Counterintuitive Consequence of Open Theism
February 22, 2009 — 13:54

Author: Andrew Moon  Category: Open Theism  Comments: 17

I was talking w/Alan Rhoda at the Central APA, and I discussed with him the following counterintuitive implication of versions of open theism which say that all future contingents (or sentences about freely willed acts) are not true.
me: I bet that Curley will take the bribe tomorrow.
Alan: I don’t think he will.
me: let’s bet!
Alan: okay.
Tomorrow comes. Curley takes the bribe.
me: I was right!
Alan: I guess you were!
It seems that when I say “I was right!” I am ascribing truth to the sentence I uttered the day before. And intuitively, I speak truth when I ascribe truth to the sentence I uttered the day before. And if it is true that I correctly ascribe truth to the sentence I uttered the day before, then I did speak truth the day before. But some versions of open theism are committed to the counterintuitive implication that I did not speak truth the day before (i.e., utter a true sentence).
Two points: not all versions of open theism have to deal with this (e.g. Hasker’s). Secondly, I take this only to be some degree of negative evidence against open theism; perhaps there is more positive evidence for open theism.
I’ve heard this objection a lot in conversation, though I haven’t seen it in the literature. If I were an open theist, I’d just say that ordinary people have false views about the future, and so they are speaking incorrectly in these cases. Are there better ways out of this problem? Has there been any literature on this?

  • Brian Boeninger

    Hi Andrew! (Good to meet you at the APA.) You might want to check out the paper by Alan Rhoda, Gregory Boyd, and Thomas Belt titled “Open Theism, Omniscience, and the Nature of the Future” (published in Faith and Philosophy, 2006). In the third section of that paper, the authors discuss what they call the “Peircean” and “Ockhamist” views about the kinds of case you have in mind – for instance, whether one can infer from the fact (on Wednesday) that “S did A on Tuesday” the conclusion that “S will do A on Tuesday” was true (when uttered) on Monday. They deny that such an inference is warranted. They count this result as intuitively correct, given various considerations concerning “rational assertibility.” Might want to consult that paper (I think it can be found on Rhoda’s website) for a nice account of how an open theist will (er, might!) respond to your Curley case.

    February 22, 2009 — 15:38
  • Andrew Moon

    hello, hello Brian! It was nice to meet you as well. thanks for the reference and the comments. I’ll probably have to add it to my “to-read” list after I finish my dissertation.
    In my case there’s no inference to the conclusion that I speak truth when I say “I was right!” That is supposed to just be intuitively true. And it is intuitively true. So there should be no worry about any inference being unwarranted.
    And the intuition isn’t that I am rational in asserting that I was right. The intuition is that the sentence I speak is true. Compare: suppose we had strong evidence that Curley took the bribe, but it is false that he did. In that case, I am merely rational in asserting that I was right. But in my case, it is stronger; it is intuitive that my sentence is true.
    Of course, this may totally miss their point, but that’s life!

    February 22, 2009 — 16:30
  • Geoff Pynn

    Believers in an open future can have:
    (a) What I said yesterday is true.
    Just not:
    (b) What I said yesterday was true when I said it.
    Error theory: when what you said at t concerns t or some earlier time, (a) is true iff (b) is true. It’s only when what you said at t concerns some moment after t that (a) can be true even if (b) is false. Ordinary speakers are blind to this very subtle distinction; that’s why “I was right” is appropriate in cases where (a) is true but (b) is false.

    February 22, 2009 — 21:14
  • jon kvanvig

    Hi Andrew, I think I give a version of this objection, aimed at Geach’s view, in my ’86 omniscience book.

    February 22, 2009 — 21:25
  • Hi Andrew – it was good to meet you at the APA.
    I think that if you think this objection has any weight, it is because you’re assuming that whatever is true is omnitemporally or timelessly so. When you let go of that, and believe in an open future, it’s just a logical consequence of the position. Regarding common sense, common folk will also concede in many cases that what was true back at the time of the (fulfilled) was not S will do A, but rather probably S will do A – the latter of which, of course, will have been true later whether S does A or not. In any case, in the last section of my F&P paper (“Three Roads to Open Theism”), I think I get into this objection as raised by Bill Craig.

