Monday, June 13, 2011

Fatalism, Indeterminacy, and No-Future Views

A paper I gave at the 2008 Eastern APA (at the Philosophy of Time Society group meeting):


Presentists and growing block theorists reject the idea of a concrete, determinate future. Such ‘No-Futurism’ is often in part motivated by an incompatibilist or libertarian view of free will and the desire to avoid the fatalism which is thought to accompany the denial of No-Futurism. However, this avenue to No-Futurism can be successfully blocked – we can show that the very assumptions which are meant to entail the incompatibility of a determinate future with libertarian freedom actually also entail, in conjunction with the common assumption that bivalence fails for future contingents, the incompatibility of an indeterminate future with such freedom. And in fact, what we are in the end forced to say is that this freedom, when conjoined with a denial of bivalence, is straightforwardly incompatible with No-Futurism in general, whether or not all of the No-Futurists’ assumptions are accepted in the first place. Far from providing a successful route to the nonexistent future, considerations from human freedom and incompatibilism actually point, if anywhere, in the opposite direction or, at least, to some other No-Future views which have their own share of problems.

I. The Open Road
Many people reject the existence of a real, concrete and determinate future because they think it leads to fatalism and hence a lack of freedom on our part. These people tend to fall neatly into one of two camps: presentists, who believe that neither the future nor the past exist (or at least that they do not do so except as parts of the present), or growing block theorists, who believe that only the future fails to exist. More carefully, presentists take it that everything which exists is present and exists at the present, whereas growing block theorists take it that everything which exists is either present or past and exists at the present or in the past. Though these two groups disagree about the ontological status of the past, they agree for the most part about the future. From now on, I will call such views ‘No-Future’ views.

Many No-Futurists hold the beliefs they do at least in part because of a widespread belief in Libertarianism about free will and the looming threat of fatalism which a determinate future is thought to provide. According to Libertarians about free will, free will both exists and is incompatible with determinism (that is, Incompatibilism is true). Libertarians will also take it that free will automatically rules out fatalism about our actions – that is, it rules out the possibility that we might have no power (at every given time) to affect what we will do in the future. So if fatalism turned out to be true, that would automatically rule out the existence of free will as understood by Incompatibilists. According to many of those who hold such views, though, the only way to avoid the fatalism supposedly provided by a determinate, real future is to get rid of it. Indeed, one often sees reasoning such as the following: “If it really is determinately the case that I will eat a burger even before I do it, then I had to do it. So there can be no determinate future featuring my eating of a burger – otherwise my eating would not have been a free action. So there is no future.”

Diekemper (2007, 429) expresses this attitude of the No-Future theorist as follows:

…if the event of the Third World War exists eternally, then in what sense is that event—prior to its occurrence—not inexorable? The [No-Future theorist’s] intuition is that there is no sense in which it is not inexorable; and so, in order to preserve the potentiality of the future, many philosophers of time have rejected the B-Theoretic doctrine of an existent future.

Tooley (1997, 45-46) describes (but does not endorse) such an argument as denying the existence of future facts because such facts would thereby be unpreventable, even by an omnipotent person. Lucas (1973) seems to endorse this sort of reasoning as, in several places, does Prior and those many who would equate eternalism with a deterministic block universe. Indeed, many people seem to take a real, determinate and concrete future to be a very grave threat to our freedom.

These sorts of lines of thought generally proceed from a largely Incompatibilist or Libertarian view of free will, according to which the predetermination of what I shall do by facts beyond my control render my actions thereby unfree. Let us state a formal, stripped down version of the Incompatibilist view as follows:

Incompatibilism: If either □N F(p) or □N F(~p), then ~Free (F(p)).

Here, ‘p’ stands in for some claim ascribing an action to me and ‘F’ is an operator making the claim future-tensed. ‘Free(F(p))’ claims that I am free with respect to whether I will perform the action described in ‘p’. Assuming a standard logic for tense, we can take ~F(p) to either entail or be equivalent to F(~p). It is important to note that the modality expressed in the modal operators here is not one of absolute metaphysical modality or logical modality. Rather, such a modality is meant to be understood in terms of the temporally-relative modality of ‘accidental’ or ‘temporal’ necessity – the modality of inevitability at a given time and future contingency. The basic idea of this sort of modality is that something is possible with respect to a given time if and only if it is compatible with the facts about that very time (plus the laws of nature and perhaps facts about earlier times). The Incompatibilist might add here that something can only be something I really have a choice about or say in if it is possible in this sense. So we should take ‘□t ’to mean roughly the same as ‘It is inevitable at t that…’ and ‘t’ to mean roughly the same as ‘It is not inevitable that it is not the case at t that…’. All modal operators in this paper should be understood to express this sort of modality and, unless stated otherwise, all verbs should be taken as present-tensed. In our statement of Incompatibilism, then, the subscript ‘N’ should be taken as referring to the current time, so that the whole claim can be read, ‘If it is either now inevitable that p or now inevitable that not-p, then I am not free with respect to whether it will be the case that p.’

