1. In a usage that is now more or less standard in philosophy, van Inwagen defines ‘fatalism’ as ‘the thesis that it is a logical or conceptual truth that no one is able to act other than he in fact does’.Footnote ¹ That thesis (‘freedom fatalism’) should be distinguished from another: the view that nothing that we do will make any difference to how things turn out. The latter understanding of the term ‘fatalism’ (I will call it ‘control’ fatalism) does, I believe, accord more closely with everyday usage. The fatalistic man is not the one who thinks it is not within his power whether to go to the shelter or to stay in bed, but, rather, the one who thinks it makes no difference which he does so far as his prospects of surviving the air raid go; for, as they say, either a bomb has his number on it or it does not. It accords more closely, too, with at least one strand in the treatment from which nearly all subsequent discussions take their lead. Thus, in his discussion of tomorrow's sea battle, Aristotle writes:

The suggestion is not that in these circumstances no one would be able to do other than he does, but that there would be no need to deliberate or to take trouble – for whatever we do will make no difference to the way things turn out.

I suspect that one of the reasons fatalism in the control sense is widely ignored is that it is felt to be too absurd to be worth taking seriously. Referring, I think, to this form of fatalism, Ryle suggests that ‘no philosopher of the first or second rank has defended fatalism’.Footnote ³ But that is not in fact true. The early Wittgenstein pretty clearly defends a form of control fatalism:

In the Notebooks Wittgenstein offers the following consideration in support of this view:

The fatalistic thought that, quite generally, how things – or, at any rate, how the really important things – will work out is in some fundamental way entirely outside our control may pander to the familiar desire to give up: to the familiar hope that nothing of substance turns on our efforts, or the wish that it did not. This may combine with the realisation that much of what is most important to us is largely beyond our control: is determined by forces, whether blind or intentional, quite independently of us. But my concern in this paper is with more strictly philosophical forces at work here. They are forces – philosophical preconceptions – that have an impact well beyond discussions of fatalism and that are, as I will suggest, often as thoroughly embedded in familiar responses to the fatalist reasoning as they are in the reasoning itself.

2. My concern in this paper is with the control form of fatalism (which I will sometimes refer to simply as ‘fatalism’.) I will, however, begin with some brief remarks about the relation between the two forms.

It may seem that any general argument that might establish that nothing I can do will make any difference to whether or not there is a sea battle tomorrow is going to establish that nothing I can do will make any difference to whether or not I raise my hand to command engagement.Footnote ⁶ ‘Nothing I can do’ – such as what? What kind of action is it that, according to this argument, will be ineffective in bringing about the raising of my hand? We have lost the image on which no matter how I twist and turn I will end up in the same, pre-ordained, place. Or rather: we have lost that image unless we are thinking of the twists and turns – what we strictly speaking do – as something distinct from anything in the world in which actions in an everyday sense occur: unless we are thinking of hand risings as much as sea battles as possible ends of our actions rather than our actions themselves. In this way, control fatalism, in its generalized philosophical form, may be a close relative of familiar forms of mind-body dualism that identify the real agent and her actions as something distinct from – ‘located in’ – the bodily human being: more specifically, a close relative of dualism in its epiphenominalist form.

Control fatalism here stands in a curious relation to a questioning of human freedom that stems from concerns about determinism. The latter is a worry about there being causal determination where we don't want it: namely, between my actions and antecedent conditions from which they flow. The former, a worry about there being no causal determination where we do want it: namely, between anything I might do and subsequent events. There may be an interesting dialectical relation between the two forms of fatalism here. In response to concerns about freedom of action we may picture the self that acts as lying outside the everyday world; but when we do this we may struggle to retain a sense of our actions as contributing to anything that happens in that world.