    February 23, 2009 — 5:33
  • Keith DeRose

    I’m inclined toward something like the error theory Geoff describes above. (And I am an open theist.) It’s worth explicitly acknowledging that because this is an error theory, there is some cost to resorting to it. Andrew is right that it counts against the view that future contingents have no truth value (which is what I hold) that we don’t just say “What I said is wrong” or “…turns out to be wrong,” but seem to appropriately say, “I was wrong.” But to my thinking, explanations like the one Geoff describes mitigate that cost fairly effectively.
    (A more evasive, slippery answer would be to say: “It depends upon what the meaning of the word ‘was’ is.”)

    February 23, 2009 — 10:05
  • Geoff:
    A technical point: I think even (a) cannot be be taken literally by the typical open futurist. For the typical open-futurist is an A-theorist. And the A-theorist has to, I think, take it that propositions are tensed, in the sense that the very same proposition that it is moderately sunny in Waco which is true now was not true eight hours ago (it’s never sunny at night). If so, then what was said yesterday, viz., that Curley will take the bribe tomorrow, is not actually true (unless Curley will take the bribe again tomorrow).
    Now, the closed future A-theorist can take care of this difficulty very easily by taking (a) to be short for (b). The open future A-theorist can’t make that move. She needs to do a little bit more work.
    Here’s another way to see the problem, at least on the Rhoda et al account. If a proposition cannot ever be true, and bivalence necessarily holds (as Rhoda et al agree) then the proposition is necessarily false. That Curley will freely take the bribe cannot ever be true, hence it is necessarily false. But if yesterday I uttered a necessary falsehood, how could what I said be true now?
    One thing the open futurist could do is to become a B-theorist. You might have thought that you can’t be an open futurist and a B-theorist. But maybe you can. You just think the basic facts are of the form “… at t”, and then say that there is a t0 such that for all t>t0, there are a lot fewer such facts.
    OK, maybe that’s crazy. A less crazy thing to do is to try to introduce an operator U such that U(p,delta) is an “updating of p” by a temporal distance delta. Intuitively, U(p,delta) is a proposition such that, necessarily, p holds now iff p will hold delta (units of time) later. But you can’t take this as the definition of U(p,delta) because on the Rhoda et al semantics, for a wide range of p (such as the proposition that Curley will freely take the bribe) it will be the case that necessarily p is false, and so the “iff” will severely underdetermine U(p,delta).
    Anyway, assuming it’s now 24 hours after the initial prediction, the semantics for (a) is that (a) holds iff U(dthat proposition which I said yesterday, 24 hrs) is now true. (There is still one technical issues here. This won’t work for negative claims, except on the Rhoda et al. semantics, because p might be compatible with time coming to an end in less than delta (units of time) while U(p,delta) is not compatible with that.)
    One might ask whether the open futurist is entitled to any such operator U, however. Here is a natural candidate for U, after all. U(p,delta) is the proposition that p held delta (units of time) ago. But if that’s our candidate for U, then U(dthat proposition which I said yesterday,24hrs) comes down to the claim that what I said 24hrs ago held. And that’s (b). So this won’t work.
    Thus, U(p,delta) needs to be defined more carefully using the syntactic structure of p. Thus, if p is the proposition that in h hours Curley will take the bribe, then U(p,h1 hours) is that Curley did take the bribe h1-h hours ago assuming h1 is greater than h. So this account requires propositions to have a syntactic structure, or else you need to work with sentences instead of propositions.
    The problem may thus have a solution. But it does mean that (a) is not straightforwardly true. What the open futurist ends up doing is giving an error theory for (b), and then a rather complex semantics for (a).
    I also don’t know whether U(p,delta) can be defined for all propositions p about the future. Have open futurists given a good argument that it can? Or is there some other way of providing a semantics for (a)?

    February 23, 2009 — 10:56
  • Geoff Pynn

    Thanks for your challenging reply. I think you are right that an open futurist is going to have her work cut out for her giving a plausible semantics for my (a). But I also think a good enough error theory for the kinds of statements Andrew is concerned with can still be had (no worse than typical error theories, at least). So, for example, I don’t see that (a*) or (a**) would face the problems you raised:
    (a*) What I said would happen did happen.
    (a**) I said that Curley would take the bribe and he did.
    New error theory: when I said Curley would take the bribe it wasn’t true that he would. But (a*) and (a**) are true. Etc.
    It would be nice to come up with a way for an open futurist to be able to say that tensed things he said (or believed) in the past about the then-future turned out to be — and hence are now — true. But the more I think about the problems you raise the less clear it seems to me how one could do that.

