) on a given occasion. It is a bit trickier to flesh out the idea that truth conditions give meaning to sentences. The cited passage makes it clear, however, that what Mellor has in mind is explicated, in part at least, by the assertion that anyone who understands the sentences that give these truth conditions, that is, that describe them, can understand the sentence the meaning of which is given by them.

Mellor's example consists of the tensed token (S) "It is now 1980" and a token of the tenseless sentence (TL) "S occurs in 1980." TL gives truth conditions, the conditions under which S is true: S is true during, and only during 1980, that is, if and only if S occurs in 1980. What are the truth conditions of a token of TL? It too is true if and only if S occurs in 1980 (TL is true if and only if TL). Thus, TL gives its own truth conditions. Do S and TL have the same truth conditions? Yes and no. Yes, in that TL describes the condition, the state of affairs, the tenseless fact to which the truth value and meaning of both are attached. No, because TL makes reference to the date of S, but not to the date of TL itself. So although TL describes the tenseless state of affairs to which the meanings of both sentences are attached, this state of affairs stands in different relations to each--it includes the date of one but not of the other. Giving the shared truth condition involves saying something about one sentence and nothing about the other. This relation is part of the truth condition of both and since it is different in each case, in this respect S and TL do not have the same truth conditions. This difference manifests itself in that while S is true if and only if it occurs in 1980, a token of TL, if true, is true regardless of when it occurs. That is why S and TL do not have the same meaning.

Meaning is related to understanding, and the difference in meaning induces a difference in what understanding consists of in each case. Understanding consists, in part, in possessing the ability to use a sentence correctly, and in particular, in being able to specify the conditions under which it can be tokened truthfully, in being able to give its truth conditions. These, on Mellor's view, are always tenseless conditions. For tenseless sentences, these conditions just are the conditions that give the sentence its meaning. As we noted, TL, for example, gives its own truth conditions, and to understand it is to know that if true, it is true regardless of when it is uttered. For tensed sentences the situation is different. They do not give truth conditions. Their truth conditions are rather given by token-reflexive tenseless sentences. Since it is in these later sentences that the information concerning when they can be uttered truthfully is contained, understanding a tensed sentence S requires, on this picture, (at least tacit) knowledge of the form "S has the truth condition TL."

Let us, then, imagine a speaker, Helen, who has the ability to make her tacit knowledge explicit, and consider a sentence in Helen's idiolect, say "It is raining now." Her understanding of this sentence consists in part in knowing when it can be tokened truthfully. How can she make this knowledge explicit? She cannot invoke Tarski's "convention T" and give the requisite specification by stating that it can be tokened truthfully if and only if it is raining now, for this does not describe the sentence's tenseless truth conditions. On Mellor's picture, a proper specification of the conditions under which it can be truthfully uttered would be: "`It is raining now' is true if and only if it is tokened at a time in which it is raining." And Helen would have to use this tenseless sentence if she wished to make her understanding of "It is raining now" explicit. In Mellor's jargon, the tenseless sentence gives the truth condition which gives the meaning of "It is raining now." I think that, when fleshed out, this is what the difference between giving truth conditions and having truth conditions give meanings comes to: grasping the meaning of a sentence presupposes the ability to think the tenseless thought (or at least to have the tenseless tacit belief) that describes its truth conditions.

This fits in well with the picture Mellor is working with. For Mellor, directly experiencing tenseless facts, for example, perceiving the thunder succeeding the lightning, is an integral part of the mechanism by which we come to understand tenseless sentences. We directly perceive precedence, that is, tenseless facts, and through these experiences we directly acquire an understanding of the sentences that describe the facts experienced--what is experienced is what is described. When a tenseless sentence says that things are thus and so, the meaning and truth of what it says are fixed by things being thus and so. There is no gap: the conditions that TL describes are the conditions its truth and meaning are attached to. With tenses, however, the situation is different. When a tensed sentence says that things are thus and so, it is not by their being thus and so that its meaning and truth are fixed, for it speaks of tensed facts when there are none. Here there is a gap between what the sentence says, and the conditions that give it meaning. We need to hook up with what gives it meaning before we can understand what it says. Thus, the gap needs to be filled in, mediated, by beliefs about the conditions that endow the tensed sentence with its meaning, that is, by the beliefs which would be expressed, if made explicit, by the tenseless sentences which give these conditions.