We may, as I just have, think of both forms of fatalism as attempts to identify the place of the agent within a picture of a ‘natural’ world that is subject to a rigid causal determination. Control fatalism may be equally at home in a quite different framework. On the version I have been considering, the self lies outside the everyday ‘material’ world: the latter being the world in which are located all the real forces – blind, mechanical ones – governing what happens. In an alternative picture – perhaps to be seen in the fatalism of Greek tragedy – while the self lies within the world of everyday happenings, the fundamental causal forces – those, at any rate, that govern things that really matter to us (sea battles, perhaps, but not arm risings) – lie outside that world, and are intentional in character. In this respect, modern fatalistic thinking in practice may be more akin to the ancient Greek picture than to the modern, science-inspired, one of where the real forces lie. In very general terms, what is at work in our lives is pictured as a moral or ‘narrative’ necessity, rather than a mechanistic one.Footnote ⁷

3. I mentioned a possible link between control fatalism and an imagery, perhaps Cartesian, of the self as lying outside the everyday world. But the sense in which such dualism ‘disconnects the individual from the world’ can be thought of in a different way: it represents us as related to the world primarily through thought about it, and only secondarily, if at all, through action in it. Descartes has taken the decisive step when he declares himself to be an essentially thinking thing: before, that is, he gets to his further reflections on the non-material nature of that thing. I believe that this face of Descartes's understanding may have a more basic and persistent hold on philosophical thinking than does his ‘mind-body dualism’ as normally construed, and that it may be the strongest narrowly philosophical impetus to fatalism: one that is still alive despite the general abandonment of such dualism.

Consider Aristotle's presentation of the fatalist's reasoning:

Discussion of this reasoning has focused on two kinds of objection to it. First, many (including, it is generally agreed, Aristotle himself) have suggested that we should reject the premise: ‘Every affirmation or negation is true or false’. And second, it is suggested that the inference from this is invalid: from the fact that it is true that there will be a sea battle tomorrow it does not follow, it is argued, that it is necessary that there will be: that there is nothing I can do to prevent it. What follows is simply that I will not do something that prevents it. In this section I will be concerned with the inference. I will return to the premise in section 5.

A certain picture of truth may seem to imply that it is only true that something will happen if it is inevitable that it will. The picture is this: if what someone says at a certain time is true (‘There will be a sea battle tomorrow’), then the state of the world at that time (the time of utterance) must be such as to ‘make’ it true: in other formulations, there must be something that already ‘fixes’ it as true, must be something ‘in virtue of which’ it is true, or must be a ‘truthmaker’ for the statement. Now this idea (in its various articulations) takes two different forms in the fatalist literature:

(i) Byrd, following what he takes to be Aristotle's view, argues that the appropriate truthmaker for the statement ‘There will be a sea battle tomorrow’ would be the sea battle itself.Footnote ⁹ This must, in some sense, be ‘already there’ if our statement about it is to be true now; and if ‘already there’ then, obviously, out of our hands. Arthur Prior appears to be working with such a picture, giving it an epistemological slant, when he writes:

A first response to Prior's reasoning is this. When Prior speaks of ‘foreknowledge’ he is clearly not thinking of the knowledge that people often have of what they are going to do. That the commander knows that there will be a sea battle tomorrow – because he has decided that there will be – does not entail that he could not prevent it. With that, it is not clear that this form of knowledge of the future – the knowledge embodied in my intentions to act – is, as Prior puts it, ‘much more like an effect of what is known than a cause of it’. Quite the reverse in fact if we can follow Anscombe's suggestion (quoting Aquinas) that a person's knowledge of what she will do is ‘the cause of what it understands’.Footnote ¹¹ At any rate, there is an important range of cases in which it is clearly closer to the mark to say ‘He did it because he said he would’ than ‘He said he would do it because he was going to’. It is not as silly as Prior supposes to treat ‘foreknowledge’ as a sort of cause of what is foreknown; and a recognition of this fact is essential to seeing that foreknowledge is compatible with preventability.

Aristotle himself speaks of the direction of fit between the truth of a statement and that of which it speaks:

This image might seem to justify the step from the truth of a statement to the inevitability of what is spoken of. But the opposition suggested by Aristotle is a false one. Thus, it could be the case both that it is today's battle that makes yesterday's announcement by the captain ‘We will engage tomorrow’ true, and that it was this announcement that was the cause of the existence of the battle. (The crucial contrast here will stand out less clearly than it should if one ‘forgets’ that statements are spatio-temporal events. I will return to this.)