On this sort of incompatibilist view we are looking at, then, causal determinism of the ordinary sort is straightforwardly incompatible with free will since if my performing a certain action at a later time is now causally determined, it will not be possible now for me to do otherwise and hence I will not have the power to do so and hence will not be free with respect to that action.

So far, however, none of this gives us any lack of free action when conjoined merely to a determinate, real future. On its own, Incompatibilism may appear prima facie compatible with a determinate future unless we add some further assumption or set of assumptions. As one might expect, though, some No-Future folks will also accept the following controversial principle (though not always stated in this precise form) :

Fatalistic Principle (FP): If Det(F(p)) then □N F(p).

Here, ‘Det’ should be read as ‘It is determinate that…’. I suppose the basic intuition behind FP (which seems on many interpretations to play the key role in Aristotle’s “sea-battle” discussion in De Interpretatione) is that if it is already true that I will perform a certain action, then this settled and hence, in order for it to be possible for me to do otherwise, I would have to be able to causally effect what is already settled (that is, the truth that I will perform that action) – an activity which to many seems impossible or at least beyond the powers of mortal ken. In other words, given that it is already true that I will A in 2010, for it to be possible for me to do otherwise than A, I would need to continue to have the ability to causally influence whether I will do A in 2010. But that would require an ability to causally influence whether it is now or was in the past true that I will do A in 2010. But since those things are already settled, it is supposed, I can have no such ability since those sorts of things cannot be causally influenced. So, on this intuition, if it is indeed true that I will perform an action it must also be necessary now that I do it – it is inevitable and completely unavoidable. And it is precisely this principle, FP, which in conjunction with Incompatibilism, entails the controversial position held by many that freedom with respect to some future action requires that it be indeterminate whether I will take such an action. From Incompatibilism and FP, then, we get:

Openness Principle (OP): If either Det(F(p)) or Det(F(~p)), then ~Free(F(p)).

And so these No-Future theorists will take it that, since there are instances where I am free to do something, there are therefore instances where my doing it in the future is indeterminate. So the Incompatibilist who accepts the Fatalist Principle will thereby be committed to OP and hence, if they believe in free will, will be forced to reject the idea that there is a real and determinate future. No-Futurism follows.

II. The Fatalistic Principle Strikes Back
That OP follows from the assumptions of the previous section is, I think obvious. But that these same assumptions are compatible with the existence of free will is, I think, taken for granted. As I hope to show now, one need not make any further substantive assumptions beyond what many No-Futurists already accept to reach the conclusion that Incompatibilism and FP, together with a common No-Futurist assumption, jointly entail that we have no determinate freedom, so that rejecting a determinate future will not help us to save free will after all.

A key assumption that we need here is one that is widely adopted by many No-Futurists, and this is just the denial of bivalence for future contingents – that is, that for any statement which neither is inevitable nor whose negation is inevitable, such a statement is neither true nor false (not that it is indeterminate whether it is true or false, but rather genuinely neither). We could take this to mean either that future contingents possess a third truth-value separate from truth and falsity, or simply that they are truth-valueless. I do not believe which one we choose will make much of a difference in what follows, so long as we deny bivalence in either case. On this sort of view, for some truth to be determinate is simply for it to be true (and vice versa) and for it to be indeterminate is simply for it to fail to be true and also fail to be false. That is, indeterminacy is a third option – a claim may be true, false, or neither. This view, then, is committed to the following principle:

Alethic Determinacy (AD): If p then Det(p).

Now what should someone who rejects bivalence think about the law of the excluded middle for future contingent propositions? That is, should we take the following as true or not:

Future Excluded Middle (FEM): Either F(p) or ~F(p).