One of Wittgenstein's mechanical images can be helpful in illustrating the point. Tenseless and tensed sentences can be likened to gear wheels that share the same axis, rotating together. Only the tenseless wheel, however, meshes with facts, with reality. The teeth of the tensed wheel are not designed to do so--there is nothing in reality they can latch on to. They get their motion (meaning and truth) by being on the same axis with the tenseless wheel which is rotated through the perfect, gapless fit that exists between it and reality. Because the meanings of tenseless--but not of tensed--sentences attach directly to the facts that endow sentences with meaning, tenseless facts cannot be traded in for truths, while tensed facts can and ought to be. The upshot of all this is that, while tensed sentences cannot be reduced, or translated, into tenseless sentences (compare section 2 above), they are nonetheless parasitic on beliefs expressed by tenseless sentences for their meaning and truth, which, as we saw, entails that an understanding of tensed sentences presupposes an ability to have tenseless beliefs--beliefs which, were they made explicit, would be expressed by tenseless sentences. It is this difference between tensed sentences and tenseless ones that enables Mellor to treat the fact/truth distinction as ordinary in the tenseless case, and as metaphysical in the tensed case. Thus, getting rid of tensed facts while retaining tenseless facts is facilitated for Mellor by a theory of understanding that distinguishes in the above manner tenseless sentences from tensed sentences.

Now, the first remark to make is that the trade-in of tensed facts for tensed truths already assumes a distinction between tensed and tenseless relations which discriminates against tensed relations and thus begs the question: tensed facts are assumed to be inferior, not to play any role in our understanding, and on the basis of this assumption are then gotten rid of. The second, more important remark, is that the above analysis of our understanding is, I believe, untenable. Indeed, from within the tenseless picture it is natural enough--something has to carry, as it were, the meaning from actual conditions which are tenseless to a sentence that is tensed, a task that only the tenseless thoughts (or tacit beliefs) can fulfill. However we understand notions like "tacit knowledge" or "tacit belief," this much is clear: if we do have the tacit knowledge that a tenseless state of affairs (the one given by TL) is the truth condition for S, then it must be the case that we could grasp, for example, the tenseless sentence TL, "S is tokened in 1980," prior to and independently of our understanding S, "It is now 1980." But this, I claim, cannot be the case. In a nutshell, the reason is that, as Mellor remarks, "however meaning may relate in general to truth conditions, it is an indisputable canon of modern philosophy to relate it to usage." (35) And if usage of a tenseless sentence requires having a grasp of a tensed sentence, then it cannot be the case that we understand it or the thought it expresses independently of the tensed sentence. Let me elaborate.

Consider the sentence "Kennedy was assassinated in 1963." To understand it or for such a sentence to transmit knowledge, one must know what "1963" refers to. It is not enough to know that it indicates a counting from a certain chosen point in time, that 1963 years separate between that point of origin and the assassination. This tenseless fact is a useless fact to anyone who does not know when the point of origin is with respect to the present. Attributing an understanding of a certain symbol to someone--a word, a number, a sentence--requires, at the very least, that that person be able to use it for various ordinary, modest ends, and most ordinary uses of dates are tied up with knowing the present date. Even if to the question "When were you born?" John correctly answers "1960," he cannot be regarded as someone who understands that he was born in 1960 if he does not know how to answer the question: "How old are you?" or if he thinks that 1960 is in the future. If I do not understand that my appointment with the dentist that is scheduled for March 3 is to take place next week, seven days from today, then I do not really know the use of a calendar.

There are cases in which we happen not to know the present date, or be mistaken about it, and fail to act successfully. Of course, if my calendar indicates that my dentist appointment is scheduled for the third, and I show up on the fourth by mistake, that does not mean that I do not understand the use of a calendar or the sentence "Your appointment with the dentist is on the third." But if I never know the present date, then I can never use information about the dates of events to act successfully, and there is no license to attribute to me an understanding of sentences that fix the dates of events. It is true that, as Mellor says, anyone who knows that "It is now T" is true during, and only during T, knows what "It is now" means. But only someone who already knows what "It is now" means, can know that "It is now T" is true during and only during T. In general, an understanding of a sentence of the form "S occurs in [T.sub.1]" depends on understanding sentences of the form "It is now [T.sub.2]."