(ii) That was one way of linking truth and inevitability. A second way of doing so is also articulated by Prior when he writes:

The ‘truthmaker’ is here pictured as a set of present conditions in virtue of which the future event is causally inevitable. On this view, assuming that the world is not fully deterministic – that there are points at which it is causally open which way things will go – there are some propositions about the future that are now neither true nor false: for there are no present truthmakers for them. To take a popular example: while it is true now that the coin will come down either heads or tails, it is neither true that it will come down heads nor true that it will come down tails.Footnote ¹⁴

Developing this picture, Lucas argues that where the future is already fixed by currently existing determining causes – and so it is possible to make certain predictions about what will happen – we are mistaken in supposing that we can have any influence on it through our actions.Footnote ¹⁵ It is only in so far as there are gaps in the causal determinism that governs most things – only in so far as the joints in nature are not totally rigid – that there is room for us to have some influence on what happens. As Zagzebski expresses the thought: ‘We have no power to prevent what is causally necessary in the future’.Footnote ¹⁶

There is, however, clearly something wrong there: something that closely parallels the confusion in Prior's remarks about ‘foreknowledge’. For while it is true that there is often a relationship between the certainty of predictions and what we take to be the possibilities of action open to us, the above reasoning gets the relationship (something like) the wrong way round. I can predict with a fair degree of certainty that my car will not start tomorrow; but it is not this – the fact that it will not start tomorrow - that makes me feel that it is not within my power to mend it. Rather, it is the fact that I am fairly certain that it is not within my power to mend it – because, perhaps, the gears are tightly jammed - that underpins my conviction that it will not go tomorrow. Conversely, it is, in a particular case, my (assumed) ability to determine the course of events by my actions that underpins my conviction that I cannot know what will happen – that the future is not ‘fixed’ – independently of my deciding what to do. Perhaps better: my conviction that a particular future event is ‘fixed’ just is my conviction that there is nothing I can do to prevent it; and so appeal to that ‘fixity’ cannot provide grounds for the idea that any action I take will be ineffective.

Behind Lucas’ reasoning lies an image of ‘nature’ as enormously powerful. We can be drawn to the image when confronted by overwhelming natural phenomena such as earthquakes or orbiting planets in the face of which we are powerless. Generalising from such cases, we may begin to fear that a remorseless deterministic necessity may govern everything that happens: may do so in a way that rules out we – weak, fragile beings that we areFootnote ¹⁷ – having any impact on the course of events. But this image of nature as a rigidly deterministic mechanism in the face of which we are effectively powerless involves a crossing of wires: one that is concealed by the focus on colliding planets and tsunamis to the exclusion of specks of dust drifting in the breeze. While the latter may be equally causally determined it is obvious in this case that this fact carries no suggestion of our powerlessness. In cases in which we are excluded from any influence on the course of events, it is not the determinism that excludes us, but the scale of the forces pitted against us.

4. The confusions on which I have focused spring, I suspect, from a deeper assumption. Those who resist fatalism by appeal to the notion of an ‘open future’ are, characteristically, seeking a ‘metaphysical underpinning’ for the idea that we may have some influence on the course of events. Here are some recent formulations of this idea:

The force of words such as ‘metaphysical’ and ‘objective’ fluctuate dangerously in philosophy. Their primary force here Footnote ²² is to indicate that what is sought is a condition of the world that renders it a possible habitat for creatures with a capacity for free action and control: something in virtue of which the world is a place fit for agency. Now, it is uncontroversial that there are particular conditions that are necessary for the agency of particular kinds of creature. For us, these will include objects and forces on a human scale. But the idea of a metaphysical condition is the idea of a condition of a deeper kind: one that would need to be satisfied for any kind of creature to make a home. Thus, when Sweeney speaks of an event as an ‘objective possibility’ her point, explicitly, is not simply that we cannot know it will occur, but that, for reasons deeper than any features of our capacities, it is unknowable. Or when it is said that ‘To qualify as contingent the predicted event …. must neither be impossible nor inevitable’Footnote ²³ the idea is not that the event must be producible or preventable by us; but, rather, that it must satisfy some deeper condition – one that has nothing to do with human capacities – that makes it a possible candidate for production or prevention tout court. The condition may be thought of as lying in the absence of fully determining causes for some future events; or, more deeply, in the passage of time itself: in the fact that, in contrast with the past, future events are not yet ‘real’.