    February 23, 2009 — 23:29
  • Geoff:
    1. Indeed, (a*) and (a**) do seem to work nicely for events. “What I said would happen” is presumably an event (it’s events that happen)–an event of Curley’s taking the bribe–and then the claim in (a*) is that that event did in fact happen. Good.
    As you note, it’s trickier with claims that are not just reports of a single future event.
    2. I was thinking a little more about my impossibility worry. On the semantics in question, it is impossible that Curley will freely take the bribe, since, necessarily, all future contingents are false. This is implausible. But maybe one can taken an error theory about that, and say that the reason we tend to say that possible Curley will freely take the bribe is that we want to modally distinguish the claim that Curley will freely take the bribe from the claim that Curley will freely draw a square circle. So I guess we can say: It will be true tomorrow that possibly Curley freely takes the bribe, but it will never be true that Curley freely draws a square circle.
    But here is a tricksy thing. One might think, one of the future contingents (at least in some possible worlds; in ours, divine promises might take care of this issue, pace Jon’s worries about these) is whether there will in fact be a future. After all, just as time can have a beginning, time can have an end. But if so, then no future tensed positive claim is true on the semantics, not even a claim of the form “it will be the case that T” where T is a tautology, since any future tensed claim entails that there will be a future. And if that’s right, then it is false that it will be the case that possibly Curley freely takes the bribe.
    Maybe you can deny the possibility of time having an end. It would be interesting if a commitment to the future’s being open were to require one to hold that it must always be determined that there will be a future.
    3. And while we’re at it, here’s another weirdness. Suppose I am in a sealed room starting at 11:30 am, and before I went in the room I bought an indeterminsitic lottery ticket, in a lottery with googolplex tickets. Let’s say that the lottery takes place at noon. Now here is the weird thing on the open semantics. I stare at my watch. It’s 11:30. I don’t yet know that my ticket is a losing ticket. It’s 11:45. I still don’t know. It’s 11:59. I still don’t know. But as soon as it’s noon, I suddenly know it’s a losing ticket, without any outside information. Weird, isn’t it?
    Moreover, suppose that it is determined that I cannot get out of the room and that I will not gain new information except about the time and that I will not lose rationality, and I know all that. (Maybe God unconditionally promises me all this.) Suppose I know the proposed open future semantics to be correct. Then at 11:30 I already know that exactly at noon I will have a justified belief that I have the winning ticket. But right now I know that it is false that I have the winning ticket. That’s weird. And come noon, I reflect on the fact that I now have a justified belief that I have the winning ticket, and that yet I once knew the proposition that I have the winning ticket to be false. That, too, is weird.
    4. So now, let me pose this question. The open futurist has to say a lot of weird sounding stuff, and in particular has to be an error theorist about a lot of things we say. Isn’t it rationally preferable just to deny one of the premises in the argument from closed future to fatalism? Maybe the premise that if it is already the case that p, then it is necessarily the case that p (on one formulation), or the premise that truth supervenes on the present state of things (on another)? These kinds of premises just do not seem to be nearly as plausible as the things that the open futurist has to deny. But perhaps that’s just my intuition.

    February 24, 2009 — 16:54
  • Geoff Pynn

    Re 2. Wow, it would be extremely bad if OFers had to say of a future contingent E that it is impossible that E will occur. I mean isn’t E supposed to be a future *contingent*!?? I would reject any semantics with this result.
    Instead, I would go truth-value-less on future contingents. So even if the OFer has to concede that Curley will freely take the bribe can never be true, it is also not false, and a fortiori isn’t necessarily false. Haven’t thought through how it would go & surely there are other problems lurking as there always are with denying bivalence.
    But more importantly I think usually when we say “it’s possible that E will occur” and E is a future contingent we intend an epistemic reading. If “it’s epistemically possible that P” means roughly “I don’t know that Not-P” then there is no error theory needed to explain how “it’s possible that Curley will take the bribe” is true provided we’re dealing with epistemic possibility. My intuitions deliver an extremely flickery and weak response when asked whether it would be bad for that sentence to be neither true nor false on a metaphysical reading. If one was going to go with truth-value-less statements about the future events why not double down and throw in truth-value-less statements about future metaphysical possibilities?
    I have a couple of thoughts about the lottery case too. But if I start writing them up I’ll never get my lecture ready for the morning!