Thus, it cannot be that being able to think tenseless thoughts facilitates understanding tensed sentences, for it presupposes it. I think the processes of acquiring tenseless concepts and tensed concepts go along hand in hand, they are inseparable. Likewise, tenseless and tensed concepts are not really separable, but are rather fused with each other, with no sharp line or even a fuzzy border area dividing them. Tensed concepts figure in every occasion that tenseless ones do, and vice versa. Mastery of the uses and senses of so-called tenseless sentences is both a prerequisite and a consequence of mastering the use of tensed sentences, and vice versa. We are taught what "It is now" means, for example, when we are told "The noise we are hearing now is that of thunder." And we learn what "e occurs at t" means when it is explained to us that we have a birthday once a year, or when we are told "Dessert will be served after the salad." But we need to have some grasp of temporal succession, that is, of dates, to understand the first sentence. For example, we need to know that the sentence we hear refers to the thunder that is simultaneous with it, that they occur on the same date; and we need to have some grasp of tense to understand that the sentence "Dessert will be served after the salad" concerns the meal we are attending now.

The problem with the token-reflexive account is that, contrary to the supposition of those who rely on it, the sentences and relations employed in handling tensed facts or relations are not themselves purely tenseless. It is true that "It is now 1997" can be explained by saying that the date of the utterance of this very sentence is 1997. But that is only because the context makes it clear that the phrase "this very sentence" refers to the sentence tokened now. Simply employing auxiliary pointing devices--replacing "this" with "this very," using italics, and so forth--still leaves uncertain which sentence is being referred to by the phrase "this very sentence." It is not inconceivable that to the question "Which sentence exactly do you have in mind?" someone will answer "The one he spoke during last year's meeting." This is not likely, but then nor is it very likely that the phrase "This very sentence," and not the word "today," will be used in fixing today's date. The point is that to the extent that token-reflexive formulations can explain tensed sentences, it is because the context makes them tensed themselves. In the above example, for instance, the "is" in "The date of this very sentence is 1997" points to the present. If the context did not make that clear, it would not be obvious what the phrase "This very sentence" denotes. So, even though the token-reflexive account seems to flesh out in tenseless terms what a tensed sentence says, its ability to do so depends on the context providing the tensed information conveyed by the tensed sentence. Thus, again, understanding a tenseless explanation turns on already possessing tensed language.

In ordinary language this mix manifests itself in the presence of tense in sentences describing tenseless relations: we say that the American Revolution occurred before, or preceded the French Revolution, or that Kennedy was assassinated in 1963, or that it is raining at the time in which this very sentence is being uttered. Tense figures even in seemingly tenseless chronologies of the form:

1712-J.-J. Rousseau is born in Geneva.

1742-Rousseau moves to Paris.

1750-Rousseau's "First Discourse" wins the contest of the Academy of Dijon.

Though it is repressed here, tense does come into play in the use of signs such as "1712."

We can conclude that Mellor's theory of meaning, which portrays the meaning and our understanding of tensed sentences as parasitic on tenseless ones, fails. It fails for reasons similar to those Mellor developed against the reductive theories of meaning: since tensed sentences have uses that tenseless sentences cannot fulfill, tenseless sentences cannot be used to capture either the meanings of or the conditions for understanding tensed sentences. More important, we see that any attempt to isolate purely tenseless relations, described by purely tenseless sentences, without recourse to and independently of any tensed terms or notion, cannot succeed. The meanings of temporal concepts, the process of acquiring temporal notions and of mastering their use, just do not exhibit any discernible elements, any dichotomies, to which the terms "tenseless" and "tensed" can be made to correspond.

For the same reason, the indexical theory cannot constitute a theory of meaning for the tenseless view of time. It too turns on the idea of purely tenseless relations, differing from the token-reflexive account only in its conception of the terms that enter into these relations--events and times rather than events and tokens--but not in its reliance on a distinction we cannot draw, between tenseless and tensed relations.

V

A Fundamental Difficulty with the View that Tense Is a Global Illusion. (36) If my analysis of the token-reflexive account and its failures is correct, then this account cannot be considered as making intelligible the assertion that tense is an illusion. In this section I wish to lay out my reasons for holding that such grounds are necessary. The problem with the idea that tense is an illusion, I shall now argue, is not so much to believe it but to understand it.