So construed, this ‘metaphysical openness of the future’ (if it exists) is something that is reflected in our lives in, on the one hand, the impossibility of our predicting certain events, and, on the other, the possibility of free actions that have some influence on them. In one familiar line of thoughtFootnote ²⁴, the former is a mark of the latter: that we cannot predict what will happen is an indication of a condition of the world – an indeterministic joint – that leaves room for our input. In opposition to such a picture, I argued that it is a mistake to suppose that the possibilities for action are underpinned by ontology in this way. The way to see whether some future event is ‘fixed’ in a sense that puts it beyond my possible influence is to reflect, not on features of the event independently of its relationship to me, but on the possible causal handles that I have on it. A parallel point applies to prediction. While the toss of a coin is sometimes offered as an example of a ‘future contingent event’ I don't suppose anyone has any idea whether how a coin lands is causally inevitable. With that, its being so is not a consideration that has any relevance to the place that the practice has in our lives. If we say that it is ‘chance’ which way it comes down, ‘chance’ must be clearly distinguished from ‘causally undetermined’. And if we say that it is not ‘inevitable’ that it will come down heads the force of this is not ‘metaphysical’: ‘metaphysical’ in the sense of picking out a feature of the world that is independent of and underpins the relations in which we stand to events like this. Its force lies in those relations to us: in the fact that neither party to the toss has any grounds for predicting how it will land nor any power to ensure that it comes down the way that she has called.

To judge that some future event is inevitable just is to judge that we can safely predict that it will happen and that there is nothing we can do to prevent it. It is not to speak of a metaphysical condition that underpins these ways of thinking of it. There is no class of events – ‘future contingent events’ – that are candidates for production or prevention tout court. And so the debate between the fatalist and his opponents over whether this metaphysical condition is ever satisfied rests on a misunderstanding. In the next section I will turn to the correlate at the linguistic level of this metaphysical condition: namely, ‘propositions about future contingent events’.

5. Aristotle's fatalist reasoning, as presented in section 3, starts from the premise: ‘Every affirmation or negation is true or false’. In some form or other, this is a common starting point of fatalist reasoning and a focus of various responses to it. In particular, recent critiques of such fatalist reasoning are dominated by responses that question whether propositions about ‘future contingent events’ – events that are, as I expressed it, candidates for production or prevention tout court – have a truth value now.

We might think of the default position, sometimes labelled ‘Ockhamist’, as being that according to which, even where it is not yet causally determined one way or the other whether there will be a sea battle tomorrow, it is still the case that one or the other of the propositions ‘There will be a sea battle tomorrow’ or ‘There will not be a sea battle tomorrow’ is true and the other false: among the various causally possible developments one is – is now – the course things will actually take. Amongst the views that have been canvassed in opposition to this proposal are these:

1. (i) On what is widely construed as Aristotle's position, while it is true today that either there will or there will not be a sea-battle tomorrow, we should reject the assumption that every proposition about the future is either true, or if not true, false. Łukasiewicz suggests that we need to introduce a third truth-value for contingent propositions about the future.Footnote ²⁵ In the same spirit, others argue that ‘the future’ should be identified with the necessary future: ‘It will happen’ means ‘It is now settled that it cannot fail to happen’.Footnote ²⁶

2. (ii) While a proposition about a future contingent is, now, either true or false, it is not yet ‘settled’ which truth-value it has.Footnote ²⁷

3. (iii) Truth must be relativized to a context of evaluation: the proposition ‘There will be a sea battle tomorrow’ is neither true nor false when evaluated today, but may express a truth when evaluated tomorrow.Footnote ²⁸