One option here is to adopt a supervaluationist branching future view, accept the truth of FEM, and save the law of the excluded middle. On this strategy, there are many possible futures branching out from the present (and hence many possible complete histories) and if a given claim is true on all branches (that is, on all histories) then it will count as true, if false on all branches then it will be false, and indeterminate otherwise. And since on every history the action described in p occurs or it does not, FEM is also true on every history and hence is true simpliciter. Supervaluation, however, will not be much help here. Consider that we can here distinguish between two notions or modes of truth – there is super-truth (truth at all possible histories) and there is local truth (truth at a given possible history). Super-truth, then, is merely local truth relative to every possible history. A statement like p, then, can be locally true relative to some possible history while failing to be super-true. Local truth, that is, does not entail super-truth. Since Det(p) is generally taken to be true just in case p is super-true, p can be locally true without being determinately true. That is, for a given possible history, it could be true that p while false that Det(p). But this means that, given the existence of some future contingent, AD fails to be locally true relative to every possible future. If we take AD to be super-true just in case it is true at every possible history, then it follows that AD is in fact not super-true.

Along similar lines, we could show that on this understanding of the branching time framework that for any statement F(p) which reports a future contingent, it ends up the case that Det(F(~Det(p))), from which it would follow via the No-Futurist’s principles that I am not free as to whether or not it will be determinately the case that p. This may be better than it not being the case that I am free as to whether or not it will be the case that p, but for someone who wants a robust Libertarian free will it does not seem all that much better. There are various moves that could be tried at this point to avoid all these apparently bad consequences, but in every case I think the rejection of AD will be more plausible than any view which happens to retain it. The truth that AD is trying to capture, it seems, is simply the triviality that if it is determinate that p then it is determinate that p. This is admittedly a rather short dismissal of other supervaluationist alternatives, but for now I will leave the discussion at that and simply assume from now on that such gambits will not ultimately deserve our acceptance.

Supposing, however, we simply stipulate that AD is true at every possible history really is necessarily true in the strongest sense and really does hold even in each of these possibilities. In that case, it is important to remember that FEM is supposed to be true at each history. But that is simply because, for each history, either one disjunct is true or the other is. Given a possible history where F(p), it is also true that either F(p) or ~F(p), and the same for any history where ~F(p). But then, by AD, in histories where F(p), it will also be true that Det(F(p)) and hence that either Det(F(p)) or Det(~F(p)). Similarly for histories where ~F(p). But then, by OP, it follows that in each of these possible histories there is no freedom. But since this holds for every possible history, it holds simpliciter and hence there is no freedom, period.

Since the supervaluationist option was a bust, the anti-bivalentist might instead suppose that FEM, like its atomic constituents, is neither true nor false. This would be a desperate move indeed but perhaps not hopeless. Let us introduce a new operator, ‘Ind’, which obeys the following rule: For any claim c, ‘Ind(c)’ is true just in case c is neither true nor false; it is false otherwise. ‘Ind’ should, then probably be interpreted as saying something like ‘It is indeterminate that…’. Since FEM, we are supposing, is neither true nor false, the following proposition will be true:

Indeterminate Excluded Middle (IEM): Ind(Either F(p) or ~F(p)).

But since ‘Either F(p) or ~F(p)’ entails that I am not free with respect to F(p), it follows that either I am not free with respect to F(p) or it is indeterminate whether I am not free with respect to F(p). That is, ‘~Free(F(p))’ will be the case or ‘Ind(~Free(F(p)))’ will be the case. The former is straightforwardly inconsistent with a Libertarian position, but what of the latter? Since ‘Ind(F(p))’ will only be true just in case ‘Ind(~F(p))’ is also, it follows that ‘Ind(Free(F(p)))’ is also true and hence it is not determinate that I am free with respect to F(p). That is, it is indeterminate whether I have free will. This might be slightly better than having no free will at all but, again, if it is better, it is not much so. After all, a genuine Libertarian is surely going to be committed to its actually being the case – to its being determinately true, in fact – that I am free, not just that it is neither true nor false that I am free. So even if this gambit does not exclude free will, we can surely describe it as excluding determinate free will and that is about as bad. From now on, I will drop the ‘determinate’ adjective and simply assume that the free will we are interested in here is supposed to be determinate. Whether FEM or IEM is accepted, then, freedom is excluded. Instead of securing freedom, it looks like our No-Futurist must reject it.