According to tenseless theorists the objective world, the world of real time, is tenseless, while the world as it figures in our minds is tensed. Before we become convinced of the validity of their view, we mistakenly think of the tenses as part of the real world, as aspects of real time. What is affected by the study and internalization of the arguments in support of the view is a geographical transfer of tense, a crossing of the border between the two worlds, of the interface that separates them: "Tense is being replaced where it belongs--in our heads." (37) This move, I hold, is serf-defeating: it involves a transgression of an interface, but can be made sense of only on the supposition that the border marked by the interface is not transgressable. Let me explain this.

As we saw, the inescapability of tense as a mode of perception, thought, and so forth, amounts to this: if we think of events in tensed terms, we cannot help recognizing that they have tensed temporal locations, that they are past, present, or future. John is contemplating the setting of the sun. He might not be paying attention when, in tensed terms, it is happening. But if he is, then he cannot help thinking that the sun is setting now, that is, that it is occurring in the present. Now, according to the tenseless view, tenses are not real, and in particular the setting of the sun is not a tensedly located event, it is not past, or present, or future. So John is not, and cannot be, regarded as perceiving a tensedly located event: how can one perceive tensedly located events if there are no such things?

That is not to say that perception is infallible, it is only to insist that one cannot be taken to perceive what is not there to be perceived. I might think I am perceiving a pink elephant, but if there is no pink elephant in my vicinity I am making a mistake, or hallucinating, or something of the like. Looking up to the night sky, John is thrilled by the sight of a shooting star. Unbeknownst to him, however, the object that he sees burning through the atmosphere is not a meteorite but an old Soviet satellite spiraling down to oblivion. Can he still be regarded as someone who has perceived a meteorite? No. He thought he saw one, and perhaps he still does, but he is wrong. In the same manner, according to tenseless theorists, if John thinks of the setting of the sun as a tensed event then he is wrong, for "tense is not the way that in reality [things] are." (38) That is what the geographical transfer of tense amounts to--it is the recognition that events are not tensedly located, that they only appear to us to be so located, that it is only in our heads that they are past, present, and future.

Thus, our position in the world is such that we cannot help perceiving events as tensedly located, though they are not--John thinks that the sunset he is seeing is taking place now, that is, is a present event, though it is not. That is the right description of John's state. Or more precisely, of the prephilosophical John. For if John has studied the tenseless view, and has been persuaded by the arguments, then he surely no longer thinks of the events he perceives as present events, for he now knows that they are not. Previously, he thought of the sunset as occurring now. Now he no longer thinks it does; rather he thinks that that is how matters appear to him.

But now a serious problem arises. Can we still maintain that tense is a mode of John's perception? I think not. Tense was a mode of his perception not in that he perceived tensedly located events; for that to be the case, the events he perceived would have to be tensedly located. Rather, it was conceived as a mode of his perception in that he mistakenly thought of the events he perceived as having a tensed location. If he no longer thinks of them as such, tense is no longer a mode of his thinking in the one and only sense in which it was.

Memory calls for a similar account. Ruth recollects the day Kennedy was assassinated--the horrifying pictures on TV, the shock. That event is now long past, or at least that is how she thinks of it. But she is wrong, she is not remembering a past event, for events are not past or present or future, she just remembers the event as past. However, once she recognizes that there are no tensed facts, that events do not have a tensed temporal location, then she no longer thinks of the event as past--she now knows it is not. "Though I inevitably remember things as past," she thinks to herself, "I know that it is only in my thoughts that that is how they appear to be, while in reality, which my thoughts are about, they are not. If `pastness' belongs to anything, it is to my memories, not to what I remember. So it is not accurate to say that tense is inescapable, for when I remember that horrid day, I no longer think of it as past."

Experience produces tensed beliefs, for example, "The train is leaving in two minutes." These beliefs are indispensable for successful action. Tenseless descriptions, such as "The train leaves at 2:04 P.M.," are insufficient. If I do not know what time it is now, this bit of tenseless information is useless, and I shall miss the train. But if, having been convinced by the tenseless view, I know that events are not tensedly located, I cannot have beliefs about their tensed locations. I might unknowingly have wrong beliefs, but I cannot have beliefs that I know are wrong, for example, I cannot believe that "The train is leaving in two minutes" is about the tensed location of an event. So we no longer have beliefs about the tensed locations of events. That is exactly what studying the tenseless view is designed to bring about. It is supposed to alter our understanding of the nature of time, and the alteration consists in our ceasing to regard events as tensedly located, that is, in the elimination of our beliefs about their tensed location.