4. (iv) While the proposition ‘There will be a sea battle tomorrow’ is (say) true now, it may change its truth value as a result of human interference in the course of events.Footnote ²⁹

5. (v) Propositions about future contingents are all false: it is false that there will be a sea battle tomorrow, and false that there will not.Footnote ³⁰

In listing these views I have spoken of whether certain ‘propositions’ are true or false. In practice, there are variations in terminology at this point. Aristotle speaks of ‘affirmations’Footnote ³¹; others speak of ‘sentences’, ‘sentence tokens’, ‘utterances’, ‘assertions’, ‘thoughts’ or ‘statements’. There are important distinctions to be drawn there: in particular, a distinction between terms, such as ‘utterance’, that denote particular spatio-temporal events and ones, such as ‘proposition’, that denote more abstract entities. (Others, for example ‘statements’ or ‘thoughts’, may be open to readings of both kinds.) We should, one might think, find no plausibility at all in the suggestion that it would follow from the fact, if it were a fact, that every affirmation is true or false that it is necessary for everything either to be the case or not to be the case; and so there would be no need for any of the alternatives canvassed by the logicians. If it were only affirmations about the future that posed a threat to my ability to influence its course, the threat would be a very limited one, and could be avoided altogether by sufficient caution in affirming things.

Those remarks will, of course, be taken to miss the point. (And Aristotle himself has a response to them, to which I will return.) For, it may be said, the threat posed by the fatalistic reasoning comes from the idea, not that something true or false has been said, but that there is something true or false there to be said; or, in Frege's terms, something there to be thought:

Notwithstanding the variations in articulation, the hold of the philosophical argument for fatalism is, on the face of it, dependent on the idea, articulated by Frege, that the real candidates for truth and falsity are something other than particular contingent happenings such as affirmations, utterances, assertions, thinkings and so on.

Setting to one side possible discomforts about such abstract entities, how would Frege's picture provide the fatalist's reasoning with what it needs? Consider someone for whom any serious possibility of truth in fatalism would be a matter of immediate concern: someone – say Aristotle's naval commander - considering the possibility of an action – one that by normal lights has every prospect of success – aimed at securing a crucial result. The initial character of his relation to the proposed course is, we can imagine, reflected in his central question being ‘Should I order engagement with the enemy?’ His question is whether a certain course of action is appropriate or inappropriate, not whether a certain proposition or thought is true or false. Prompted by the fatalist, however, we remind him that, whether or not it is his question, there is something in the vicinity for which, as Frege expresses it, the question of truth arises: namely, a thought that there will be a sea battle tomorrow (or: that he will order engagement). After all, it is clear that others, for example a reporter on shore, can consider a possible engagement tomorrow in a spirit of predictive speculation: ‘Will there or won't there be a battle?’ or ‘Will he or won't he issue the order to engage?’ And further, based on her previous experience of the captain – experience of him, perhaps, as cautious and wise – we can imagine that the reporter has the thought ‘There will be no battle’. The thought is there to be had by her. But now, if ‘there to be had’ by her, then, on Frege's picture, ‘there to be had’ by anyone – including, of course, the captain. If he continues to deliberate this must be at the price of a more or less self-induced blinkering.

But this reasoning gets things the wrong way round in a manner parallel to that encountered in the reasoning of Prior and Lucas. It is (perhaps subject to slight qualification) true that there is a tension between the enquiring and deliberative attitudes towards a possible future occurrence: that I cannot simultaneously treat the occurrence of a battle tomorrow as an object of speculation and as a subject for deliberative reflection. And it may be true that the commander could come into relation to the thought about tomorrow's battle in the Fregean way: could stand back and ask, in the mode of enquiry, ‘Will there be a sea battle tomorrow?’Footnote ³³ But there is no necessity in his doing so: no sense in which this is demanded of him. Indeed, his doing so may involve, not intellectual honesty, but a serious dereliction of duty. Such speculation is morally excluded for the captain by the fact that he must decide whether there will be a battle.Footnote ³⁴