    February 24, 2009 — 20:52
  • Geoff:
    I’m sorry. I thought you accepted Alan Rhoda’s semantics (since the thread started with a discussion with Rhoda), and so everything I said was predicated on that. On a non-classical logic semantics, things will be different, and these linguistic problems will be somewhat less problematic. However, it seems to me that changing logic should be a last resort, when we have an argument all of whose premises are utterly undeniable, etc. But the premises in the arguments for open future aren’t like that.

    February 25, 2009 — 7:49
  • Modality and open future

    I’ve been thinking what open future (OF) views can say about the modality of statements about the future. There are two OF semantics, which I’ll call N and F. Suppose Curley now exists, and that Curley’s freely taking the bribe is open. On the N semant…

    February 25, 2009 — 10:17
  • Matt Benton

    Andrew: I think that John MacFarlane’s relativist semantics is *one* way for the open futurist to handle this problem of retrospective truth/falsity assignment… See his “Future Contingents and Relative Truth,” Phil Quarterly 2003.
    (I have only scanned the other comments here, so I’m sorry if someone already gestured at this.)

    February 25, 2009 — 19:44
  • Geoff Pynn

    I fear that I may have been even more misleading — I’m not even a believer in an open future! In fact, I’m a determinist, and (insofar as I understand the meaning of the term as it’s used by philosophers) a fatalist. So I don’t really have a dog in this fight.
    However, it does seem to me that (a) OFers shouldn’t be too threatened by the fact that “I was right” is felicitous wrt some past future-directed assertion, and that more generally (b) facts about the semantics of future-directed sentences shouldn’t decide the issue. That’s what got me interested in the thread initially. And now, perversely, I’ve acquired an interest in defending OFers against the challenges you’ve posed!

    February 25, 2009 — 19:58
  • Alan Rhoda

    In response to Andrew’s original question, I would concede that retrospective applications of the word “true” do lend some support to an Ockhamist semantics, but I think that support is very slight. For one thing, it doesn’t follow from the fact that people often say thinks like “you were right” that you were in fact, at that time, right. For another, there is linguistic evidence that points in the other direction (I’m thinking here of how people typically speak in lottery cases: “Is it true now that you will not win? No, because I might win.”)
    Matt’s right that MacFarlane’s relativistic semantics is an option for open futurists. I doubt, though, that it’s all that helpful for open futurists who are theists (i.e., open theists). On the plausible assumption that God’s beliefs are true simpliciter, as opposed to true-with-respect-to-moment-of-evaluation-E, relativistic semantics will not be of help to us in unraveling the dilemmas of freedom and foreknowledge.

    February 26, 2009 — 12:40
  • Matt Benton

    Hi Alan: I’m not sure I follow your last point.
    I’d expect the open theist to think that God’s beliefs about future-tensed propositions won’t be true simpliciter, and their truth-conditions could track those of the relativist semantics. (The open theist will likely have a different approach to divine “beliefs” about future propositions: presumably God has perfect probabilistic credences, not anything like on/off beliefs. And if so, all such credences are true, but those credences may change over time.)
    This might force the issue in favor of God being “in time” in a way similar to us: but then again, I think most open theists already think that.

    February 26, 2009 — 13:00
  • Alan Rhoda

    Hi Matt,
    Perhaps I should clarify what I mean by ‘true simpliciter’. I don’t mean timelessly true or omnitemporally true. Rather, what I mean is true in virtue of what exists simpliciter, where the latter means what exists (period). In other words, it means what exists from an absolute, non-relativistic perspective (the proverbial “God’s eye” perspective). Now, if the “God’s eye” perspective is, in fact, God’s perspective, then relativistic semantics doesn’t apply.
    While I wouldn’t say that God is “in” time, I do think that God experiences succession. This is compatible with the above account of truth simpliciter so long as we don’t make the mistake of assuming that an absolute perspective is ipso facto a timeless or temporally invariant perspective.

    February 26, 2009 — 14:55