This, however, is not a position we can readily accept. A little while back we paused to emphasize that tense is inescapable in the sense that we cannot help conceiving events and objects as tensedly located. Experiences and beliefs that are tensed in this sense are not a skin that philosophical wiggling can help us shed and get out of; it is not possible to peel tense away from language and thought. Quite the contrary, the assertion that tense is an inescapable mode of perceiving and thinking about events and objects describes a feature of our cognition which philosophical inquiry presupposes. To use an analogy reverted to by Mellor, no amount of theorizing can establish that we do not perceive colors, for the fact that we do is part of the language and conceptual stock that makes theorizing about colors possible at all. It is the fact that we see colors that can get going in the first place a discussion concerning the reality of colors. Similarly, we do think of events as tensedly located--we remember events of the past, and perceive those of the present, and anticipate future ones--and we cannot help doing so, any more than we can help seeing the colors of objects.

For the tenseless view, that we inescapably perceive, think, and speak about events as being past, present, and future, is not just a matter of contingent fact; it is a logical condition for its very stateability: it is only against this fact that the assertion that the tenses are not real can make sense. But as we just saw, this assertion implies that we can escape tense (John knows the sunset is not a present event) and is thus self-refuting: it undercuts the condition for its being meaningfully stateable. To put it in other words, the distinction between mental tensed time and actual tenseless time turns out to constitute an instrument by which we escape that which, ex hypothesi, is inescapable; it undercuts the very hypothesis that makes it viable, a consequence that puts into question its coherence.

Now, the tenseless theorist might object that the above argument proceeds too quickly and misses a simple but important point. According to him, before we study the tenseless view we do not distinguish between believing that things actually have a tensed location and thinking that that is only how things appear to us. We regard tensed beliefs as beliefs about the tensed locations of things, and suppose that we experience, think, and speak of events as tensedly located because they really are. In light of the tenseless view we no longer see matters in this way. The result of the analysis is the realization that things only appear to be tensedly located. Having tensed beliefs is reinterpreted to mean that it inescapably appears to us as though events and things have a tensed location although they do not. But then, says the tenseless theorist, the tenseless view, far from being in conflict with the assertion that tense is an inescapable mode of thought and speech, is still committed to it. Moreover, endorsing the view does not require abandoning tensed beliefs, only realizing that these concern the appearance of things, rather than the way things really are. So it might be protested that the criticism made above, with its charge that the tenseless view divests us of our tensed beliefs, is misguided. All that tenseless theorists are claiming is that conceptual analysis teaches us that tenses cannot be real. Admittedly, we perceive events as though they are tensedly located, and "we must think and speak as if there were [tensed facts]." (39) But this is nothing but the way the inescapable global illusion that tense is manifests itself. Realizing this, we see that philosophical analysis is not a means of actually experiencing the tenseless reality that is beyond this illusion, only of inferring that there is such a reality.

Similarly with other illusions, the rejoinder continues. On very hot sunny days roads look wet at a distance. We cannot help seeing them as though they were spotted with puddles. But there is no mystery here. We know the road is dry, and can explain why it appears to be wet, by describing the effect the heat emitted from the road's surface has on the refraction index of the air right above the road, as a result of which light rays coming from the sun curve, and it appears as though the sun is reflected from a puddle. That is what an optical illusion is--things appear differently from the way they really are, and they continue to appear differently even after we learn the truth of the matter. Of two lines of equal length, embedded in a certain geometrical configuration (the Muller-Lyer diagram), one looks longer than the other, and they continue to appear unequal in length even after it is verified that they are not. This fact does not make us suspect that perhaps they are unequal after all, rather, it compels us to explain why they appear to be unequal. Similarly with tense, goes the rejoinder. Events are not tensed, they do not have a tensed location, but they appear to be tensed, and they continue to appear tensed even when we discover, in this case through conceptual analysis, that they are not. This claim, insists the tenseless theorist, is entirely innocuous. Philosophizing does not open impossible channels of cognition, but it can modify and correct our understanding of the mechanism and reality underlying our cognition. That is its purpose.