III. Why No-Futurists Must Accept the Fatalistic Principle
To get out of this mess, it seems we must reject one of the following: free will, Incompatibilism, FP, AD, or FEM. Since I cannot see how anyone would plausibly accept all the other assumptions and yet reject FEM as false, I will leave that option aside. The live options, then, are these: either give up free will, reject Incompatibilism, give up the Fatalist Principle, or accept bivalence and reject AD. Although I would opt for both of the latter two options, it is perfectly open to the Libertarian No-Futurist to take only one. That option, however, cannot be the rejection of the Fatalist Principle – the No-Futurist is committed to this principle by their very views about the future and, hence, the only way for a No-Futurist to adopt a Libertarian position is for them to accept bivalence and reject AD. What I claim, then, is that the following is true:

No-Future and Fatalism (NFF): If No-Future then FP.
NFF, of course should probably be restricted in such a way that it is true only of future truths – that is, as far as facts about the future are concerned, if No-Future, then if these future things are determinate they are necessary. Take this as understood. But how does FP follow from a No-Future view? Here is one way to go – we can say that it follows largely due to the principle that Truth Supervenes on Being (TSB) – a difference in truth value for a given proposition across two worlds requires a difference in the objects or instantiations of properties between those two worlds (that is, a difference in what there is and how things are between them). I take this to be an intuitively obvious principle (even though some may want to deny it) – after all, if the truth of a statement floats freely of the way the world is, we lose any grip on its really being about anything in the world in the first place.

But now, given TSB, if we believe No-Future we are going to be hard pressed not to accept FP as well. After all, if there is no future then all facts about the future must be grounded in the present (or the present together with the past) plus any relevant nomological laws. So if it is determinately true that p, then that fact is grounded in (let’s say) present facts. But for p to be temporally contingent, it must be temporally possible at this time both that p and that ~p. And that is only so if the intrinsic facts about the present (along with, perhaps, the past) do not determine that p. But if that were so, then given No-Future the present facts could not ground the truth that p. But, again, given No-Future and TSB, if that were so then it cannot be determinately true that p. So if it is determinate at this time, it is not possible at this time for it to be otherwise. Hence for the No-Futurist, NFF follows straightforwardly from TSB.

None of the various versions of the kinds of truthmakers No-Futurists give to ground truths about the future will work here. Take, for instance, the view that what grounds truths about the future are simply the current states of things plus the laws of nature. This is straightforwardly ruled out by Incompatibilism as a way of grounding truths about future free actions. Consider also the view that there are primitive, irreducible future-tensed properties such as being such that one will do A possessed by individuals or the world which ground truths about the future (see Bigelow 1996). The having of some such properties rather than some others clearly will be temporally necessary. Appeals to temporally distributional properties will seem to fare similarly, as will verificationists or dispositionalist accounts of the grounding of future truths. If any of those posited truthmakers now ground in the present what will happen in the future, such future occurrences are inevitable – that is, they are temporally necessary.

An ersatzist about times such as Bourne (2006) or Crisp (2007) – who treats times as abstract objects arranged in order via primitive ersatz earlier-than relations – will not do any better here. Such a view treats future truths as being grounded in what is represented as being the case by an ersatz time which is ersatz-later than the ersatz time which correctly represents the present. So if some such time represents p then it is true that p at the corresponding distance from the present time. But this is just as inevitable as any of the other options. These states of abstract objects are, after all, present states – as in all the other options, they render my future actions temporally necessary. Similar sorts of views which make future truths depend on God’s memories or will or some such divine mental state will obviously not do any better.

The argument for NFF I have been giving is similar to ones that have been given in the literature attempting to show that Libertarians and their ilk must reject bivalence or at least determinacy for at least some future-tensed statements. Rhoda, Boyd, and Belt (2007), for instance, argue that the guarantee by present conditions of the truth of future-tensed statements about my actions is straightforwardly inconsistent with Incompatibilism (though they do not use this exact terminology). And Michael Rea (2006) has skillfully argued that, if Presentism is true and all statements about the future are determinate then, since future truths about someone’s actions have not in the past been even partly grounded in the person such truths are about (since they were true before that person even existed), such a person never has and never will have a choice about such actions. And hence such a person will not be free with respect to such actions.

Rea considers ways someone might try to resist fatalism, but most of the ways out of principles like FP end up being available only to one who actually believes in a real, concrete and determinate future. At this point, though, one might think that if TSB is rejected for statements about my future contingents (as might be done by certain Molinists) then we would have a way out of NFF. But, as Rea (2006, 521-522) has in effect argued, even if TSB is false (as claimed by Merricks 2007), if No-Future is true and all statements of future contingents are now either true or false, then such ‘future contingents’ will indeed be temporally necessary. After all, these ungrounded truths are still present (or past) truths and, being ungrounded, they are not made true by anything outside of the present (or past). So the fact that F(p) is a wholly present fact, not dependent on anything future, and this fact guarantees or fixes what will occur in the future. Hence, any future contingents will be inevitable. So whether or not TSB is true, NFF follows once we accept a No-Future view.