And so responses to the fatalist reasoning that involve variations on the theme that ‘Thoughts / propositions asserting the occurrence of a future contingent event are not true now’ come too late. These responses simply take on the Fregean notion of a ‘thought’ in a way that perpetuates a deeper distortion in that reasoning. For it is not the truth (or the falsity) of the thought ‘There will be a sea battle tomorrow’ that is in tension with the commander's practical deliberation; but rather, the speculative attitude towards the future event: the attitude in which his thinking is guided by the question of truth or falsity. The apparent power of the fatalist reasoning derives from the idea that the demand for speculative reflection overrides any demand for practical deliberation. That is not an idea that we need to accept. And the suggestion that such reflection may result in something other than a judgement of truth or falsity (as in the various proposals summarised above) is neither here nor there from the perspective that I have presented. Conversely, that truth, as such, is not the problem is seen in the possibility of exchanges of the form: ‘Is it true that there will be a sea battle tomorrow?’ the sailors ask the captain, to which he responds ‘Yes’. The possibility of this affirmation does not rest on (or imply) that it is necessary that there will be and so that there is no need to deliberate. For there is (we may assume) nothing to prevent him changing his mind if further reflection casts things in a different light.

Frege's ‘thoughts’ are essentially public, in the sense of being accessible to, and sharable by, anyone: there for anyone to think. In contrast with the material and perceptible event of my thinking or asserting something, the thought that is expressed in these acts is, in no sense, peculiarly mine. The thought is something with which anyone can come into the same relation. While some may worry about a perceived ontological extravagance in such a picture, my concern has been different. My focus has been on the relation of these entities to such material and perceptible events as thinking, asserting or deciding: on the ways – the range of different ways – in which these thoughts may enter into the lives of individuals at particular times and places. If we let go of the idea of the priority of the judging mode, we can acknowledge the possibility, for example, of the commander recognising that another may be in a position to predict whether there will be a battle without this compromising his sense that whether there is or not is in his hands. While the same thought is there to be related to by both, nothing dictates that the captain must relate to it in the mode ‘Is it true or is it false that I will order engagement?’ rather than in the mode ‘Should I or shouldn't I do so?’ Or consider a familiar perspective that is notably absent from standard treatments of fatalism: that of another with whom the agent is potentially in conversation. While we can imagine such a person adopting an enquiring or a manipulating stance towards the agent the stance that would normally be expected in friendship is different. Would the truth of determinism, or the ‘reality of future events’, entail that the appropriate stance towards one's friends was one of disinterested enquiry? The answer, I take it, is clear. That it is totally predictable that, in the absence of my intervention, there will be a sea battle tomorrow does, in itself, nothing whatever to suggest that it would be inappropriate for me to attempt to reason with the captain: does nothing whatever to suggest that my question should be ‘Will he or won't he order one?’ (‘Will there be a battle or won't there?’), rather than ‘Should he or shouldn't he order one?’ (Central though it is to one of the debates in this area, I will resist the temptation to ask what would be appropriate if I, the bystander, happen to be God.)

6. I have spoken of Frege's thoughts, as Frege himself insists we must, as entities existing quite independently of us. While this is called for by the role that they are to play in the fatalist reasoning, it also, I have suggested, creates a problem. For, to the extent that we picture a particular thought as a freestanding entity with which we may come into relation through thinking it, we explicitly leave room for the option of not coming into relation to it, or of coming into relationship with it in a way quite different from that involved in an assessment of it as true or false. And once these options are acknowledged the possibility that we sometimes ought to refuse to come into relation with the thought in that way comes clearly into focus.

Aristotle's own discussion does, on the surface, remain more consistently rooted in the material and perceptible than do approaches inspired by the Fregean imagery. Here is how he presents the matter:

Aristotle's suggestion is mirrored in recent responses to concerns about a Fregean notion of a ‘proposition’. Thus, Bernstein writes: ‘We can analyze “proposition p is true at time t” as “had anyone uttered at t the sentence that expresses proposition p, he would have expressed a true proposition”’.Footnote ³⁶

Aristotle and Bernstein are pretty clearly far too relaxed in these remarks. While there may be no difficulty in the idea of someone ten thousand years beforehand having produced the sounds ‘There will be a sea battle at Salamis on 20^th September 480 B.C.’ or ‘Xerxes will order an attack at dawn on ….’Footnote ³⁷, there is certainly room for questions about what he would have been doing in producing those sounds. After all, while I have uttered the sentence ‘There will be a sea battle tomorrow’ a number of times recently I have not, by presumably anyone's lights, in doing so said that there will be a sea battle tomorrow or expressed any proposition, true or false. And there are other kinds of case to be remembered here: utterances of the same sentence but with the distinctive rising pitch of a question, someone practicing English grammar, or someone telling the future by plucking petals off a flower ‘There will be….. There won't be ……’.

There is more to saying something than the uttering of certain words.Footnote ³⁸ Just as acting requires conditions (engaging with the enemy requires an enemy to engage with), so does asserting or thinking. And so, when Aristotle suggests that ‘there is nothing to prevent someone's having said ten thousand years beforehand that this would be the case’ we should reflect on just what we need to imagine as having been in place for anything that happened then to have been a saying of this – as opposed, on the one hand, to something other than a saying, and, on the other, to a saying of something quite different.

There are also questions about the relevance to one contemplating action of such deeply counterfactual reflections about hypothetical assertions. (Questions that roughly mirror ones about the relevance of the existence of a Fregean ‘thought’ to one contemplating action.) But let us consider a captain who supposes that, in one way or another, appropriate circumstances were in place and the affirmation was made. How should that reflection impact on his thinking? Well, just what is the context and who is this speaker? If we are to picture the case as one in which a speaker affirms something (in contrast with something closer to a mere tossing out of sentences) we are, perhaps, going to have to think of the speaker as, more or less, superhuman. Perhaps, then, it was a god announcing his intention to ensure that there is a sea battle on this day whatever the captain should decide. Or perhaps it was a god who takes more of a back seat – or a super scientist with a wondrous computer into which he has fed all the data – who predicts the whole situation, including the fact that there is a causal chain in place that will bypass the captain; or, alternatively, that the man in charge of the fleet will not be a lunatic, and seeing that it would be sheer lunacy for the captain to instigate action, confidently announces that he won't. Or perhaps the predictor was one who, having made his firm prediction, would, had the prediction been of disaster, certainly have informed the captain and tried to persuade him to follow a different course.

While uttering the words ‘There will be a sea battle tomorrow (in 10,000 years)’ – out loud or to myself – may be more fully under my control than bringing about a sea battle, for such a public or private mouthing to be a saying or thinking that there will be a sea battle requires more: requires a ‘location in the world’ from which one speaks.Footnote ³⁹ And any bearing that the saying or thinking should have on the captain as he faces his decision will depend crucially on the character of that ‘more’. Failure to acknowledge the importance of a ‘location in the world’ is seen in Frege's idea of the public availability of thoughts; and in the early Wittgenstein's finding intolerable the idea that one part of the world might be ‘closer to me than another’. The idea that the thinker is, in this sense, not of this world may be the most persistent legacy of Cartesianism.

7. My doubts about common treatments of fatalism have focused on two assumptions. First, on the assumption of a primacy of thought over action: in the sense that an enquiring stance towards a possible future event or state of affairs necessarily takes precedence over a deliberative one. This presumption finds a pure, and very general, expression in the idea that the possibility, or lack of it, of our influencing the course of events is grounded in the ‘metaphysics of time’ (in the enquiries of the most fundamental kind of thinker: the philosopher): grounded, for example, in the fact, if it is a fact, that future events are not yet ‘real’. I have focused, second, on the assumptions involved in a readiness to consider ‘thoughts’ or ‘propositions’ in independence of thinkers or asserters: to consider grammatical sequences of English words in abstraction from any context of use. That the doubts that I have been raising are largely absent from recent discussions of fatalism is, I think, testimony to the hold that this pair of preconceptions has on contemporary philosophical thought.