Well, the reason this attempt to salvage the tenseless view falls has to do with the following crucial difference between tense and optical illusions: in the case of optical illusions we can specify, and in fact attain, conditions under which our perception is veridical. It is only by reference to the way a truly wet road looks under such conditions that talk of a road appearing to be wet makes sense. Similarly, it is only due to the fact that there are circumstances under which we can in fact verify that the two lines in the Muller-Lyer diagram are equal in length that we can so much as describe the diagram as an optical illusion. To say meaningfully that the lines appear unequal in length, we need to have the notion "equal in length." Thus, only if there are circumstances that equip us with this notion, circumstances under which we could correctly say of the lines that they are equal in length, can we speak here of an illusion, of a situation in which they appear unequal though they are not. More generally, the word "illusion" designates a distinction between the way something looks under one set of circumstances, and the way it looks under different conditions, for example, between the way a patch of asphalt appears at a distance on a hot day, and the way it looks when observed from close by; or between the way two lines look when they are embedded in a certain configuration with other lines, and the way they look outside of this configuration. In fact, applications of the notion "illusion" make tacit use of two veridical perceptions. For example, we invoke both (a memory of) a perception of a wet road and a perception of a dry road to describe both how the road looks--it looks wet, and how it really is--it is dry.

The hypothesis that experience, thought, and language may be systematically misleading is not new. However, it was always necessary to invoke veridical cases in order to raise it. Descartes assumed we knew what dreaming is when he articulated the possibility that what we take to be our waking experiences are in fact dreams. Ayer had to presuppose our acquaintance with bent sticks to so much as suggest that a stick in water looks bent, that is, that the perception of a stick in a glass of water is, as he put it, "delusive." (40) To repeat, the word "illusion" describes a distinction between the way things appear under different circumstances, and refers to those situations in which things appear the way they would if different circumstances were to obtain. The upshot of this is that the word "illusion" evokes (tacitly) conditions other than those that actually obtain.

But what could such conditions possibly be in the case of tense? The problem is that with tense we have no clue, cannot have a clue, as to what conditions other than those that obtain could be. In fact, the very purpose of employing the word "illusion" in this case is to assert that such circumstances, circumstances in which the illusion were somehow neutralized, cannot so much as be conceived. Here we can not draw the distinction between the way things appear under certain circumstances and the way they really are, because we cannot describe circumstances under which they would appear differently from the way they do, or from the way they are. Again, it is entailed by the very claim that tense is an illusion that we cannot under any circumstances have access to veridical perceptions. If we could, tense would no longer be an illusion! The same is the case with the way we think and speak about events in the world, not only with the way we perceive them. This, of course, is a serious problem, for it turns out that speaking of tense as an illusion amounts to asserting that there is no nontensed realm of experience, thought, or discourse which would furnish the acquisition of the concepts required for describing tenses as illusions. If tense is an illusion, then we cannot have access to that the contrast with which warrants describing tense as an illusion. Clearly, this is an application of the term "illusion" which we cannot make sense of--it is incoherent. On the one hand, it tacitly suggests that tenseless reality is something that under certain conditions we can gain access to, but on the other, its function is to indicate that this reality is always and forever veiled from us and out of reach. Thus, using the word "illusion" in this context requires that we remove ourselves from our heads and take our inevitably tensed minds for a stroll in the tenseless fields of reality. And that even philosophers cannot do.

To sum up, the use made of the word "illusion" in the statement that tense is an illusion cannot be the ordinary use we make of it in other contexts. One important feature of its ordinary use--the conceivability of veridical perceptions--is absent. This observation does not constitute a blanket argument against all versions of the tenseless view. It merely aims to highlight that, contrary to the way it is often introduced, the thesis that tense is an illusion is not self-explanatory. A theory is necessary in order to make it intelligible. The token-reflexive account of tensed sentences might have functioned as such a theory were it not for its own structural flaws. There are other arguments with which philosophers attempt to expound and defend the tenseless view of time. The scope of this paper does not permit a discussion of them. I can only remark that I have not found them to be more promising than the arguments studied here and that thus, the idea that tense is an illusion remains, in my view, incoherent.

As indicated in the introduction, I believe the tenseless view is driven by an important insight, namely, that the present is not ontologically privileged. More than anything else, what gets tenseless theorists going is the need to react to the attempts of tensed theorists to anchor the distinction between the present and the past and future in some metaphysical theory which explains how the present is more real. I believe that the reaction of the tenseless camp is a valid one, but only so long as it remains a criticism of a doctrine that cannot be made sense of. When this reaction evolves beyond the scope of a mere criticism and turns into the doctrine that tense is an illusion, it begins to suffer from incoherence as well. (41)

Correspondence to: Department of Philosophy, Faculty of Humanities, Bar-Ilan University, 52900 Ramat-Gan, Israel.

(1) I am not sure that Einstein had something like the tenseless view in mind when he said these words. I think that what he intended was that the impression that being past, present, or future are absolute properties, rather than properties that are relative to a frame of reference, is an illusion. This, will become evident, is very different from what tenseless theorists mean by the claim that tense is an illusion.

(2) A term first introduced by Hans Reichenbach in his Elements of Symbolic Logic (New York: Free Press, 1966).

(3) Derek Parfit, "Rationality and the Metaphysics of Time," unpublished paper, 7.

(4) It should be mentioned that some tensed theorists hold that the past, as well as the present, is real. All of them agree, however, that the future is not.

(5) David H. Mellor, Real Time (Cambridge and New York: Cambridge University Press, 1981), 4-5.

(6) Parfit, "Rationality and the Metaphysics of Time," 7.

(7) Mellor, Real Time, 6.

(8) Michael Dummett, "A Defense of McTaggart's Proof of the Unreality of Time," in Truth and Other Enigmas (Cambridge, Mass.: Harvard University Press, 1978), 356.

(9) Parfit, "Rationality and the Metaphysics of Time," 3.

(10) Mellor, Real Time, 58. To this Mellor adds: "i.e., there are no tensed facts." Parfit finds the denial of tensed facts both unnecessary and indefensible. According to him there are such facts, and the sentences that state them "can be straightforwardly true of false." But tensed statements are subjective, in the sense that they are perspectival: "they describe reality from a particular subject's point of view"; Parfit, "Rationality and the Metaphysics of Time," 7.

(11) Mellor, Real Time, 24.

(12) Mellor, Real Time, 140.

(13) Nelson Goodman, The Structure of Appearance (Cambridge, Mass.: Harvard University Press, 1951), 295.

(14) This formulation of the example is from Dummett, The Reality of the Past, Truth and Other Enigmas (Cambridge, Mass.: Harvard University Press, 1978), 373.

(15) Mellor, Real Time, 74.

(16) Using "this very sentence" rather than "this sentence" would not make much of a difference, for there might still be some uncertainty as to what the new phrase, "this very sentence," refers to. The context might be such that it is clear that it refers to the sentence being uttered now, but in that case the response is in effect tensed. I shall return to this point later.

(17) Mellor, Real Time, 75.

(18) Mellor, Real Time, 6.

(19) Mellor, Real Time, 92.

(20) The second chapter of Mellor's book Real Time provides a detailed and helpful elucidation of this distinction.

(21) Parfit, "Rationality and the Metaphysics of Time," 1.

(22) Mellor, Real Time, 34.

(23) Mellor, Real Time, 39.

(24) Mellor, Real Time, 40.

(25) Mellor, Real Time, 40, 34.

(26) Mellor, Real Time, 40.

(27) Mellor, Real Time, 41.

(28) Mellor, Real Time II (London and New York: Routledge, 1998), 32.

(29) Mellor, Real Time II, 31.

(30) Mellor, Real Time II, 34.

(31) Mellor, Real Time, 140.

(32) Mellor, Real Time, 24.

(33) Parfit, "Rationality and the Metaphysics of Time," 7.

(34) Mellor, Real Time, 75 (italics in the original).

(35) Mellor, Real Time, 75.

(36) Some tenseless theorists take tense to be perspectival (like the "hereness" of a place) and claim that we are subject to an illusion if we fail to recognize its perspectival nature. I address this interpretation of the assertion that tense is an illusion in a paper entitled "Space and Time--Some (dis)Analogies," Iyyun 49 (January 2000): 61-72.

(37) Mellor, Real Time, 92.

(38) Mellor, Real Time, 24.

(39) Mellor, Real Time, 78.

(40) This fact is, of course, the basis of Austin's adamant criticism (in his Sense and Sensibilia [Oxford: Oxford University Press, 1962]) of Ayer's attempt to draw from such cases the conclusion that all we ever perceive are sense data.

(41) I am greatly indebted to Hilary Putnam, Derek Parfit, Charles Parsons, Yemima Ben-Menahem, and Ruth Weintraub for helpful comments on earlier versions of this paper. I would also like to thank the Sydney Edelstein Center for the History and Philosophy of Science at the Hebrew University in Jerusalem for a fellowship that made writing this paper possible.