IV. The No-Futurist’s Options
Given the truth of NFF, the No-Futurist is thereby stuck with FP. Hence, per the discussion of the previous sections, for the No-Futurist, the only options left are to abandon No-Futurism (so they can reject FP), abandon free will, accept compatibilism, or take bivalence on board. So for someone committed to both No-Futurism and Incompatibilist freedom, the only option is to accept bivalence and reject AD. Presumably, this would have to involve invoking some kind of perhaps entirely non-semantic metaphysical indeterminacy where F(p)’s indeterminacy is compatible both with its truth and its falsity – that is, where it really is either true or false, and hence bivalence holds for it, yet it is simply indeterminate which. On this view, indeterminacy would not be some third option over and above truth and falsity, but rather a kind of primitive state where the world is simply indeterminate between the two options. The problem here, of course, would be to make sense of this sort of metaphysical indeterminacy and the idea that something could be true but not determinately so.

Another bivalence-accepting strategy would be to take ‘~F(p)’ as true, and differentiate this from ‘F(~p)’. So far I have been taking these two claims to be either equivalent or for the former to entail the latter. However, were one to differentiate them, denying that one entails the other, and take all ‘will’ claims such as ‘F(~p)’, where they express future contingents, to be false, one could still affirm the truth of ‘~F(p)’. In this case, the inference from IEM to the denial of free will is rendered invalid. The basic idea of this view, then, is to accept bivalence and still get out of my argument by, in effect, denying that ‘will’ and ‘will not’ sentences are genuine contradictories. Contraries, yes, but not contradictories. This strategy, of course, has its own problems dealing with the way we ordinarily treat ‘will’ and ‘will not’ as contradictories and dealing with truth-value links between previous predictions about what were once future contingents and the truth of propositions about these same events occurring in the present..

So, from what we have seen, far from requiring a No-Future view of time, Libertarianism is straightforwardly incompatible with it (or at least with many common forms of it). To escape from this incompatibility, I have suggested two strategies the No-Futurists might make use of, both of which will involve a lot of work, if not substantial difficulties. Whatever the case may be, the denial of bivalence for propositions about future contingents in order to uphold our free will is a self-defeating gesture and No-Futurists who deny bivalence in the name of saving free will are instead attacking the very thing they set out to defend.


Aristotle, De Interpretatione
Asher, N., J. Dever, and C. Pappas (draft) “Supervaluations Debugged”
Bigelow, John (1996) “Presentism and Properties” in Tomberlin, James (ed.), Philosophical Perspectives 10: Metaphysics. Oxford: Blackwell, 35–52.
Bourne, Craig (2006) A Future for Presentism. Oxford: Oxford University Press.
Cameron, Ross (draft) “Truthmaking for Presentists”
Crisp, Thomas (2007) “Presentism and the Grounding Objection”, Noûs 41.1: 90-109.
Diekemper, Joseph (2007) “B-Theory, Fixity, and Fatalism”, Noûs 41.3: 429-452.
Ludlow, Peter (1999) Semantics, Tense, and Time. Cambridge, MA: MIT Press.
McGee, V. and B. McLaughlin (1995) “Distinctions without a Difference”, Southern Journal of Philosophy 33 Supplement: 203–251.
Merricks, Trenton (2007) Truth and Ontology. Oxford: Oxford University Press.
Naylor, Margery Bedford (1980) “Fatalism and Timeless Truth” in Inwagen, Peter (ed.), Time and Cause: Essays Presented to Richard Taylor, 49-65. Dordrecht: Kluwer Academic Publishers.
Parsons, Josh (2000) “Must a Four-Dimensionalist Believe in Temporal Parts?”, Monist 83.3: 399-418.
__________ (2004) “Distributional Properties” in Jackson, F. and G. Priest (eds.), Lewisian Themes, 173-180. Oxford: Oxford University Press.
Rea, Michael (2006) “Presentism and Fatalism”, Australasian Journal of Philosophy 84.4: 511-524.
Rhoda, A., G. Boyd, and T. Belt (2006) “Open Theism, Omniscience, and the Nature of the Future”, Faith and Philosophy 23: 432–59.
Tooley, Michael (1997) Time, Tense, and Causation. Oxford: Oxford University Press.
Varzi, Achille (2007) “Supervaluationism and its Logics”, Mind 116: 633-676.
Williamson, Timothy (1994) Vagueness. London: Routledge.

No comments: